Queensland Bulk Water Supply Authority t/as Seqwater v Rodriguez & Sons Pty Ltd

1. [1]

   THE COURT:

1. Introduction

1. [2]

   In 2014 a small business which had suffered loss as a result of the flooding, Rodriguez & Sons Pty Ltd (“Rodriguez” or “the plaintiff”), commenced a class action seeking to recover damages from the operators of the Wivenhoe Dam. Relevantly for present purposes, it alleged that the operators had been negligent in failing to commence releasing water from the Wivenhoe Dam several days earlier than they did, in anticipation of the heavy rainfall, so as to avoid the need to release much larger volumes in a short period after the rainfall caused large inflows and a surge in the dam level. The rising level caused fears for the safety of the dam, necessitating releases which, when combined with high flows from downstream tributaries, rendered extensive flooding of urban areas inevitable.

2. [3]

   A trial held before Beech-Jones J extended from December 2017 through much of 2018, the last written submissions being received in May 2019. The principal judgment (one of many judgments delivered in the proceedings), was published on 29 November 2019 [1] and ran to more than 1,500 pages. The judgment was divided into 15 chapters, each with its own paragraph numbering. References to passages in that judgment will take the form “Rodriguez (22) Ch 1 [22].” Subsequent judgments have also given rise to issues addressed by the notice of appeal, the latest being Rodriguez (29) delivered on 7 May 2021, just 10 days before the hearing of the appeal commenced. [2]

3. [4]

   Answers were given to numerous common questions which had been identified as appropriate for separate determination. The judgment upheld the claims of the plaintiff in negligence against the State of Queensland, and two statutory agencies responsible for operating the dams, namely Queensland Bulk Water Supply Authority trading as Seqwater (“Seqwater”) and SunWater Ltd. Other than the State of Queensland, each of the parties filed notices of appeal, although the plaintiff’s appeal was confined to a narrow issue. Final orders have been made with respect to the plaintiff itself in Rodriguez (23), [3] in an amount (payable by Seqwater) in excess of $100,000, and in Rodriguez (29) with respect to three other members of the class, whose claims raised particular issues not raised by the plaintiff’s case. These matters are relevant to whether, and to what extent, Seqwater required leave to appeal, discussed below.

4. [5]

   Shortly before the appeals were listed for hearing, Rodriguez (in its representative capacity and in its own right), SunWater and the State of Queensland settled all claims between them. As a consequence, the only outstanding issues are those arising between Rodriguez (in its representative capacity and in its own right) and Seqwater. The issues for determination by this Court fall within a far narrower compass than those confronting the primary judge.

5. [6]

   The trial judgment being divided into 15 chapters, it was inevitable in such a large factual case that findings would be made on specific issues which were then relied upon in the assessment of higher level issues, including ultimately the resolution of questions as to the existence and scope of the duty of care owed by each defendant, whether the duty was breached by each defendant and its employees, whether such breaches caused the harm suffered by the plaintiff, and the assessment of damages. At each of these stages, the judgment meticulously cross-referenced both later and earlier findings, so that the trail of the reasoning was readily ascertainable. That was as true of the complex technical evidence and the computer simulations relied upon by the plaintiff as it was of the factual circumstances as they occurred in January 2011. The tasks to be undertaken by this Court have been immeasurably assisted by the care and ultimately the clarity of the primary judgment.

2. Issues on appeal

1. [7]

   As the primary judge noted, practically no aspect of the case presented by the plaintiff was not in dispute at the trial. For example, each defendant challenged the claim that it owed a duty of care to the plaintiff and group members. When Seqwater filed its notice of appeal, it challenged findings by the judge as to its duty of care. However, those challenges (grounds 4 and 5) were abandoned shortly prior to the hearing of the appeal. Nevertheless, there remained 30 grounds, some with sub-grounds. A further amended notice of appeal, filed on 10 May 2021, included two new grounds 31 and 32. The issues in dispute on the appeal primarily focused on the applicable standard of care and the findings as to breach of duty.

2. [8]

   As the primary judge explained, three factual issues predominated at trial, namely that: [4]

3. [9]

   The activities of the flood engineers were regulated by a Flood Operations Manual (the Manual) which had been revised in November 2009. [5] The plaintiff’s case was identified by the primary judge in broad terms as follows:

4. [10]

   Three issues as to the Manual arose on this appeal. The first was a matter of legal principle, namely how to construe the Manual. Was it to be construed by the court as a legal document, or was the correct approach to determine whether the conduct of the flood engineers was consistent with an understanding of its terms which was reasonably open in the circumstances? Because, as will be explained, the Manual did not create legally enforceable standards, the latter approach should be adopted.

5. [11]

   The second and third issues involved specific issues in the application of the Manual. The second issue was whether it was reasonable to construe the Manual as not permitting reduction of the level of Wivenhoe Dam below Full Supply Level (FSL) in anticipation of a possible future flood event. If that were so, the plaintiff’s case as to releases between 2 and 6 January 2011 was difficult to uphold. The third issue was whether assessments of the appropriate “strategy” (as described in the Manual) for managing the dams during a flood event were to be undertaken by reference to expected inflows, without taking account of likely outflows. The strategies were defined by the level to which it was expected water would rise. The plaintiff contended that, in assessing likely dam levels, the engineers were required to assume “no releases”; that meant estimating the likely level to which water would rise during the adopted forecast period on the basis that no water would be released.

6. [12]

   However, at the forefront of Seqwater’s case on appeal was the legal submission that the standard of care to be applied was not that of reasonable care under s 9 of the Civil Liability Act 2003 (Qld), but rather the attenuated standard required of public authorities under s 36 of the Civil Liability Act. Section 36, if applicable, imposed a higher hurdle for the plaintiff in establishing that Seqwater acted tortiously in the exercise of its functions. Establishing the standard by which Seqwater’s conduct was to be judged was a necessary preliminary to considering the factual elements of the claim in negligence. The primary judge found that s 36 of the Civil Liability Act was not engaged: as a result, the standard of care applied in judging Seqwater’s conduct was the ordinary standard of a failure to take reasonable precautions against a foreseeable and not insignificant risk of harm, as required by s 9 of the Civil Liability Act. However, if engaged, s 36(2) provided that an act or omission was not “wrongful” unless it was “so unreasonable that no public or other authority having the functions of the authority in question could properly consider the act or omission to be a reasonable exercise of its functions.”

7. [13]

   The challenge to the finding that s 36 was not engaged was identified as ground 1 in Seqwater’s appeal. Determining that issue is a preliminary and necessary step before addressing the factual basis of the alleged negligence. That in turn requires a consideration of the legal framework under which Seqwater operated.

3. Leave to appeal

1. [14]

   The issues sought to be raised by Seqwater undoubtedly warrant a grant of leave to appeal, if required: indeed, leave was not opposed and could be granted by consent. However, it is desirable to explain why leave may be required.

2. [15]

   The Supreme Court Act 1970 (NSW), s 101(2)(e), provides that an appeal shall not lie from an interlocutory judgment of the Court in a Division except by leave. Separately, s 103 requires leave for an appeal from a decision of any question or issue ordered to be decided separately from any other question or issue. Where the answers to the separate questions result in the proceedings being finally determined, it has been held that an appeal lies as of right. In Plymouth Brethren (Exclusive Brethren) Christian Church v The Age Company Ltd, [6] the principal majority judgment held that such an appeal was as of right, whilst noting that the issue had not been raised or argued. [7] That statement relied upon a discussion in the judgment of Brennan J in O’Toole v Charles David Pty Ltd, [8] a case dealing with questions directed to be answered separately by a judge of the Federal Court, in a matter which was removed to the High Court before any answers were entered in the record of the Federal Court and thus raising a question as to the availability of appellate jurisdiction under s 73 of the Constitution. The other member of the majority in Plymouth Brethren had been party to a decision which took a different view and expressly declined to revisit the question, on the basis that if leave were required it should be granted. [9] Support for the conclusion that leave was not required may be found in the dissenting judgment at [120]-[123].

3. [16]

   On one view, the question is to be resolved by the specific provision in relation to stated questions, namely s 103 of the Supreme Court Act. That section appears to remove the issue from the ambit of the generic requirement with respect to an interlocutory order. Giving priority to the specific provision, the need for leave does not depend upon whether it is possible, because of the way the question is answered, to make a final order consequential upon the answer, or whether that is not possible. However, at least on one view, the function of s 103 is to sidestep the issue which troubled Brennan J in O’Toole, which was whether the answer to the separate question affected the rights and liabilities of the parties, or whether the answer was non-conclusive, in the sense that the judge who reserved the question for separate determination would not be bound by the answer in resolving the proceeding: O’Toole at 258. Only the former would result in a judgment, decree or order from which an appeal might lie to the High Court: O’Toole at 259. If that were in fact the function of s 103, there would be no inconsistency with the general approach to interlocutory orders: whether leave was required would turn on the legal effect of the order. Thus an appeal would lie as of right where the answers did give rise to a final order, despite the unqualified language of s 103.

4. [17]

   The issue is complicated by the fact that the present matter involves a class action. Although a final order has been made as between the plaintiff and Seqwater, which should entitle Seqwater to an appeal as of right, the answers to the common questions have not given rise to final orders involving members of the class, other than three members, who are not parties to the appeal.

5. [18]

   The proceedings brought by the plaintiff in a representative capacity engaged Part 10 of the Civil Procedure Act 2005 (NSW). Part 10 deals with appeals in s 180, but only by providing for appeals by the representative party. It does not vary the rights or procedures governing appeals by individual defendants, nor appeals by a plaintiff in its own right. It does, however, permit Rodriguez to bring an appeal in its representative capacity, which it did with respect to a limited issue concerning the calculation of damages. Presumably because an appeal is a fresh proceeding, s 180(5) appears to envisage that the class members have the right to opt out under s 162: that issue was not addressed in the present case. [10] The Uniform Civil Procedure Rules 2005 (NSW) (“UCPR”), Pt 58 does not take the matter any further. Nor does s 180 address the capacity of a representative plaintiff to be a representative respondent to an appeal.

6. [19]

   Similar legislation with respect to class actions is found in Pt IVA of the Federal Court of Australia Act 1976 (Cth), and Pt 4A of the Supreme Court Act 1986 (Vic). However, the issue as to the nature of a judgment which determines the final outcome for the plaintiff, but not for class members, appears not to have arisen.

7. [20]

   The correct view may be that to the extent that the appeal seeks relief with respect to the plaintiff’s judgment against Seqwater, which is in an amount in excess of the threshold in s 101(2)(r) of the Supreme Court Act, it is an appeal against a final order and may be brought as of right. However, to the extent the relief sought challenges answers to common questions affecting the basis on which the claims of class members are determined, Seqwater requires leave, pursuant to s 103. Because the issues form a common substratum for the individual and representative claims, and because the representative claims collectively involve a sum many times greater than the threshold, there should be a grant of leave to address the answers to the common questions.

4. Jurisdiction and choice of law

1. [21]

   As the primary judge noted, the events in question all occurred in Queensland; the plaintiff is a body corporate located in Queensland, as are the defendant statutory authorities and the State of Queensland. The primary judge suggested that the Court’s jurisdiction to hear the matter may derive, at least in part, from the Jurisdiction of Courts (Cross-Vesting) Act 1987 (Qld), s 4(1). That section purports to vest jurisdiction in the Supreme Court of another state or territory with respect to “State matters.” A “State matter” is one in respect of which the Queensland Supreme Court has jurisdiction otherwise than under a law of the Commonwealth or of another state: s 3(1).

2. [22]

   However, there is no doubt that the New South Wales Supreme Court also has jurisdiction as to the subject-matter of the claims, which were claims in tort, including causes of action in negligence, trespass and nuisance. Jurisdiction over the defendants depended on service of the initiating process. So far as personal jurisdiction over the defendant Queensland entities is concerned, they were served with the initiating statement of claim and filed appearances. Accordingly, the jurisdiction of the Supreme Court did not depend on the Queensland Act, nor on the reciprocal s 9 of the Jurisdiction of Courts (Cross-Vesting) Act 1987 (NSW), providing that the Supreme Court may exercise jurisdiction in relation to a matter cross-vested by a law of another state.

3. [23]

   The real issue in this case was identification of the correct State law to be applied. There was no dispute that the substantive law of Queensland governed the liability of both the State and the Queensland authorities, including, so far as necessary, the Acts Interpretation Act 1954 (Qld) in relation to questions of liability. [11] That Act states that it applies to “all Acts” (s 2), “Act” being defined to mean, relevantly, an Act of the Queensland Parliament: s 6. Conversely, the Interpretation Act 1987 (NSW) does not apply to statutes enacted by other State legislatures. In any event, the only choice of law questions arose in relation to identification of the interest payable on elements of the damages awarded and the apportionment of costs. For reasons which will be discussed in relation to that topic, the law of Queensland was the applicable law.

5. Factual background

1. [24]

   To understand the extensive evidence concerning the storm event in south-east Queensland in January 2011, and the effects of releasing water from the dams, it is necessary to outline the general topography of the affected areas. This was described by the primary judge in the following terms: [12]

2. [25]

   A plan outlining the different catchments was conveniently set out by the primary judge. (Brisbane is shown on the far right of the map, downstream from Moggill gauge, which was the last measuring point for flows in the Brisbane River before they reached the urban areas of the city itself.)

3. [26]

   The total catchment area is in the order of 13,570km2 of which approximately half, some 7,000km2, flow into Wivenhoe Dam (some via Somerset Dam): Ch 2 [39], [40]. The remainder produces flows into tributaries below Wivenhoe. As may be seen from the plan, Brisbane River flows broadly south-east from Wivenhoe Dam until it meets the Bremer River at which point it turns to flow north-east through Brisbane to the sea at Morton Bay. The other major tributary is Lockyer Creek, which joins the Brisbane River below Wivenhoe Dam at Lowood.

4. [27]

   Somerset Dam was built first, construction being completed in 1956; the construction of Wivenhoe Dam was completed in 1984. The dams have three functions. First, and taken together, they are intended to supply water to south-east Queensland; secondly, they mitigate flooding in the Brisbane River valley and, thirdly, there is a small hydroelectric plant which feeds electricity into the south-east Queensland grid. The third function was irrelevant in the present case, except that the ownership of the dams vested in Seqwater did not allow that authority to control the hydroelectric plant or its operations.

5. [28]

   The distinction between the first and second functions was identified by the designation of a FSL for each dam. Although the focus of the case was on the operation of Wivenhoe Dam, one particular issue concerned the operation of Somerset Dam. Wivenhoe Dam was considerably larger than Somerset, having a storage capacity at FSL approximately three times that of Somerset. As will be seen from the plan, the outflow from Somerset travelled along the course of the Stanley River a short distance into Wivenhoe.

6. [29]

   The spillway of Wivenhoe Dam had a crest at an elevation of 57m. [13] Above the crest were five radial gates which could be raised to allow the release of water over the spillway. Full Supply Level for Wivenhoe was 67m. The top of the core of the dam was 80m but was described as “not resistant to overtopping”; [14] thus, once the dam level reached 80m there was an expectation of a structural failure with potentially “catastrophic consequences.”

7. [30]

   Overtopping was not likely to occur because the dam had an auxiliary spillway below 80m on which were situated three “fuse plug” embankments designed to erode once water flowed over them. Although the erosion was intended to occur in a “controlled manner”, erosion would result in a large and uncontrolled discharge of water. The “trigger levels” for the erosion of the three fuse plug embankments were stated as 75.7m, 76.23m and 76.78m respectively. However, the seeming precision of the trigger levels was qualified by the statement that initiation of the fuse plugs was expected to occur when the lake exceeded those levels by 0.1m-0.15m. [15]

8. [31]

   In broad terms, the water supply function of Wivenhoe was to be met by maintaining the reservoir volume at the FSL of 67m. The area above 67m was identified in the Manual as “temporary flood storage”. However, to avoid uncontrolled discharge of water, it was necessary to keep the water level below 75.7m. The strategies for management of the water level during a “flood event” were identified in the Manual and will be discussed below.

9. [32]

   As may also be seen from the Brisbane River Basin plan, there are two major tributaries of the Brisbane River below Wivenhoe Dam, but above the city of Brisbane. The first is Lockyer Creek, which flows into the Brisbane River at Lowood. There is a gauge at Lowood which measures combined flows from Lockyer Creek and Wivenhoe releases. The second tributary, the Bremer River, meets the Brisbane River some distance below Lowood. The combined flows from Lockyer, Bremer and Wivenhoe are measured at a gauge upstream from the major conurbation, known as Moggill gauge. Rates of flow likely to cause urban inundation are measured by the readings taken at Moggill.

6. Operation of the dams: legal structure

1. [33]

   The references to relevant legislation discussed below identify the provisions at the date of the events giving rise to the plaintiff’s claims, namely January 2011. The early (pre-2007) legislative scheme is not dealt with here, but in Part 10(2) in addressing the Flood Operations Manual.

2. [34]

   There is no dispute that the “ownership” of both Wivenhoe Dam and Somerset Dam was vested in Seqwater. That was effected by notices published in the Queensland Government Gazette of 26 June 2008, pursuant to a power conferred on the responsible Ministers under the South East Queensland Water (Restructuring) Act 2007 (Qld) (“Restructuring Act”), s 67. [16] Nothing was said to turn on the concept of “ownership” in this context. It at least provided a basis for Seqwater to exercise its statutory functions in circumstances where its juristic nature was not entirely clear. Various provisions of the Restructuring Act bore upon submissions regarding the engagement of s 36 of the Civil Liability Act and, in particular, whether Seqwater was a “public authority constituted under an Act”, and what functions it exercised in carrying out flood mitigation activities. It is convenient at this point to address that legislative framework.

3. [35]

   The purposes of the Restructuring Act were set out in s 3:

4. [36]

   Pursuant to s 6, four “new water entities” were established, including “the Queensland Bulk Water Supply Authority” (Seqwater). Section 6(2) stated that a new water entity was “not a body corporate” and did not “represent the State.” Nevertheless, each new water entity was stated to have “all the powers of an individual” including the power to enter into contracts, and the power to acquire, hold, dispose of and deal with property: s 7(1)(a) and (b). Further, “it” was said to have the powers given to it under an Act, but subject to any limitations under an Act: s 7(2) and (3). A new water entity was able to sue and be sued in the name given under s 6(1): s 7(4). [17] Section 14(2) provided that the entity must have a “board”, but was not constituted by the members of its board.

5. [37]

   Despite the express terms of s 6(2), Seqwater was undoubtedly a trading corporation for the purposes of s 51(xx) of the Constitution: see Communications, Electrical, Electronic, Energy, Information, Postal, Plumbing and Allied Services Union of Australia v Queensland Rail, [18] dealing with relevantly identical legislation establishing Queensland Rail. It is not aptly described as a corporation aggregate or a corporation sole, but it is clearly an entity having a separate juristic personality. It may well be a corporation for other purposes; however, the only question for present purposes was whether it was a public authority for the purposes of another Queensland statute, namely the Civil Liability Act. No party submitted that s 6(2), defining it as an entity other than a corporation, had any consequence for this purpose.

6. [38]

   Chapter 2, Pt 1 of the Restructuring Act, identifying the establishment, powers and functions of the new water entities, contained three provisions which were relied upon in considering whether the terms of s 36 of the Civil Liability Act were engaged; it is convenient to set them out in full.

7. [39]

   First, the generic powers of the new water entities were set out in s 7:

8. [40]

   The functions of the new water entities, including Seqwater, were identified as follows:

9. [41]

   Chapter 2, Part 2 of the Restructuring Act (encompassing ss 14-26) dealt with the appointment and membership of the boards of the entities; Part 3 dealt with senior executives. Each new water entity had a board which was “responsible for the way the entity performs its functions and exercises its powers”: s 15(1). A board was to consist of no fewer than two, nor more than five, members appointed by the “responsible Ministers”. The chairperson and deputy chairperson were appointed by the responsible Ministers (ss 17 and 18), who might also end the appointment of a person “at any time” for any reason or none: s 19(4). (The term “responsible Ministers” was defined to mean both “the Minister administering this Act” and “the Minister administering the entity.” [19] )

10. [42]

    Chapter 2, Part 4 recognised that a new water entity is a statutory body under the Financial Accountability Act 2009 (Qld): s 34(1)(a). Part 4 of the Restructuring Act dealt with “reporting and accountability”. Division 1 provided that each new water entity was a statutory body under the Financial Accountability Act and under the Statutory Bodies Financial Arrangements Act 1982 (Qld), and was a unit of public administration under the Crimes and Misconduct Act 2001 (Qld). Division 2 imposed obligations to report to the responsible Ministers. Division 3 provided for annual reports to be prepared in accordance with the Financial Accountability Act. Relevantly for present purposes, Div 4 provided for “Strategic and operational plans”.

11. [43]

    Division 4 was comprised of ss 44-51. Section 44 recognised that the requirements of the Division overlapped with those of the Financial Accountability Act. Section 45 provided:

12. [44]

    Section 47 recognised that agreement between the responsible Ministers and the board might not be reached before the start of a financial year, in which case a draft plan was taken to be the entity’s strategic or operational plan, until s 48 came into effect. That provision read as follows:

13. [45]

    Section 49 required that the entity “must comply with its strategic and operational plans for a financial year”, although there was power to modify a plan: s 50. Finally, s 51 provided for the contents of an operational plan (though not a strategic plan):

14. [46]

    Every financial year a new water entity other than the water grid manager (not being a commercial enterprise and not seeking to make a profit) must pay to the State an “annual return” in an amount determined by the Minister but not exceeding its estimated net profit for that year: ss 52-54.

15. [47]

    Chapter 2, Pt 6 envisaged that a responsible Minister might give a direction in the public interest which was not in the entity’s commercial interests and which was designated a “community service obligation”: s 56. Such obligations were to be set out in the operational plan: s 57(3). Sections 58-63 gave the responsible Minister a degree of direct control over the operations of a new water entity – most notably, s 61 permitted the Minister to give the board of an entity a written direction “in relation to the entity” where, “because of exceptional circumstances”, it was “in the public interest” to do so. To that extent, the obligation to carry out functions commercially (s 11) and the freedom from direction by the government (s 63) were partly qualified. Under Part 9, when a new water entity expires, 99 years after its creation, the State is to be its successor: s 64. The entity was not liable to pay specified State taxes: s 81.

16. [48]

    Chapter 3 Part 4 dealt with matters incidental to the application to the new water entities of the Water Act 2000 (Qld) (Water Act) and the Water Supply (Safety and Reliability) Act 2008 (Qld) (Water Supply Act). Sections 72 and 73 provided that each entity other than the water grid manager was “a water service provider” and was to be registered on the register of water and sewerage service providers required to be kept by “the regulator” under the Water Supply Act, s 13, although the ordinary procedures governing applications for registration [20] did not apply to the new water entities. For the most part, however, Seqwater’s operation of Somerset and Wivenhoe was authorised and regulated by the unqualified operation of those Acts.

17. [49]

    Returning to the terms of s 9(2), it may be noted that there was no express reference to the flood mitigation functions of the dams. However, the term “water activities” in s 9(2)(a) was defined in the Dictionary (Sch 3) to mean activities mentioned in the definition of the same term in the Water Act. That definition was in the following terms:

18. [50]

    The Water Act, s 19 stated that “[a]ll rights to the use, flow and control of all water in Queensland are vested in the State”. It is an offence for a person to “interfere with water” unless authorised to do so under that Act or a similar law of another state or territory: s 808(2). However, by s 107A(1) the holder of a “resource operations licence” is authorised “to interfere with the flow of water to the extent necessary to operate the water infrastructure to which the licence applies.”

19. [51]

    The Water Act, s 38 permitted the relevant Minister to prepare a “water resource plan” for any part of Queensland, for purposes which might include “defin[ing] the availability of water for any purpose” and “provid[ing] a framework for sustainably managing water and the taking of water”. After the consultation process set out in ss 39-49A, including the preparation and publication of a draft plan, the Minister might prepare a final water resource plan, which would have effect when approved by the Governor in Council: s 50. Section 95 then permitted the chief executive of the relevant department to prepare a “resource operations plan to implement a water resource plan for any water in [all or part of] the plan area”. A draft resource operations plan, which would become a final resource operations plan through the operation of s 103, was to (among other things) “identify any water infrastructure to which the draft plan is intended to apply and how it will be operated”: s 98(1). Resource operations licences were to be granted by the chief executive, in accordance with the resource operations plan, on the day the plan comes into effect: s 107. If the plan set out a process for the granting of licences to meet “future water requirements”, licences might also be granted in accordance with that process: s 108. A resource operation licence could be held by the owner of the water infrastructure to which the licence applied or, if the owner were a subsidiary, its parent company: s 107A(3).

20. [52]

    Seqwater held a resource operations licence for the Stanley River Scheme under the Moreton Resource Operations Plan (Operations Plan), which implemented the Water Resource (Moreton) Plan 2007. The licence relevantly provided:

21. [53]

    Attachment 5 to the Operations Plan described Wivenhoe and Somerset Dams. Chapter 5 was largely concerned with water allocation and water sharing rules in the Central Brisbane River and Stanley River Water Supply Schemes, but it also contained s 72, which set minimum operating levels for infrastructure in the schemes and provided that releases could not be made from that infrastructure unless “necessary” to “meet minimum flow rates specified in section 75” or “supply downstream demand”. There was no reference to releases for flood mitigation purposes but neither party suggested that they were not necessary for either purpose, or that Seqwater was in breach of that condition. Seqwater’s position was that s 72 did not apply to water above the FSL of the dams.

22. [54]

    There is a question whether the phrase “as detailed in the Plan – Chapter 5 – Central Brisbane River and Stanley River Water Supply Schemes” modified “to interfere with the flow of water”, in which case the grant authorised interference in the manner detailed in that chapter, or whether it merely described “the Stanley River Water Supply Scheme”. The better view may be that it did the latter. Subject to condition 3, condition 1 of the licence required the holder to comply “with the operating arrangements and supply requirements detailed in the Plan Chapter 5”. Condition 3 applied the “transitional arrangements” in schedule 1 of the licence in the event Seqwater was unable to comply with the requirements of the plan on the day the plan commenced. The transitional arrangements relevantly required Seqwater to comply with any “interim program” approved by the chief executive under s 13 of the plan. None of that is consistent with the clause beginning “as detailed” itself imposing an unqualified condition on the authority of the licence holder. That result would also sit awkwardly with the provisions of the Water Act concerning the imposition of licence conditions (ss 109, 110) and the consequences for licence holders of noncompliance: ss 119, 813.

23. [55]

    In the event, Seqwater was unable to comply with the requirements of the plan, including s 72, and an interim program was approved. The interim program stated that, despite s 72, Seqwater would “continue to make releases from infrastructure for ... flood mitigation”. That part of the program had an indefinite duration and remained in effect in January 2011. It follows that by its licence Seqwater was authorised to interfere with the flow of water to the extent necessary to operate the dams, including for flood mitigation purposes, and that authorisation was not conditioned by s 72.

24. [56]

    The provisions of the Water Supply Act relevant to flood mitigation were carved out of the Water Act by the enactment of the former. They were relevantly as follows. Section 370(1) provided that the owner of a dam might be nominated by regulation “as an owner who must prepare a manual (a flood mitigation manual) of operational procedures for flood mitigation for the dam”. As the explanatory note to the Water Supply (Safety and Reliability) Bill 2008 explained, “a dam nominated in the regulation will be a dam which was constructed for the purpose of flood mitigation”, and a flood mitigation manual “ensures that such dams make controlled releases of water for flood mitigation purposes in accordance with pre-agreed conditions”. Responsibility for approving and, if necessary, amending the manual rested with the chief executive, who was empowered to consult with an “advisory council” before doing so: ss 371, 372. Unlike the Brisbane and Area Water Board Act 1979 (Qld), which expressly required the Water Board (the former owner and operator of Wivenhoe and Somerset Dams) to prepare and follow flood mitigation manuals, [21] the Water Supply Act did not expressly require an owner or operator of a dam to comply with its flood mitigation manual. However, s 374(2) provided that the owner and operator, and its employees and agents, would not “incur civil liability for an act done, or omission made, honestly and without negligence in observing the [manual’s] procedures”. Where that immunity was engaged, “the liability attaches instead to the State”: s 374(3).

25. [57]

    It appears that at the time the relevant version of the manual was approved there was no regulation in force pursuant to Water Supply Act, s 370 requiring Seqwater to produce a flood mitigation manual. Section 589 of the Water Supply Act did not give continuing effect to subordinate legislation made under the equivalent provisions of the Water Act (relevantly the former s 496, in the same terms as s 370). It is therefore not clear whether Seqwater was actually required to produce a manual for approval, but in any event there is no question that the relevant version of the manual was in fact approved under the Water Supply Act, s 371.

26. [58]

    Against the background of this statutory scheme, it is convenient to turn to the defence pleaded by Seqwater invoking s 36 of the Civil Liability Act: the rejection by the primary judge of this “defence” being the first ground of the appeal.

7. Civil Liability Act – s 36 (Ground 1)

1. [59]

   Section 36 of the Civil Liability Act appears in Ch 2, Pt 3, headed “Liability of public and other authorities and volunteers”. [22] It contains two brief definitions relating specifically to Division 1 of Pt 3. They are as follows:

2. [60]

   Although the primary judge put to one side the operation of s 35, it is convenient to set the provision out at this stage:

3. [61]

   Section 35 adopted “principles” to be applied in determining whether a public or other authority has breached a duty; at the very least, the principles are relevant as part of the statutory context in which s 36 is located. In particular, s 35 expressly recognised that an authority may allocate financial and other resources to the exercise of particular functions, that the functions are to be “decided” by reference to “the broad range of its activities”, and that an authority may be expected to act in compliance with general procedures and standards.

4. [62]

   Section 37, being the final section in Div 1, reads as follows:

5. [63]

   This provision is of more limited relevance, but it illustrates the need to have regard to the knowledge of the authority of particular risks, the materialisation of which may give rise to harm and, at least implicitly, to the need for principles of attribution to determine the scope and extent of such knowledge.

6. [64]

   Against this contextual material, the operative provision relied upon by Seqwater was s 36, which provides:

7. [65]

   The primary judge commenced by dealing with an issue raised by Rodriguez, namely that s 36 was confined to cases where the alleged liability was based on a cause of action for breach of statutory duty. [23] In support of that contention, Rodriguez had relied upon a judgment of Dalton J in Hamcor Pty Ltd v Queensland. [24] Hamcor had upheld such a limited operation of s 36; Seqwater contended that Hamcor was wrongly decided, a submission which the primary judge accepted. [25] (Hamcor was appealed but the operation of s 36 was not addressed as the judge’s finding as to the operation of s 36 was not in fact dispositive. [26] ) On this appeal, Rodriguez filed a notice of contention re-agitating its submission below that Hamcor had been correct on this point, and that, as the relevant claim against Seqwater was in common law negligence (not for breach of a statutory duty carrying the remedy of damages for loss caused to an individual by the breach), s 36 was not engaged.

8. [66]

   In rejecting the reasoning of Dalton J, the primary judge noted that she had applied a rule of construction derived from the judgment of Kitto J in Board of Fire Commissioners (NSW) v Ardouin. [27] Thus Dalton J had stated:

9. [67]

   Although this passage followed her conclusion, it was apparent that Dalton J applied a “jealous interpretation” to the words of s 36. Noting that the heading to the section referred to proceedings against public or other authorities “based on breach of statutory duty”, she read down the operative terminology (which does not refer to duty) as limited by the reference to breach of statutory duty in the heading. Justice Dalton stated: [28]

10. [68]

    Whilst the heading forms part of the statute, [29] and can be relied upon in construing the operative provision, the manner in which it was applied is not persuasive. The term “duty” is defined in the Dictionary to the Civil Liability Act to mean a duty of care “in tort”, “under contract”, or another duty under statute or otherwise that is “concurrent with a duty of care in tort or under contract”. The fact that the defined term “duty” is not used in s 36 suggests that the section was not limited to any particular cause of action. As the primary judge observed, “the definition of ‘function’ in s 34 is not confined to a ‘duty’.” [30] Further, the phrase “breach of statutory duty” used in the heading is at best a paraphrase of the third limb of the definition of duty, the meaning of which is not without its own difficulties. Finally, s 36(1) makes explicit provision for the application of the section, and does so by reference to “function” rather than the type of duty which is alleged to have been breached.

11. [69]

    The better view is that s 36 should not be read down as applicable only to a proceeding in which the cause of action is a breach of statutory duty. The reasoning of the primary judge was correct.

12. [70]

    It is also doubtful that the principle of statutory interpretation adopted by Kitto J in Ardouin provides assistance in this case. The statutory provision in issue in Ardouin was s 46 of the Fire Brigades Act 1909-1956 (NSW) which provided:

13. [71]

    Taylor J stated: [31]

14. [72]

    In short, too much weight should not be given to the language of Kitto J presuming that words were chosen with “complete precision” in circumstances where all members of the Court complained of poor drafting. In identifying the “strict meaning” that should be given to the immunity, Kitto J stated in language reflected in the other judgments: [33]

15. [73]

    That distinction drawn in Ardouin may readily be applied to s 36. Thus, Seqwater would not obtain the protection of s 36 if a flood operations engineer, making an urgent trip in his work vehicle to undertake flood control work, had an accident as a result of negligent driving. On the other hand, a decision to release or not release water, clearly a power conferred on the authority which was capable of exercise only by a body having the necessary statutory function, would attract the protection. The powers in issue were not those which must generally attach to a corporation or authority as a result of its institutional existence. Rather, the powers in question fell squarely within those accorded protection under Ardouin. These were undoubtedly functions of the authority conferred for the core purposes for which it was established.

16. [74]

    The judge’s determination that s 36 was not engaged relied on two propositions. The first involved acceptance of Rodriguez’ submission that the reference to “functions” in s 36 did not apply to any statutory functions of Seqwater. In particular, the primary judge noted that the chapeau to s 9(2) of the Restructuring Act, conferring functions on water entities, referred to “the following functions to the extent they are consistent with its operational and strategic plans”. He held in Ch 2 that as Seqwater had been unable to produce any such plans it had not established that s 9(2) was engaged:

17. [75]

    As discussed above, Seqwater was an authority established by statute, namely the Restructuring Act, which established four new water entities. The subsequent provisions of the Act applied generically to all new water entities, except where otherwise indicated. Relevantly for present purposes, the powers of each were defined in s 7, set out at [39] above. Further, the functions of the new entities were generically conferred under s 9, set out at [40] above. They included, by reference to the definition of “water activities” in the Water Act, flood prevention and floodwater control. They also included water supply services.

18. [76]

    Although the primary judge held that s 9(2) did not confer any functions on Seqwater, that did not leave Seqwater bereft of relevant functions; rather the judge held it was able to conduct activities in relation to the operation of the dam resulting from its ownership of the dam. As discussed above, it held a resource operations licence which authorised it to interfere with the flow of water in the rivers for flood mitigation purposes. The whole case against Seqwater assumed it had such a function, and that it was required to exercise its powers in accordance with a flood mitigation manual.

19. [77]

    The non-engagement of s 9(2) would not necessarily have provided a basis for excluding the operation of s 36 with respect to Seqwater. Even if the functions referred to in s 36 must be those derived from statute, not being powers available to any person in the exercise of ordinary activities, it is nevertheless apparent that the functions exercised by Seqwater were dependent upon the statutory creation of Seqwater as the owner of the dam. However, the reasoning that no functions were conferred on Seqwater under s 9(2) should also be addressed.

20. [78]

    The chapeau to s 9(2) conferred functions on each new water entity created by the Act, “to the extent they are consistent with its operational and strategic plans”. The primary judge found that Seqwater had neither a strategic plan nor an operational plan in January 2011. Although Seqwater challenged that finding on the appeal, it produced no evidence to support a conclusion that such a plan had existed. However, it does not matter whether that finding of the primary judge is accepted: the question is what effect that finding had in relation to the operation of Seqwater and, in particular, the conferral of functions under s 9(2).

21. [79]

    Rodriguez’ construction of s 9(2) turned on the proposition that the identified functions could only exist if “consistent with” operational and strategic plans and, therefore, if the entity had such plans. Only once the entity had such plans, and only during the period in which it had such plans, did it have the identified functions. If the provision had used the phrase “to the extent they are not inconsistent with” the plans, it would be understood that the functions were held absent inconsistency. That criterion could be satisfied in the absence of plans, but to confer functions only to the extent they are consistent with plans would require the existence of plans. Thus it was submitted, “consistent with” does not carry the same connotation as “not inconsistent with”.

22. [80]

    The underlying syntax of this construction may be accepted; however, the construction should not be accepted if the effect is to restrict the activities and operations of the entity so as to deprive it of its role as defined by the Restructuring Act.

23. [81]

    Rodriguez, in seeking to uphold the construction which negatived the existence of such functions, noted that the entity would still have the ordinary powers of “an individual” pursuant to s 7. However, those powers were not at large. They were “subject to any limitations under an Act”, which must include the Restructuring Act: s 7(3). Thus, although Seqwater may have had the powers of an owner of land, it did not follow that it could use or manage the land in any way that it chose. More importantly, the significance of the functions conferred by s 9(2) must be viewed within the statutory and regulatory framework for the management of the water supply of south-east Queensland and for flood mitigation, particularly with respect to the residents of Brisbane and surrounding areas. To construe s 9(2) as not conferring any function on Seqwater unless it was, at any point in time, in compliance with the requirements of Pt 4 in respect of operational and strategic plans, would be to subvert the statutory scheme. There was no provision stating that consequence. That consequence cannot have been contemplated by the Parliament; there is nothing in the legislation from which it is possible to infer any such unexpressed intention. Established principles of statutory interpretation set out by Project Blue Sky [35] preclude the inference of such a consequence.

24. [82]

    On any view, Seqwater in fact undertook its flood mitigation functions by establishing a Flood Operations Centre and exercising its powers to release water in a controlled manner. In doing so it was, on Rodriguez’ own case, subject to a duty to apply the Flood Control Manual according to its terms. The Manual, discussed in detail below, imposed strategies for the operation of Somerset and Wivenhoe Dams. It is not possible to read the reference to a “function” in s 36 as not encompassing the flood mitigation activities undertaken in January 2011.

25. [83]

    In rejecting Seqwater’s reliance on s 36, the primary judge accepted a second contention put forward by Rodriguez, expressed in the following terms:

26. [84]

    The words “true vicarious liability” were said to have been taken from Rodriguez’ submissions, but had earlier [36] been sourced to the judgment of Fullagar J in Darling Island Stevedoring and Lighterage Co v Long. [37] Generally speaking, the cases in which a vicarious liability is imposed on someone who is not the immediate wrongdoer depends on the relationship between the two parties. Thus vicarious liability may be imposed on an employer for the conduct of an employee. However, other relationships, such as that of a principal and independent contractor, may not give rise to vicarious liability under the general law. That situation may be varied by statute. [38] Further, there are circumstances where the general law imposes what is described as a “non-delegable duty” on a particular party, which, it has been said, “enables a plaintiff to outflank the general principle that a defendant is not vicariously responsible for the negligence of an independent contractor.” [39]

27. [85]

    Fullagar J in Darling Island Stevedoring stated: [40]

28. [86]

    In other words, the liability of the employer depends on the wrongful act of the employee. In contrast, Kitto J in the same case adopted the “master’s tort” theory of vicarious liability which imputed to the employer the acts of the employee, but not the liability of the employee, thus requiring a breach of duty on the part of the employer. [41]

29. [87]

    Kitto J explained that the concept of “vicarious liability” was coined by Pollock in 1877 to describe liability for the act of another, not for the wrong or liability of another. [42] Kitto J explained that understanding in the following terms: [43]

30. [88]

    The approach of Fullagar J to the concept of vicarious liability was accepted as correct in Hollis v Vabu Pty Ltd, [44] and later affirmed in Prince Alfred College Inc v ADC. [45] However, it is not necessary to rely on that reasoning for present purposes.

31. [89]

    Liability based on the tort of the employee may be understood as an expansion of the principle stated by Kitto J, not a denial of it. As explained by the Full Court of the Federal Court (Davies, Gleeson and Edelman JJ) in Pioneer Mortgage Services Pty Ltd v Columbus Capital Pty Ltd: [46]

32. [90]

    In its defence, Seqwater pleaded reliance on s 36 in relation to each allegation of breach of duty. Accordingly, the primary judge returned to this question in Ch 11, in dealing with standard of care:

33. [91]

    The foregoing explanation of the judge’s reasoning must be applied to the functions of Seqwater. Although the analysis of the functions was not limited to determining whether s 36 was engaged, it was applied in this context. It is not entirely clear how the analysis fitted into the approach to s 36. On one view, it was necessary to determine the functions of Seqwater which were engaged in the present case in order to determine how s 36 operated with respect to the wrongful exercise of those functions. Thus, Rodriguez’ case appeared to be that the statement of functions in s 9(2) did not operate, so that the conduct of Seqwater was limited to the “Manual of Operational Procedures for Flood Mitigation at Wivenhoe Dam and Somerset Dam”, a document which was required to be applied by the owner of the dams, namely Seqwater.

34. [92]

    It is possible that the description of vicarious liability as “true vicarious liability” was intended to distinguish between the liability of an employer and other circumstances of vicarious liability. No doubt it is true that liability may be imposed on one party for a particular purpose, but not for other purposes. However, in the present context the significance of the distinction is not apparent. Rather, the judge’s reasoning set out at [83] above sought to contrast the language in s 16(1) of the Civil Liability Act, stating that “[a] person is not liable in negligence for harm suffered by another person as a result of a materialisation of an inherent risk”, with the language of s 36 which does not refer to absence of liability. Thus s 36 was said to address a “narrower topic”, namely the acts and omissions of the authority, not its liability.

35. [93]

    Finally, the primary judge also considered the operation of s 22 of the Civil Liability Act which provides as follows:

36. [94]

    With respect to this provision, the primary judge reasoned as follows:

37. [95]

    However, to say of a particular act or omission that it is not wrongful must entail the conclusion that the responsible authority will not be liable for that act or omission. Accordingly, to describe an act as “not wrongful” does not appear to have any different effect from describing the authority as “not liable” for any harm caused. The point of distinction must therefore be the fact that the subject-matter of consideration is the act, rather than liability. Thus vicarious liability is the liability of the authority, but the wrongful act is not, on this supposition, the act of the authority. It is possible that this reflects the distinction between the servant’s tort theory and the master’s tort theory noted above. Nevertheless, this description of the underlying philosophy of vicarious liability did not assist the plaintiff in the present case, for three reasons.

38. [96]

    First, the primary judge held that Seqwater owed a duty of care to the plaintiff. Thus all that needed to be attributed to Seqwater were the acts and omissions of its employees, which would have given rise to a breach of duty on its part if they were wrongful.

39. [97]

    Secondly, for the purposes of s 36(1) it is clear that the proceeding instituted by the plaintiff against Seqwater was based on an alleged wrongful exercise of the flood mitigation function of Seqwater. The term “is based on” does not identify a specific cause of action, nor necessary elements of a tort, but is wide enough to incorporate factual allegations which, if established, would demonstrate liability on the part of Seqwater. Those actions included the acts or omissions of its employees undertaking flood mitigation operations.

40. [98]

    Thirdly, the term “public or other authority” includes “the Crown” and “a local government” as well as any other public authority constituted under an Act. Each of those limbs identifies a juristic entity which must act through the agency of others. Section 36(1) therefore assumes a wrongful exercise or failure to exercise a function by one of those bodies, which will perforce be undertaken through the acts or agency of a real person. To remove from the scope of its protective operation the concept of wrongfulness as applicable to the acts or omissions of those through whom an authority must operate would be to remove such a large area of potential operation of the section as to subvert its apparent purpose.

41. [99]

    The primary judge considered that the negligence of the flood engineer would be attributed to the employer (Seqwater), so as to render Seqwater liable, but that s 36(2) would not be engaged because the act or omission of the flood engineer was not the act or omission of Seqwater or, perhaps, it was the act or omission of Seqwater but was not a wrongful act or omission of Seqwater.

42. [100]

    On either view, there is an element of circularity in this reasoning. The finding that Seqwater owed a duty of care with respect to flood mitigation and that the acts of the flood engineers employed by it were to be attributed to it should have led, even on the master’s tort theory, to liability on the part of Seqwater for breach of its own duty. In other words, the acts constituted potentially wrongful conduct of Seqwater, so as to engage s 36(1). The concern underlying the master’s tort theory was that it imposed liability without fault, because the master did not owe a duty which was breached; that concern is absent in the case of an authority acting through agents.

43. [101]

    Further, whether the judge’s reliance on s 22 accurately stated either the effect of s 22 or the scope of vicarious liability is by no means clear. If a public authority can only exercise a function through the medium of trained professionals, it might be surprising if the standard of care applied to it differed from that applied to its agents. However, if that which is attributed pursuant to vicarious liability is the wrongful conduct of the employee, then Seqwater would be liable if the engineer’s conduct is wrongful, and will not be liable if the engineer’s conduct is not wrongful. It is not clear what significance the operation of s 22 might otherwise have in considering the engagement of s 36. A more difficult question might arise were the flood engineer liable despite the element of protection granted by s 22, but Seqwater was not because it obtains the higher standard of protection provided by s 36(2). However, that possible anomaly does not arise because the engineers were not sued.

44. [102]

    Assuming that the functions of Seqwater included flood prevention and floodwater control, Seqwater was obliged to carry out those functions, having regard to the interests of Rodriguez and others with downstream interests in avoiding inundation. It employed professional engineers as its agents to carry out those functions. If the functions were carried out negligently, according to general law principles, Seqwater would be liable. However, the standard of care would be that identified in s 36(2). The contrary reasoning of the primary judge in this regard should not be accepted.

45. [103]

    The primary judge held that s 36 was not engaged without reaching a conclusion as to whether Seqwater was “a public or other authority”. However, both because it is not easy to isolate particular aspects of the operation of s 36 without some risk to the proper construction of the section as a whole, and because Rodriguez sought to re-agitate the issue on appeal, it is necessary to explain why this element in s 36(1) is satisfied in relation to Seqwater.

46. [104]

    The primary judge noted that there was “a substantial dispute” as to whether Seqwater was a public authority, [50] referring to the discussion of when a body may be a public authority in Re Anti-Cancer Council (Vic); Ex parte State Public Services Federation [51] and, through the discussion in that case, principles identified in Renmark Hotel Inc v Federal Commissioner of Taxation. [52] The judge’s conclusion was expressed in the following terms:

47. [105]

    The basis of the final observations was found in the earlier conclusion that the statutory functions of Seqwater “do not include flood mitigation and the conferral of permission to interfere with the flow of water and operate the dams cannot therefore be ancillary to any such function”. [53] The final observations reflected that reasoning. As noted above, that reasoning has been rejected.

48. [106]

    The application of s 36 is satisfied to the extent that the proceeding brought by Rodriguez was “based on” an alleged wrongful exercise of, or failure to exercise, a “function” of a public authority. Contrary to Rodriguez’ submissions, in determining when and at what rates to release water from each of the dams, Seqwater was exercising a relevant function under s 9(2), namely carrying out water activities, and was doing so under its statutory licence. It follows that, contrary to the plaintiff’s second submission, the function was a function “of” Seqwater.

49. [107]

    The third challenge raised by Rodriguez relied on the contention that Seqwater was not a public authority. The primary judge referred to a statement from the judgment of Mason CJ, Brennan and Gaudron JJ in Re Anti-Cancer Council: [54]

50. [108]

    Re Anti-Cancer Council concerned an attempt by a public service union to require the Industrial Relations Commission to entertain an industrial dispute between the union and the Council. The Commission had jurisdiction if the employees of the Council were “employed in any State instrumentality or other undertaking carried on by public authorities … under any State charter, statute, enactment or proclamation of the State of Victoria.” The High Court held that the Council, while not a State instrumentality, had a sufficient public aspect to constitute it a corporation within the terminology set out above and carried on activities under a State statute. Different statutory contexts will give rise to different characterisations of particular bodies. However, as in the case of Re Anti-Cancer Council, it is clear that a public authority for the purposes of s 36(2) need not be the State or an emanation of the State.

51. [109]

    The importance of context is demonstrated by recent cases in the United Kingdom. Much of the relevant case law in the United Kingdom has been directed to the question whether a particular body falls within the terms of s 6(3) of the Human Rights Act 1998 (UK) which makes it unlawful for a public authority to act in a way which is incompatible with the European Convention on Human Rights. For that purpose, a public authority is defined to include a court or tribunal, and “any person certain of whose functions are functions of a public nature”. In applying that criterion, the courts have distinguished bodies which are “governmental in a broad sense of that expression”, [55] or the functions of which are treated as public, from those “hybrid” bodies which have both public and private functions. Of the latter Lord Nicholls stated in Aston Cantlow:

52. [110]

    Similar criteria will be applied in considering whether an authority is a “public authority”. However, to determine that a body is governed by human rights principles will not necessarily determine whether, in another case, the same body will be subject to judicial review, or whether it is entitled to protection from liability in tort. [56]

53. [111]

    There has been a debate as to whether a non-statutory, non-governmental body exercising “public functions” may be subject to judicial review. The focus of debate has been the decision of the English Court of Appeal in R v Panel on Take-overs and Mergers; Ex parte Datafin Plc. [57] This Court has doubted that judicial review is available with respect to decisions of “private bodies which do not exercise functions conferred by government, whether under statute or otherwise”. [58] In NEAT Domestic Trading Pty Ltd v AWB Ltd, [59] the High Court held that a privately operated corporation with power to veto exports of wheat did not exercise authority “under an enactment” for the purposes of the Administrative Decisions (Judicial Review) Act 1977 (Cth) when refusing approval to a competitor. The majority held that the relevant statute did not prevent the company giving preference to its own commercial interests over the interests of an applicant. [60] The issue need not be addressed further as the question is not one of the availability of judicial review, nor is the juristic nature of Seqwater, or its functions, comparable with the corporate nature of AWB and its functions.

54. [112]

    In the course of oral submissions on the appeal, Rodriguez drew attention to two decisions of this Court dealing with the application of s 43A of the Civil Liability Act 2002 (NSW). Because there are aspects of that section which differ from s 36 of the Queensland Act, it is convenient to set the section out in full:

55. [113]

    The significant variation for present purposes is that s 43A, unlike s 36, is limited to the exercise of or failure to exercise a “special statutory power”, a term which is defined in s 43A(2). On its face, there is no such limitation to be found in s 36. However, if s 36 is read down so that it only applies with respect to the kinds of power identified in s 43A(2)(b), the difference may be one of little moment. As noted above, in Board of Fire Commissioners v Ardouin the High Court limited the scope of a protective provision to conduct for which the statutory authority was required, being authority which did not inhere in the creation of a corporate personality.

56. [114]

    In Puntoriero v Water Administration Ministerial Corporation, [61] a potato farmer sued the Water Corporation in negligence for the supply of contaminated water which had allegedly damaged his potato crop. The question was whether a protective clause in the Water Administration Act 1986 (NSW) excluded liability for loss or damage suffered as a consequence of the exercise of a function, including the release of water, if the function were exercised in good faith and for the purposes of executing the Act. [62] The Court held that the protective provision did not apply because, as explained by Gleeson CJ and Gummow J, “[t]he supply of water by the Corporation to the appellants was not the exercise of a function which of its nature involved [interference with persons or property].” [63] As further explained by McHugh J:

57. [115]

    It will be necessary to return to the application of these principles for another purpose; the present purpose is simply to note that the significant constraint in s 43A, referring to exercises of special statutory powers, may also impliedly limit the operation of s 36.

58. [116]

    While some degree of broad guidance may be obtained from the authorities, the relevant criteria will need to be determined in the context of the specific authority under consideration. For present purposes there are two statutes in question, namely the Restructuring Act (from which the institutional nature of Seqwater is to be derived) and the Civil Liability Act (providing a degree of protection from suits in tort). Whether Seqwater constituted a public authority depends on an analysis of its constituting legislation, namely the Restructuring Act. As has been noted, its title “The Queensland Bulk Water Supply Authority” appears to engage the term “authority” as found in the Civil Liability Act. As an entity which can sue and be sued, it has legal personality and, as found by the primary judge, it owed a duty of care to the plaintiff.

59. [117]

    It is aptly described as a “public authority” because it is established under statute, has the functions and powers conferred by the statute, has no corporators or individuals who would benefit from the exercise of its powers as members of a corporate body, is responsible for the supply of water and other services relating to the water industry and is run by a board appointed by the responsible Ministers. Although the characterisation of Seqwater is required for the purposes of the Civil Liability Act, the language of that Act in s 34(c), referring to “any public authority constituted under an Act”, cannot be read down so as to exclude a body having the characteristics of Seqwater.

60. [118]

    Rodriguez submitted that Seqwater’s obligation to undertake its functions “as a commercial enterprise” demonstrated it was not a public authority. That obligation presumably means that it must charge for the supply of water at a rate likely to cover the cost of supply and return an appropriate profit, although the actual exercise of those powers will be subject to the strategic and operational plans required to be approved by the responsible Ministers. That is merely an application of the “user pays” principle, which is commonplace in the provision of public services.

61. [119]

    Rodriguez further submitted that the effect of s 11(2), together with the requirement in s 54(1) that Seqwater provide an estimate of its “net profit” for each financial year, demonstrated that Seqwater was intended “to operate as a profit oriented trading enterprise” a factor which, it submitted, was relevantly decisive. [64] However, the facts that it was not described as a corporation, had no shareholders, reported net profit pursuant to a statutory definition of that phrase, and pursuant to accounting standards applicable to an entity under the Financial Accountability Act, and was to report to the responsible Ministers, belie the proposed characterisation. If it carried out its functions so as to return a profit, the sole beneficiary was the State’s Consolidated Revenue. This factor did not demonstrate it was not acting as a public authority, but rather the reverse.

62. [120]

    The functions of Seqwater demonstrate beyond real doubt that its dominant purpose was to supply water and regulate the flow of waters out of the dams for the benefit of residents and businesses in south-east Queensland. As the primary judge correctly stated, [65] the fact that it was required to carry out its functions “as a commercial enterprise” [66] did not prevent the functions being those of a public authority and carried out for the benefit of the public or a section of the public.

63. [121]

    The term “public authority”, as used in s 36, is not defined. Accordingly, it should be understood in the context of the regulation of water supply and flood control set out above. These considerations demonstrate that Seqwater was a public authority within the meaning of that term in s 34, and therefore in s 36 of the Civil Liability Act.

64. [122]

    The primary judge was in error in failing to find that s 36 was engaged with respect to the claim in negligence against Seqwater. Ground 1 in Seqwater’s notice of appeal must be upheld.

65. [123]

    Once it is established that s 36 applied to the proceeding brought by Rodriguez against Seqwater, it was necessary for Rodriguez to establish that the exercise, or failure to exercise, the relevant function constituted an “act or omission [which] was in the circumstances so unreasonable that no public or other authority having the functions of the authority in question could properly consider the act or omission to be a reasonable exercise of its functions”, pursuant to s 36(2). That provision sets a standard for determining whether any act or omission was wrongful. Such provisions have also been characterised as conferring a “qualified immunity”; [67] however, that term suggests the provision operates as a defence, which may be misleading. Rather, in a claim based on negligence, s 36(2) sets the standard to be applied by the Court in determining whether there has been a breach of the authority’s duty of care.

66. [124]

    The statutory history supports the conclusion that s 36(2) adopts the language of Wednesbury unreasonableness, being the level of unreasonableness required to invalidate a discretionary action of a government authority. [68] However, to source the language of s 36(2) to a ground of judicial review provides no definitive answer to the question as to how precisely the lowered standard is to be understood and applied. There are several points to be made in this regard.

67. [125]

    First, if the same question is to be asked in assessing a claim in negligence against the authority as would be asked in proceedings for judicial review, one would need to assess whether the act or omission was an invalid exercise of a discretionary power. It was this point which led to discussion of the English cases referred to in Curtis v Harden Shire Council [69] at [265]-[272]. However, the focus of Lord Hoffmann’s reasoning in Stovin v Wise, [70] from which the language of s 36(2) appears to have been drawn, was on whether the local authority owed a duty to take action. There is a risk of confusion inherent in the use of the term “duty” in this context. In administrative law, a distinction is drawn between a power to act (which contains an element of discretion) and a duty to act (which creates an obligation). In some circumstances the conferral of a power may be coupled with a duty to exercise the power, at least in particular circumstances. In principle, this use of the term “duty” is distinct from the common law concept of a duty of care, breach of which may give rise to a liability in damages. The latter can arise regardless of the validity of the act or omission in question. Rather, as explained by Mason J in Sutherland Shire Council v Heyman [71] the general law may impose a duty of care in the exercise of a power. The circumstances in which this will occur were identified by McHugh J in Crimmins v Stevedoring Industry Finance Committee [72] in the following terms:

68. [126]

    After referring to Stovin v Wise, [73] McHugh J continued in Crimmins:

69. [127]

    As a matter of statutory construction, the general law principle may be displaced: that is, the statute may evince an intention that a particular authority is not to be subject to a general law duty of care in carrying out (or failing to carry out) its statutory functions. Whether Seqwater was subject to a general law duty of care was an issue at trial, but no longer is. By abandoning grounds 4 and 5, Seqwater accepted that it was required to exercise due care in carrying out (or failing to carry out) its functions in relation to flood mitigation and the release of water from the dams.

70. [128]

    There is a third sense in which the concept of duty is used. By way of extrapolation of the general law, a statute conferring a power may impose an obligation, breach of which carries a liability in damages, independently of the common law tort of negligence. Rodriguez did not allege that such a cause of action for breach of statutory duty arose in the present case.

71. [129]

    The observations of McHugh J in Crimmins at [82], three years before the first Civil Liability Act, must now be qualified by the widespread existence of provisions such as s 36. [76] In any event, despite the differing rationales, the dichotomy between private law and public law notions of duty was never so clear or universal as suggested in Crimmins. A duty imposed by statute not to mislead or deceive in trade or commerce is an example. Thus, if the general law imposes an obligation to pay damages to those injured by the failure of a public authority to take a particular act, or to act in a particular manner, it may be said that the law imposes a duty to act, or to act in the prescribed manner. At that point, the distinction between the private law and public law concepts of duty is at least diminished. The point is illustrated by the summary in the judgment of Gibbs CJ in Heyman at 445:

72. [130]

    Mason J stated in Heyman at 459-460:

73. [131]

    It may be said that where the court has concluded there is a duty to exercise reasonable care by acting, or by acting in a particular way, the public authority is subject to a duty to take such action. However, the language of manifest unreasonableness operates, in public law, by reference to a discretionary power in circumstances where the court has no power to determine whether the power should or should not have been exercised, but only to decide whether the authority acted within the legal limits of its discretion in acting as it did. In accordance with public law principle, a decision of an administrative agency will be set aside if it has acted in a way which the law does not permit, a question which is to be determined by the court. The exercise of an available power, where choices must be made, is not for the court, but only for the authority, so long as it remains within the bounds of lawfully available choice. On that view, the effect of a provision such as s 36(2) is to remove from the court the power to determine what is and is not a breach of the duty of care, that is to determine what precautions a reasonable authority would have taken in the circumstances, thus limiting the court to the question whether the step actually taken or not taken by the authority fell outside the range of choice available as a matter of law. It has been said that, “the approach resembled the test for apprehended bias, in that rather than assessing reasonableness for itself, the court was required to ‘view the matter through the eyes of a responsible public authority’.” [77]

74. [132]

    The point of distinction is described in administrative law terms as the difference between review for legal error and review on the merits. As explained by Brennan J in Attorney-General for the State of New South Wales v Quin: [78]

75. [133]

    Often, in the absence of reasons, the approach to unreasonableness amounting in the words of Brennan J to “an abuse of power” will follow from an analysis of the kind applied by Dixon J to the Commissioner of Taxation in Avon Downs Pty Ltd v Federal Commissioner of Taxation: [79]

76. [134]

    Applying these principles, the distinction between the general approach to breach of duty under s 9 of the Civil Liability Act and that applicable to a public authority to which s 36 applies may be identified as follows: under s 9 it is for the court to determine whether, on the evidence before it, it is satisfied that the defendant failed to take precautions against a risk of harm which a reasonable person in its position would have taken; under s 36, the court must be satisfied that the authority, acting on its understanding of the relevant circumstances and applicable law, adopted an approach to the exercise of its functions which fell outside the range of reasonably available options. Following the observation in Curtis, counsel for Seqwater drew an analogy with a challenge based on a reasonable apprehension of bias, where the question is not whether the court has an apprehension of prejudgment, but whether the court is satisfied that a fair-minded observer with knowledge of the relevant facts might have such an apprehension. [80] Although the analogy should not be pressed too far, it provides an illustration of the different roles that the court undertakes.

77. [135]

    There is no assistance to be obtained by paraphrasing the statutory language, but there are two aspects of the language which warrant clarification. The court in Curtis adopted the following propositions:

78. [136]

    There has been little consideration of equivalent language in other jurisdictions, but in a case involving smoke from a controlled burn of land in a national park which ruined a nearby grape crop, McLure P stated (Buss JA agreeing): [81]

79. [137]

    McLure P was correct to avoid a paraphrase or substitution of the statutory language with other words. In particular, it would invite error to reformulate the statute by reference to subsequent explanations by the High Court of the unreasonableness standard in administrative law cases. Further, some concepts commonly used in the law do not lend themselves to exegesis: an example is the standard of proof in a criminal trial, beyond reasonable doubt. However, what can be done by way of clarification is to note the context in which the standard will be applied. Thus, where a person suffers injury caused by the conduct of another, it may appear on the evidence that the impugned conduct falls into one of the following categories, viewed prospectively:

80. [138]

    Each category may need reformulation where the complaint is of a failure to act. There are situations where morality and the law take different courses and the law does not impose a duty to act. However, that did not arise in this case. The purpose of the powers conferred on Seqwater to operate the dams involved management of the water supply for south-east Queensland and flood mitigation on the Brisbane River. It had only one tool to achieve both purposes, that was turning the tap on or off at the dams. No question of distinguishing which acts from omissions affected its legal responsibility.

81. [139]

    Viewed in this way, one can see a progression in terms of culpability: categories (i) and (ii) do not engage tortious liability in negligence in accordance with s 9 of the Civil Liability Act; (iii) engages liability under s 9, and (iv) alone will engage liability under s 36(2). Thus the schema suggested above reflects a scale of increasing culpability. As McClure P said, (iv) imposes a higher burden on the plaintiff than does (iii). Lack of reasonable care, assessed in accordance with s 9, does not demonstrate liability where s 36(2) is engaged.

82. [140]

    This exercise reflects the observations of Gleeson CJ in Plaintiff S157/2002 v Commonwealth of Australia, [82] dealing with the basis of judicial review in the face of a privative clause:

83. [141]

    Because the primary judge did not apply the s 36(2) standard in assessing Seqwater’s conduct, it will be necessary for this Court to consider that exercise. However, in the absence of any notice of contention, the extent to which this Court can make findings is limited. The limitations are discussed in part 18 below. That will attract further questions, including as to the appropriate identification of the relevant “act or omission”, whether singular or plural, for the purposes of s 36(2). Such matters are conveniently addressed in the factual context in which they arise. Ground 1 of Seqwater’s appeal should be upheld.

8. Vicarious Liability (ground 2)

1. [142]

   The second ground of appeal challenged the finding of the primary judge that Seqwater was vicariously liable for the conduct of its employees, Mr Malone and Mr Tibaldi, in conducting flood operations in January 2011. The ground was somewhat incoherent. It was said to arise only if s 36 of the Civil Liability Act was not engaged. If s 36 were engaged, it could only be because the engineers were acting as the agency through which Seqwater undertook its flood mitigation functions. Whether it was vicariously liable for their conduct had nothing to do with the operation of s 36. Nor was there any coherent explanation as to why, as its employees, it would not be liable for negligent conduct in the course of the carrying on their duties as employees. In any event, as it has been held that s 36 is engaged in the present case, the ground is not relied upon and nothing further need be said about it.

9. Negligence – Overview

1. [143]

   Because the wrong standard was applied, findings of breach of duty will need to be reconsidered. To the extent such an exercise is available, pursuant to Rodriguez’ limited notice of contention, it is convenient first to address the basis on which numerous breaches of duty were found.

2. [144]

   The plaintiff’s case at trial turned squarely on acceptance of the expert evidence of Dr Christensen, a civil and environmental engineer with 32 years of experience in hydrology, dams, flood control and hydraulic modelling. His expertise was properly accepted by the primary judge. His evidence was contained within a number of reports and he spent some 22 days in the witness box. Dr Christensen produced no fewer than 10 “simulations” setting out flood operations which, in his view, a reasonably competent flood engineer “would or must” have pursued, but based upon various assumptions and with varying temporal elements. Although there was much disputation over Dr Christensen’s methodology, some of these issues were resolved at an interlocutory stage and errors corrected. As to matters of substance, the primary judge accepted Dr Christensen’s evidence on a number of outstanding matters which are no longer in dispute.

3. [145]

   The case in negligence resolved itself into two main parts. The first was based on an assumption that the engineers were negligent in terminating the December 2010 Flood Event and closing down the flood operations centre on 2 January 2011, at a point when the level of Wivenhoe had returned to 67.1m, but there was further rain forecast. On the assumption that the flood operations centre should have remained open, Dr Christensen modelled the preferred operation thereafter in what was known as “simulation C”. That simulation involved anticipating future rainfall and dropping the level of the dam significantly below FSL (67m). Whether that course was available in conformity with the Manual was a major issue in the proceedings. The primary judge ruled that that course was available, a ruling challenged on the appeal by grounds 12 and 13.

4. [146]

   The second part turned on Seqwater’s proposition that the flood engineers were not negligent in closing down the flood operations centre which had operated during late December. In that event, Rodriguez accepted that a further flood event, with accompanying releases, could not commence, in accordance with the Manual, until the dam level rose above 67.25m. A second flood event was declared at 07:00 on 6 January 2011. At 21:00 a “gate operation strategy” was developed during that day and releases commenced in the late afternoon of 7 January. Dr Christensen prepared two simulations, accepted by the primary judge, namely simulations F and H, each of which commenced on 8 January 2011 and which did not differ materially.

5. [147]

   In Ch 1 the primary judge identified the key issues in Rodriguez’ case in the following terms:

6. [148]

   There were elements of professional judgment required of the flood engineers in various aspects of the assessments to be made as to appropriate releases from the dams. It followed that the acceptance of an appropriate simulation by way of a reasonable counterfactual did not necessarily establish that what the engineers in fact did was unreasonable. Three of the four flood engineers (Messrs Ayre, Malone and Tibaldi, but not Ruffini) gave evidence as to what they believed they were doing and were entitled to do at each of a number of relevant times when they were on duty. The judge did not accept much of that evidence, treating it as self-justifying reconstructions in circumstances where, at least to an extent, the engineers accepted that they had no recollection of actual thought processes, other than those recorded in situation reports and retained copies of computer modelling runs.

7. [149]

   However, even applying the s 9 standard of reasonable care, Rodriquez’ case in negligence needed to be assessed objectively by reference to what the engineers in fact did, rather than any memory or reconstruction of underlying thought processes. Their conduct was to be assessed objectively by reference to what they knew or ought to have known at the relevant times.

8. [150]

   The point of distinction may be illustrated by way of an example. The primary judge found that the engineers should have made releases from Wivenhoe Dam based on anticipated rainfall as predicted by four-day forecasts issued by the Bureau of Meteorology. If, as appears to have been the case on some days, releases were not made, the question is not so much why the engineers did not do so, but rather what the consequences would have been for ongoing flood operations had they done so. In not having regard to the 4-day forecasts, if that be the case, an engineer was deprived of information which was available and should have been taken into account. Whether the use or failure to use forecasts constituted a breach of duty did not turn on the engineers’ actual recollected thought processes.

9. [151]

   The finding that the engineers did not take reasonable care by not taking account of the four-day forecasts in determining releases did not fix upon any failure to use those forecasts in a way expressly required by the Manual. Rather, as found, it involved a failure to construe the Manual in a way which required that forecasts be taken into account, leaving to the flood engineers the task of determining the weight and significance that the available forecasts were to be given in the determination of strategies and water releases. Thus it was necessary to focus on the conduct of the flood engineers. It may have been unreasonable to read the Manual as not requiring that the forecasts be taken into account; alternatively, it may have been reasonable to read the Manual as not requiring that the forecasts be used in any particular way, but thereafter unreasonable to have given them little or no weight. Ultimately, what was to be established was not that Dr Christensen’s simulated flood operations were reasonable, but that no reasonably competent flood engineer in the actual circumstances facing the engineers could have acted otherwise.

10. [152]

    Further, as will appear below, it is important to identify the precise acts or omissions which constituted the alleged negligence. For example, if Rodriguez failed to establish that ending the flood operations on 2 January was negligent, then the failure to take various steps which Dr Christensen’s simulation C mandated between 2 January and 6 January cannot have been negligent. That latter proposition was accepted by the respondent during the appeal. [83] In circumstances where simulation C was not engaged, the next relevant counterfactual simulations (F and H) did not commence until 00:00 on 8 January. Similarly, if releases which commenced on 10 January and inevitably caused inundation of urban areas were necessary to save the dam structure, those acts were not negligent. Rather, the negligence might, on that scenario, be found in the failure to release larger amounts, not causing inundation, or causing less inundation, at an earlier point in time, namely on 8 and 9 January.

11. [153]

    Because the standard of appropriate conduct was found to be governed by the Manual, a finding which is not challenged, it is convenient to turn first to the proper construction of the Manual and how it applied at various times.

10. Flood Operations Manual

1. [154]

   In Rodriguez (22) Ch 3, the primary judge dealt with the Flood Operations Manual and the Flood Procedure Manual. The latter is not presently material. Of the former the judge said, “[t]he significance of the Manual to these proceedings cannot be overstated”: [2]. After setting out the content of the Manual section by section, the judge then turned to its construction, which commenced with the heading “Interpretative Approach and Reasonable Interpretations”. He noted:

2. [155]

   The difference encapsulated within this heading was most clearly identified in the following passage:

3. [156]

   It will be necessary to consider the effect of s 374 of the Water Supply Act in seeking to understand the legal significance of the Manual. Before turning to that issue it is convenient to set out the submissions of the parties, as identified by the primary judge, as they were in substance adhered to in the course of the appeal. Under the heading “Strict or Reasonable Interpretation?”, the primary judge set out those contentions as follows:

4. [157]

   The judge both resolved the question posed by the submissions of the respective parties, and disavowed the need to resolve the question. He stated first:

5. [158]

   The judge then eschewed the significance of this resolution of the debate:

6. [159]

   Each of the briefly stated conclusions was footnoted by reference to relevant sections of other chapters in which reasons were given for the particular findings. However, one consequence of the judge eschewing the need to determine which was the correct approach to the construction of the Manual was that no specific ground of appeal addressed the conclusion set out at [126]. Rather, in the section of the notice of appeal dealing with “Manual Construction” (and encompassing grounds 6-15) the appellant contended in a chapeau covering each of the grounds:

7. [160]

   As noted above, the geography of Ipswich, Brisbane and the Brisbane River means that there is a risk of floods, sufficient to justify spending millions of dollars constructing and maintaining Somerset and Wivenhoe Dams. The dams were built in such a way that if waters rise close to or exceeding the dam crest, there is a chance the structure will fail. That would cause catastrophic flooding in Brisbane and Ipswich.

8. [161]

   One purpose of both dams was to mitigate flooding – which is to say, speaking generally, to lower the peak of any flood. This is achieved by temporarily storing flood waters, and releasing them over a longer period of time, so that the peak outflow from the dam is less than the peak flow of floodwaters in the absence of the dam. All the water that flows into the dam catchments must, sooner or later, be released; the flood mitigation purpose is achieved by altering the timing of that release.

9. [162]

   Although the litigation emphasised the dams’ flood mitigation purpose, the dams also served the purpose of supplying water to Brisbane and south-east Queensland. This may be seen in the title of the report presented to the Queensland Parliament in 1934 by the Special Committee which was “appointed to Investigate and Report upon Brisbane Water Supply and Flood Prevention” as well as in the term “Full Supply Level” which is basic to the operation of Wivenhoe Dam in particular.

10. [163]

    Those two principal purposes are opposed to each other. Each dam has a finite capacity to hold water (it will be explained below why the capacity of Wivenhoe is slightly over 2,655,000Ml). That capacity can be used to store water for consumption. Alternatively, the dam may be kept empty, available for the temporary storage of flood waters. Maximising water storage would lead the dam to be kept as full as possible for as much time as possible. Maximising the dam’s capacity for flood mitigation would cause the dam to be left empty for as much time as possible.

11. [164]

    Those basic considerations entail the following consequences.

    1. (1)

       First, and above all else, each dam must be managed to prevent a catastrophic structural failure.

    2. (2)

       Secondly, so long as there is no threat to the structural integrity of the dam, the main flood mitigation purpose is to prevent urban inundation downstream.

    3. (3)

       Thirdly, there must be some mechanism for regulating the compromise between the dam’s water storage purpose and its flood mitigation purpose.

12. [165]

    It will be seen below that the Manual reflects those basal considerations. For the most part, it is convenient to focus upon the larger, downstream Wivenhoe Dam.

13. [166]

    Construction of Wivenhoe occurred pursuant to the Wivenhoe Dam and Hydro-Electric Works Act 1979 (Qld) (“Wivenhoe Act”), following severe flooding in 1974. When completed, the land which was “part of or relevant to” the project was vested in the Brisbane and Area Water Board (a body corporate created by s 9 of the Brisbane and Area Water Board Act 1979 (Qld)).

14. [167]

    Section 32 of the Wivenhoe Act mandated the preparation of “a manual of operational procedures in relation to each reservoir ... for the purpose of flood mitigation pending completion of the Wivenhoe dam project”. The manual was not effective until approved by the relevant Minister: s 33(1). Section 34 and 35 gave defences. Section 34 seems to have been limited to the construction phase of the project, but it provided that “[n]o person shall be held liable for damages claimed in respect of loss or injury alleged to arise from (a) the carrying out of flood mitigation procedures in accordance with the approved manual (as amended to the material time) prepared under section 32”. Section 35 provided that, inter alia, Brisbane and Area Water Board “shall not be liable, absolutely or vicariously (a) for flooding or sending water upon any land by reason of the construction or provision of any works, being part of the Wivenhoe dam project ... unless it be shown that the flooding [or] sending of water ... is due to or arose out of the negligence of such of them as would, but for this section, be so liable”.

15. [168]

    The Brisbane and Area Water Board Act 1979 (Qld) required the preparation of a manual of operational procedures, and expressly preserved the operation of the manual prepared under the Wivenhoe Act until that manual had ceased to be effective: s 106(2)(b). Sections 107 and 108 were as follows:

16. [169]

    In broad terms, there were two available views of the legal status of the Manual. On the one hand, it could be seen as imposing legally enforceable obligations on the flood engineers; on the other hand, it could be seen as a set of guidelines or best practices, with limited legal significance. The latter characterisation is sometimes identified as “soft law”. [99] That is not to say the terms of the Manual were without legal significance, but only that its statements did not mandate action or prohibit other action. Such documents take many forms: for example where large numbers of decisions are to be made on a daily basis (such as claims for social welfare benefits), a manual may seek to create consistency in decision-making; where decisions are to be made hurriedly and without time for reflection (such as use of firearms by police) a manual may provide guidance in advance as to when a particular course of action is appropriate. The Flood Operations Manual fell into a different category. Although it may have been expected that it would be applied only in rare circumstances and where there was ample time for reflection and consideration, it provided a list of priorities, and strategies to effect those priorities. Its terms were in part prescriptive and in part guided the exercise of professional judgment.

17. [170]

    The primary judge stated in opening that: [100]

18. [171]

    No regulation of the kind identified in s 370(1) was relied upon. Rather, the legal status of the Manual appeared to depend upon a series of transitional provisions as noted in part 6 above. Of immediate relevance, s 613 of the Water Supply Act provided that a flood mitigation manual approved under s 497 of the Water Act and in force immediately before the commencement of the Water Supply Act, was taken to be a flood mitigation Manual approved under s 371. [101] The revision of the Manual in force in January 2011 had been approved by the chief executive on 22 December 2009, such approval taking place by notice in the Gazette, published on 22 January 2010, pursuant to s 371(2) of the Water Supply Act. The approval was stated to be for a period of five years. [102]

19. [172]

    Otherwise, Ch 4, Pt 2 of the Water Supply Act, headed “Flood mitigation”, provided only for amendment of the Manual and for regular reviews. The final section, s 374 read as follows:

20. [173]

    Seqwater was entitled to the protection conferred by s 374(2), which, if engaged, would have transferred liability to the State. It would be engaged only if Seqwater acted “honestly and without negligence in observing the procedures” set out in the Manual. However, the dual requirement of both honesty and absence of negligence resulted in limited consideration being given to the operation of this provision. As noted above, the primary judge held that s 36(2) of the Civil Liability Act was not engaged in the present case; however, on the basis that it was engaged, the standard to be applied in determining whether Seqwater was negligent in observing the procedures in the Manual should have been the standard of Wednesbury unreasonableness identified in that provision. As the only relevant cause of action turned on the existence of negligence, there was no liability to attach to the State in the event that Seqwater obtained protection under s 374(2) because negligence was not established.

21. [174]

    For present purposes, s 374 is material because the primary judge relied upon it as a basis for conferring on the Manual a legal status requiring that it be given the construction properly to be accorded to a legal document. However, the preferable course is to identify the legal status of the document without regard to its consequences. That invites attention first to whether “as an instrument” it falls within the terms of the Statutory Instruments Act 1992 (Qld). A statutory instrument is one identified in s 7 of that Act:

22. [175]

    It is far from clear that the Manual fell within any of the particular categories listed in s 7(2), and no attention was paid to this issue in the course of the appeal. The section is to be construed in accordance with the Acts Interpretation Act 1954 (Qld). Section 2 of the Acts Interpretation Act says that it applies to “all Acts”; it is arguable, though again not beyond doubt, that the reference to an “Act” in the Acts Interpretation Act includes a reference to a statutory instrument: s 7(1).

23. [176]

    In any event, similar principles should apply to determining the intended scope and operation of the Manual whatever its precise legal status. Two factors suggest how a court should approach the construction of the Manual. First, the sole audience for the Manual was the group of flood engineers who would be operating the dams during a “flood event”. Secondly, as the evidence revealed, the Manual was drafted by the flood engineers, with the involvement of officers from the Department of Environment and Natural Resources. The Chief Executive of the Department was required to approve the final document. The manner of formulation is apparent from some of the language in the Manual, examples of which will be considered below.

24. [177]

    The Manual is an 86-page document. It describes itself as “Revision 7” and is dated November 2009. The primary judge dealt extensively with earlier revisions of the document, mostly in Chapter 4, especially insofar as they treated the role of forecast rainfall differently, but consideration of them may be deferred for present purposes.

25. [178]

    Notice of the approval of the revision of the Manual in place in January 2011 was published a year earlier in the Queensland Government Gazette. [103]

26. [179]

    The primary judge said, correctly, that “the parties were in sharp dispute about a vast number of issues concerning the construction and application of the Manual”, but that about the only matter as to which the parties were agreed was “the necessity for flood engineers to follow the Manual during flood operations save for the possibility of following its own procedures for departure from its requirements when the safety of the dams is threatened”: Ch 3 [2]. His Honour gave a lengthy description of the manual over some 37 pages in Ch 3 [5]-[101]. This was followed by a much longer analysis of the disputed provisions. In light of the narrowing of issues on appeal, a substantially shorter description will suffice.

27. [180]

    The Manual was replete with grammatical and syntactical glitches. Capitalisation was haphazard. The passages reproduced below are verbatim, with errors left uncorrected (and not marked “sic”). That is not said as a significant criticism. The drafting was by engineers, who were evidently and understandably much more focused upon its substance than its form.

28. [181]

    The first 50 pages of the Manual (putting to one side preliminary pages dealing with the history of revisions and the table of contents) comprises 10 sections. These are followed by 36 pages comprising 11 appendices.

29. [182]

    The Preface is best reproduced in full:

30. [183]

    It will be seen that section 1.1 of the Manual (i) stated that compliance with the Manual is mandatory, (ii) identified in descending order of importance five specified objectives, and (iii) contemplated relatively rapid (7 days) discharge of floodwaters.

31. [184]

    Section 1.2 of the Manual defined twenty terms, including:

    1. (1)

       “Flood Event” is “a situation where the Duty Flood Operations Engineer expects the water level in either of the Dams to exceed the Full Supply Level”;

    2. (2)

       “FSL” or “Full Supply Level” means “the level of the water surface when the reservoir is at maximum operating level, excluding period of flood discharge”;

    3. (3)

       “Flood Operations Engineer” means “a person designated to direct flood operations at the dams in accordance with Section 2.4 of this Manual”;

    4. (4)

       “Senior Flood Operations Engineer” means “a person designated in accordance with Section 2.3 of this Manual under whose general direction the procedures in this Manual must be carried out”, and

    5. (5)

       “Duty Flood Operations Engineer” means “the Senior Flood Operations Engineer or Flood Operations Engineer rostered on duty to be in charge of Flood Operations at the dams”.

32. [185]

    Section 1.7 provided that the Manual contained the operational procedures for Wivenhoe Dam and Somerset Dam for the purposes of flood mitigation and must be used for the operation of the dams during flood events. Section 1.5 stated that operating in accordance with the procedures in the Manual “shall give the protection from liability” provided by s 374 of the Water Supply Act.

33. [186]

    Section 2, titled “Direction of Operations”, dealt with the roles of Duty Flood Operations Engineer and Senior Flood Operations Engineer, which was a live issue between the defendants at trial (turning principally on the fact that Mr Ayre, an employee of SunWater, was the Senior Flood Operations Engineer) but was given less significance on appeal. Section 2.8 made provision for how the procedures in the Manual might be departed from (essentially, requiring the Senior Floor Operations Engineer to attempt to contact the Chairperson of Seqwater and the Director General of the Department of Environment and Resource Management). It was common ground that those steps were not taken in January 2011, although at one stage there was contemplation that they might be taken.

34. [187]

    Section 3, titled “Flood Mitigation Objectives”, identified five objectives explicitly listed in descending order in identical terms as found in section 1, including the paragraph immediately following the list. The section then stated:

35. [188]

    The last two sentences reproduced above are consistent only with the “forecast peak water level” being a reference to the actual forecast peak level during a flood, rather than a forecast peak which would occur if all of the gates were closed for the remainder of the flood event.

36. [189]

    Some explanation of each objective was given in sections 3.2–3.6. The catastrophic consequences of the structural failure of a dam were emphasised in section 3.2, and their structural safety “must be the first consideration” in flood mitigation operations. Section 3.3 stated:

37. [190]

    Sections 3.3-3.6 dealt with lower level objectives, including (in section 3.5) the desirability of the dams being full for water supply purposes following a flood event, and (in section 3.6) the requirement to give consideration during the drain down phase to minimising impacts on flora and fauna and, in particular, “strategies aimed at reducing fish deaths in the vicinity of the dam walls are to be instigated, provided such procedures do not adversely impact on other flood mitigation objectives.” (It will be seen that one aspect of flood operations on 2 January 2011, which the primary judge found to be a breach of duty, was the closing of the gates so as to enable volunteers to seek to return fish into the dam.)

38. [191]

    Section 4 identified four magnitudes of flooding: minor, moderate, major and extreme.

39. [192]

    Section 5, titled “Flood Monitoring and Forecasting System”, was mostly directed to the system of field stations within the within the five main catchments above and below the dams (see [25] above) which sent rainfall and river heights to the Flood Operations Centre. It identified the Real Time Flood Model (or RTFM) which was to be used to “estimate likely dam inflows and evaluate a range of possible inflow scenarios based on forecast and potential rainfall in the dam catchments”.

40. [193]

    Section 6 dealt with Communications, both between statutory agencies and with the public. Section 7 dealt with the review of the Manual.

41. [194]

    All sections have hitherto been short documents of no more than three pages. Section 8 was titled “Wivenhoe Dam Operations” and was 18 pages long. It was the largest section by far in the Manual. Most of the contested questions of construction or interpretation arose from this section. (The 18 pages of section 8 are reproduced as Appendix A to this judgment, but key aspects are noted below.)

42. [195]

    Section 8.1 included the following paragraph:

43. [196]

    Section 8.2 dealt with the two main ways (“Radial Gates and an Auxiliary Spillway”) in which water might be released during a flood event. The radial gates were controlled by the Flood Operations Engineer. Their arrangement is shown diagrammatically in Appendix A, Manual p 20.

44. [197]

    The Auxiliary Spillway comprises three plugs which are designed to fail shortly after the water height exceeds 75.7, 76.2 and 76.7m, and thereafter to release substantial flows of water in an uncontrolled way down three separate spillways until the depth returns to 67m. This is quite different from the operation of the radial gates. The release down each spillway cannot be altered, and is determined by the width of the spillway, as shown in the spillway rating table (taken from Appendix C of the Manual):

45. [198]

    Thus if the water reaches 76m and the first spillway plug is eroded, then the entire flood capacity of the dam – the volume of water between 67m and 75.5m (1,160,000Ml) – will be released down the first spillway, starting at a rate of around 1873m3/s, and thereafter reducing as the water level is lowered. If the water level reaches 76.5m and the second spillway plug is eroded as well, then the same will occur, save that this time the water will be released down two spillways, starting at a rate of in excess of 5400m3/s – which will of itself, irrespective of the other water being released from the radial gates at Wivenhoe not to mention the waters in the Lockyer and Bremer, cause damaging floods to urban areas of Ipswich and Brisbane. Potentially worse than that, until the fuse plug embankment is restored, the dam would have no flood storage capacity for any future flood event.

46. [199]

    The significance of the foregoing is threefold.

    1. (1)

       The balance of the section was directed to the aspects of releasing water which were in the control of the Flood Operations Engineer, namely, the five radial gates.

    2. (2)

       The fuse plugs and the auxiliary spillways introduce a further basal objective in flood operations, which is to prevent the water level from exceeding 75.7m and thereby prevent the destruction of the first fuse plug and at higher levels the second and third plugs, which could cause urban inundation, and which would, until the plug or plugs were restored, reduce the flood storage capacity of Wivenhoe to zero.

    3. (3)

       Consequently, although the dam is capable of storing water up to 80m, once the height exceeds 75.7m and the fuse plugs are eroded, the only way the water height will exceed that level is if inflows exceed the releases down the spillways. That is possible (during the rain events of January 2011, inflows exceeded 10,000m3/s on two separate occasions) but even if that occurs, there will continue to be uncontrollable releases down one or more spillways until FSL is reached. That in turn means that although Appendix C described the flood capacity of Wivenhoe Dam as 1,980,000Ml at a depth of 80m, a more significant level is 75.5m (20cm below the level at which the first fuse plug will erode). The flood capacity at 75.5m is only 1,160,000Ml.

47. [200]

    Section 8.3, headed “Initial Flood Control Action”, provided:

48. [201]

    Section 8.4, headed “Flood Operations Strategies”, identified four “strategies” for Wivenhoe known as W1, W2, W3, and W4. The strategies were “based on the Flood Objectives” of the Manual, and the five objectives from section 3 were repeated, once again in descending order of importance. The section reiterated that:

49. [202]

    The section continued:

50. [203]

    (A slightly different table was reproduced in Rodriguez (22) Ch 3 [39]. That appears to have been taken from a draft which was circulated in around September 2009 prior to the finalisation of the Manual and this superseded flowchart is also reproduced at Rodriguez (22) Ch 4 [121]). The table is the same, save that in the orange and red ovals are the words “Use Strategies W1 and W2 as appropriate”, “Use Strategies W1 and W3 as appropriate” and “Use Strategies W1, W3 and W4 as appropriate”. The rejection of those words in the Manual in its final form is of some slight significance in understanding the way each strategy operates. It was discussed by the primary judge at Rodriguez (22) Ch 4 [107]-[137].)

51. [204]

    Pausing there, two important things may be noted:

    1. (1)

       First, the choice of strategy depended on the likely level in Wivenhoe Dam, as well as (in the case of W2 and W3) the likely maximum downstream flows at Lowood and Moggill.

    2. (2)

       Secondly, it is clear that the strategy once selected was not fixed for the duration of the flood event. Rather, the strategy which was selected might change from time to time depending on the up to date predictions of water levels and downstream flows. That was the force of the paragraph reproduced above commencing “Strategies are likely to change during a flood event ...”.

52. [205]

    Contrary to Seqwater’s submissions at trial, it is tolerably clear that the estimation of the likely level of water in Wivenhoe Dam is to be informed by rainfall forecasts. This is clear from the words “which are to be made using the best forecast rainfall and stream flow information available at the time” and “[s]trategies are changed in response to changing rainfall forecasts”. It is reinforced by many other references to rainfall forecasts and predictions in the Manual, discussed further below.

53. [206]

    Contrary to Rodriguez’ submissions and the construction adopted by the primary judge, it seems unlikely that the criterion separating the choice between strategies W2 and W3, which turns on the maximum flows at Lowood and Moggill, is to be assessed on the assumption that no water is being released from Wivenhoe. That does not make sense. The flow rates of 3500m3/s and 4000m3/s represent actual flow rates in the real world. They have been chosen because they represent thresholds above which there will be inundation of urban areas. No sensible purpose would be achieved by basing operational decisions during a flood event upon whether the critical flow rates at Lowood and Moggill would be exceeded if one assumed, almost certainly contrary to the fact, that no water was being released from Wivenhoe.

54. [207]

    The ensuing seven pages of the Manual identified each of the strategies, including substrategies W1A, W1B, W1C, W1D and W1E (within W1), and W4A and W4B (within W4). Each strategy commenced with a boxed heading in larger typeface identifying the “Primary Consideration” of that strategy. These are:

55. [208]

    Each strategy had “conditions” in a separate box and in bold typeface. Some of the “conditions” are requirements for the particular strategy to be applicable (notably, the predicted water level in the dam and to that extent reflecting the information in the flow chart). Other “conditions” described the operation of the strategy itself, including the maximum release rate and the purpose. Each repeated the substance of the “primary consideration” which had already been mentioned a few lines earlier in the document.

56. [209]

    Each strategy then contained a sentence commencing “The intent of [the relevant strategy]”, as follows:

    1. (1)

       “The intent of Strategy W1 is to not to submerge the bridges downstream of the dam prematurely (see Appendix I).”

    2. (2)

       “The intent of Strategy W2 is limit the flow in the Brisbane River to less than the naturally occurring peaks at Lowood and Moggill, while remaining within the upper limit of non-damaging floods at Lowood (3,500m3/s).”

    3. (3)

       “The intent of Strategy W3 is to limit the flow in the Brisbane River at Moggill to less than 4000m3/s, noting that 4000m3/s at Moggill is the upper limit of non-damaging floods downstream.”

    4. (4)

       “The intent of Strategy W4 is to ensure the safety of the dam while limiting downstream impacts as much as possible.”

57. [210]

    The substrategies within W1 and W4 all reflected considerations of infrastructure. The five substrategies W1A – W1E all turned on water levels in the Dam (67.25, 67.5, 67.75, 68 and 68.25) and maximum releases, of 110m3/s, 380m3/s, 500m3/s, and 1900m3/s. The latter were determined by the levels at which downstream bridges could be kept open. Within the substrategies there was further detail, based on the combination of flows from Wivenhoe Dam and Lockyer Creek, which were intended to keep specified bridges open.

58. [211]

    Similarly, the distinction between strategy W4A and W4B turned on whether the lake level would exceed 75.5m, and thereby trigger the first bay of the fuse plug. There is some slight significance in the drafting of these substrategies. W4A is expressed to apply when “Lake Level between 74.0 and 75.5m AHD” and W4B when “Lake Level greater than 75.5m AHD”. However, it is reasonably clear that both levels are predicted maximum levels. That is because strategy W4B was primarily directed to minimising the possibility of fuse plug initiation. It dealt in terms with a scenario described as “Potential to keep lake level below EL 75.5 by early opening of the gates and/or varying the operational procedure at Somerset”. The details of the substrategy do not matter; its significance is that, despite the apparent reference to the actual lake level being above 75.5m, the document only makes sense if that is understood as a reference to a prediction that levels may in the future exceed 75.5m.

59. [212]

    The description of strategy W1 concluded with the words in bold and large typeface:

60. [213]

    Section 8.5, headed “Gate Closing Strategies”, was a short section of some importance to the findings of breach on 2, 3, 4 and 5 January, and will be dealt with when dealing with the grounds of appeal challenging the judge’s findings relating to the steps taken on those days.

61. [214]

    Section 8.6, headed “Gate Operation Sequences”, explained over slightly more than five pages the way in which the five radial gates were to be opened. Save in extreme flood events, the section stated that the gates were to be opened in 50cm increments, one at a time, so that the middle gate (gate 3) was opened earliest and the gates at the edges (gates 1 and 5) were opened last. The intention is for “Flow in spillway to be as symmetrical as possible”. Some details of this section are relevant to particular submissions, but these may be deferred for present purposes.

62. [215]

    Sections 8.7 and 8.8 are short paragraphs dealing with modifications to the gate openings during and after the triggering of a fuse plug.

63. [216]

    Section 9 of the Manual dealt with “Somerset Dam Operations”. This is relevant to grounds 14 and 15, but its details are best deferred until those grounds are addressed. It followed the same general structure as section 8, including a “Somerset Flood Strategy Flow Chart” enabling the choice of one of strategies S1, S2 and S3, each of which had “conditions” and a statement of intent.

64. [217]

    Section 10 dealt with Emergency Flood Operations, including the possibility of loss of power and/or communications. It included table 10.2 which identified water levels, gate openings and the flow rate of water discharged, described as the “radial gate opening sequence”. In substance, it involved a progressive opening of gates, starting when the water depth is 67.5m when the first increment of gate 3 is opened (releasing 50m3/s) continuing to a depth of 75.3m when all five gates are fully open and 10,160m3/s is released.

65. [218]

    Appendix C, the most important of the appendices for present purposes, included the following table:

66. [219]

    It will be seen that water levels between 57m and 80m (in 50cm intervals) are correlated with storage capacity and flood capacity.

    1. (1)

       Storage capacity commences at 414,000Ml at 57m, rising to 3,132,000Ml at 80m. At 67m, the storage capacity is 1,165,000Ml.

    2. (2)

       Flood storage capacity is left blank at heights of 57m – 66.5m, is stated to be “0” at 67m, and thereafter rises to 1,980,000Ml at 80m.

67. [220]

    One aspect of the construction of the Manual turned upon the absence of entries in the column for “Flood Capacity” at depths below 67m; this was said to support a construction which precluded discharging flood water so as to reduce the dam level below 67m, an issue discussed below. For present purposes, it suffices to observe that the Manual identified the storage capacity of Wivenhoe Dam as 3,132,000Ml, of which 1,980,000Ml is “Flood Capacity”. (Of course, if water levels reached 80m, or anything above 75.7m, it was expected that one or more of the fuse plugs would be eroded.)

68. [221]

    The parties debated a “vast” (to use the language of the primary judge) number of issues of construction of the Manual. His Honour dealt with these with care and over literally hundreds of paragraphs. It was necessary to do so because issues of construction bore upon the assumptions underlying the simulations modelled by Dr Christensen, as well as the assessment of breach of duty.

69. [222]

    As noted at the outset, this appeal raises many fewer issues than were presented to the primary judge. It is convenient to identify relevant principles before turning to the issues raised by specific grounds of appeal.

70. [223]

    First, the need for “clearly defined procedures” in a flood event is explicitly stated in the Manual’s opening paragraph. It was obviously undesirable for flood operations engineers to be debating the meaning of the Manual during a flood event. While there is room for argument on a number of issues, that debate should not obscure the fact that some things are fundamental and beyond argument.

71. [224]

    Secondly, the Manual was, as Seqwater submitted and Rodriguez did not deny, “a practical document addressed to engineers, not lawyers”. It falls within the principle that “documents addressed to practical people skilled in the particular trade or industry” ought “to be construed in light of practical considerations, rather than by a meticulous comparison of the language of their various provisions such as might be appropriate in construing sections of an act of Parliament”. [104]

72. [225]

    Thirdly, that approach accords with orthodox approaches to construction. Very commonly, contracts, wills, deeds or other legal documents give rise to a host of difficulties, not all of which need to be resolved in order to determine the particular dispute arising in litigation. The point was made in this Court, in a contractual context, in Zhang v ROC Services (NSW) Pty Ltd [105] by reference to Professor Carter’s work: [106]

73. [226]

    The Manual identified four key volumes of water in the dam by reference to water levels plus a range of flow rates at Lowood and Moggill. The procedures contained in the Manual depended upon those integers relating to conditions in the real world.

74. [227]

    The water levels of 74m and 68.5m referred to predictions used to identify the primary objective and the particular strategy to be deployed. If predicted dam levels were to exceed 74m, then the primary objective was to preserve the dam. If predicted dam levels exceeded 68.5m, then the primary objective was to prevent urban inundation. If the predicted dam levels were lower, then subordinate flood mitigation objectives (such as preserving the downstream bridges) came to the forefront.

75. [228]

    So much is clear from, inter alia:

    1. (1)

       the reiterated hierarchy of objectives in descending order of importance;

    2. (2)

       the flow chart reproduced above guiding decision making;

    3. (3)

       the existence of four strategies, reflecting different primary considerations, delineated by water levels in the dam and flow rates downstream;

    4. (4)

       the heavy emphasis of the “primary consideration” of each strategy and its intent; and

    5. (5)

       the bold mandatory words at the end of the description of strategy W1 “If the level reaches EL68.5 m AHD in Wivenhoe Dam, switch to Strategy W2 or W3 as appropriate”.

76. [229]

    Further, bearing in mind the Manual’s purpose, and the textual inconsistencies and infelicities, it is convenient to step back from the detail and appreciate that the Manual contained “operational procedures” for the operation of two dams during flood events, noting that all that a flood engineer could do was to make decisions when and at what rate to release water. The Manual gave a structure to the exercise of the only power the flood engineers had.

77. [230]

    The flood mitigation purpose of the dams was to reduce the flow rates into downstream areas. It was the rate, rather than the absolute volume, which would cause damage (given enough time, an indeterminably large volume of water will proceed downstream; a flood (by definition) is a large amount of water in a relatively small timeframe). However, the role of the dam as infrastructure for the purpose of flood mitigation also turned on volume; it permitted flood waters temporarily to be stored and later to be released, thereby reducing the maximum flow rate. Concepts within the Manual, such as “peak outflow” and “peak inflow” reflect the significance of flow rates.

78. [231]

    Plainly it would be catastrophic were the dam to fail; hence W4. But, subject to preserving the integrity of the dam, the most significant consideration, which was reflected throughout the Manual, was the desirability of reducing the inundation of urbanised areas. That was because a larger number of people would be affected and greater damage would be suffered than by the inundation of less densely inhabited rural areas and rural infrastructure such as the low level downstream bridges. Indeed, the critical distinction between strategies W2 and W3 on the one hand and W4 on the other turned on the maximum release of 4000m3/s at Moggill. As explained in bold under strategy W3:

79. [232]

    If the predicted maximum dam level exceeded 68.5m but not 74m, then there was sufficient risk that further inflows would result in urban inundation that the key objective became taking steps to prevent urban inundation. If the predicted maximum dam level were to exceed 74m, there would be sufficient risk of further inflows which may cause structural collapse to conclude that the key objective would be to preserve the dam. That meant taking steps which would probably cause hundreds of millions of dollars of damage to Ipswich and Brisbane, so as to minimise the catastrophic risk of a structural collapse of the dam (which would cause even greater loss).

80. [233]

    Essentially, the procedure described in the Manual involved the engineers predicting as best they could what the likely maximum amount of water was going to be. Then, appreciating that inflows might exceed that prediction, they were to regulate outflows by reference to the primary consideration as identified by the Manual. That is to say, the Manual answered the basic question a flood engineer must ask during a flood event:

81. [234]

    The importance of degrees of tolerance may be seen in a number of ways.

    1. (1)

       To commence with an extreme example, it is plain that water levels could not be measured to the nearest millimetre (the levels were derived by observation) from a gauge at a distance which was not measured in centimetres, but in decimetres, and required a deal of interpolation; on occasion the water level at the gauge may be affected by wind or waves.

    2. (2)

       More importantly, the Manual fairly consistently identified different strategies, and different sub-strategies and other decisions, by reference to increments of dam levels of 25cm. No decision or criterion in the Manual turns on a dam level at any finer gradation (such as +/- 5cm or +/- 10cm).

    3. (3)

       Further, the fact that the Manual is relatively indifferent to dam heights within a 25cm range is reflected in the provision that, notwithstanding that a flood event commences when a flood engineer is of the opinion that the dam level will exceed FSL, gates are not to be opened unless the dam level exceeds 67.25m. Yet, if the dam is at FSL, there is no expectation that an engineer would declare a flood event every time it rains and the water level goes to 67.05m; indeed, there could be little point in declaring a flood event unless the level was at, or expected shortly to reach, 67.25m before which the Manual prohibited opening a gate.

    4. (4)

       A striking indication of the relative lack of precision in water levels for the purposes of making decisions or determining criteria for the Manual is the fact that not only the FSL, but two critical levels, 68.5m and 74m, are expressed in whole or half numbers of metres. (That is to say, Strategy W4 obtained at a threshold of 74.0m, not 73.75m or 74.25m, and Strategy W3 applied at a threshold of 68.5m, not 68.25 or 68.75, let alone smaller variations.)

82. [235]

    The issue of tolerances will be directly relevant in considering the breach found to have occurred on 2 January 2011, when the flood engineers terminated the December flood event while Wivenhoe was at 67.1m, 10cm above FSL.

83. [236]

    The most important issues in the appeal concerning the Manual resolve to:

    1. (1)

       how was the predicted maximum water level to be determined in a flood event (in particular, should regard have been had to future releases, and to forecast rain and, if so, qualitatively or quantitatively)? This issue incorporates a temporal horizon and varying degrees of uncertainty.

    2. (2)

       when (if at all) may flood releases reduce the water level to below AHD 67m?

    3. (3)

       when does a flood event end?

11. The “no release” assumption – (ground 8)

1. [237]

   The primary judge addressed the “no release” assumption extensively, over some ten pages of reasoning: Rodriguez (22), Ch 3 [201]-[237]. The issue is whether the references to “maximum storage levels” in sections 8.3 and 8.4, the determination of “likely levels” in the flow chart and the “predicted” levels in the “conditions” of each strategy were determined by assuming no releases from the dam were made.

2. [238]

   The adoption of the no release assumption could make a major difference to the identification of the appropriate strategy for managing flood operations. Essentially this was because the floodgates were capable of releasing a very large amount of water in a relatively short period of time. In order to appreciate the force of this, it is essential to be able to relate flow rates, typically measured in m3/s, to volumes of water, which are usually measured in megalitres (Ml).

3. [239]

   A cubic metre of water is 1000 litres. Hence 1000 cubic metres of water is 1Ml, and 1000m3/s is a flow rate involving one megalitre every second, which is the same as saying 3600 megalitres every hour. A convenient rule to convert m3/s into megalitres per hour, is to multiply by 3.6.

4. [240]

   Suppose during a flood event the gates are opened so that 3,000m3/s is being released (this might be consistent with strategy W3 to keep the flow at Moggill less than 4,000m3/s if the flows from Lockyer and Bremer are below 1000m3/s). 3,000m3/s is 3Ml/s = 10,800Ml/h. Over a 24hr period, that equates to 24 x 10,800Ml, or 259,200Ml, which is considerably more than 10% of the absolute flood storage capacity of the dam, and a little over 22% of the flood storage capacity of the dam at 75.5m. If the dam were at 75.5m, 259,200Ml represents a depth of slightly more than 1.5m, and its release would lower water levels by about 1.5m to slightly less than 74m. (Appendix C states that flood capacity at 75.5m is 1,160,000Ml and flood capacity at 74m is 910,000Ml; the difference of 250,000Ml is only slightly less than 259,200Ml.) This calculation puts inflows to one side and in this respect is artificial, because if 3000m3/s were being released, it may be expected that there would be substantial inflows. However, the point of the example is to indicate how quickly significant flood storage capacity can be made available, even at levels of release which do not by themselves involve inundation of urban areas.

5. [241]

   The “no release” assumption is also, in a sense, counterintuitive. It applies during a flood event. Yet that is precisely the time when one would expect the radial gates to be open, releasing hundreds or thousands of cubic metres of water every second.

6. [242]

   Another way of appreciating the significance of the no release assumption is that if (as occurred throughout most of 8 and 9 January) water were being released at around 1300m3/s, then applying the no release assumption over a 24hr period equates to inflating the amount of water in the dam by 24 x 3.6 x 1,300 = 112,320Ml. If the water level was 68.5m, then that equates to almost an additional metre of water in the dam. (Appendix C states that flood capacity at 68.5m is 171,000Ml and flood capacity at 69.5m is 290,000Ml; the difference of 119,000Ml is only slightly more than 112,320Ml, and thus the latter corresponds to an increase from 68.5m to almost 69.5m.)

7. [243]

   Finally, if the no release assumption is applied for periods longer than 24 hours, then its impact is correspondingly greater. The primary judge found that the flood engineers had to apply the 4-day PME rainfall forecasts, which involved inflows of water into the dam over 4 (or more accurately 5 or 6) days, and so the effect of the assumption is to quadruple (or more than quadruple) the volume of water which is treated as if it had not been released.

8. [244]

   None of the foregoing is to deny the difficulties created by sudden, extreme rainfall events. The peak inflows into Wivenhoe in January 2011 were in excess of 10,000m3/s on two separate days, and there was a 13 hour period throughout which inflows exceeded 5,000m3/s. If rain is causing inflows of, say 8,000m3/s, and it is desired to keep outflows to, say, 3,000m3/s to avoid downstream urban inundation, then it is necessary to have the capacity to store, temporarily, 5000m3/s of water – which is 18,000Ml of water every hour.

9. [245]

   The parties, rightly, regarded the “no release” assumption as a critical element of the reasoning of the primary judge. They devoted substantial written submissions to it and senior counsel for Rodriguez addressed it at length on the first day of his oral submissions, returning to it more briefly on the following day. [107]

10. [246]

    The primary judge first addressed section 8.3, and noted that none of the defendants addressed this aspect of the Manual in their submissions. His Honour said that it was “self-evident” that the initial assessment of predicted flood flows should be undertaken without regard to releases. That accorded with what Mr Ayre said was commonly done, and Mr Malone’s actual first RTFM on 6 January 2011. His Honour relied on Mr Pokarier’s evidence to the contrary as an aspect which “significantly undermined my preparedness to accept the various opinions he expressed on the Manual”: at Ch 3 [203].

11. [247]

    Turning to section 8.4, the primary judge recorded four submissions made by the plaintiff: (i) consistency of approach between sections 8.3 and 8.4; (ii) consistency with determining the “maximum” storage level, on a “conservative” basis; (iii) maximising the flood mitigation potential of the dam, by telling the flood engineers whether the dam is capable of storing the whole flood without releases, and (iv) better achieving the objective of W3 and W4, noting that those strategies did not dictate any minimum rates of release.

12. [248]

    The primary judge identified three reasons advanced by the defendants: (i) the absence of any express reference to the no release assumption coupled with the expressed exclusion of Wivenhoe Dam releases in two bullet points at the beginning of section 8.4; (ii) the reference in the flowchart to the “likely” level of the dam, which “must include the proposed releases by the adoption of an iterative modelling process”, without which the “likely” lake levels would be distorted, and (iii) “the almost uniform chorus of evidence, other than Dr Christensen, to the effect that ‘no release’ modelling is unheard of in the field of flood operations whereas iterative modelling is commonplace”.

13. [249]

    The primary judge thereafter summarised the evidence of the witnesses: Ch 3 [210]-[220]. The summary included his Honour’s criticisms of aspects of their evidence, which need not be addressed here. His Honour’s reasoning on this point is elaborate: Ch 3 [221]-[237]. His Honour regarded a water control manual which used predicted water levels to select strategies as “relatively unique”, and reiterated that at least some modelling based on the no release assumption was in fact undertaken and was presumably regarded as having utility.

14. [250]

    Concerning the references to “predicted maximum storage level” and “likely” storage level, his Honour said at [223]:

15. [251]

    However, the principal reason advanced by the primary judge was the problem of “circularity” if the no release assumption were not adopted, stating:

16. [252]

    The primary judge considered that a further problem with modelling maximum storage levels based on releases was that it introduced much more uncertainty, because, over the future covered by the modelling, the gate operations would change and could not be reliably predicted. His Honour said that the flash flooding which occurred in the Lockyer Valley on 10 January 2011 supported the no release assumption, because if dam operations had been determined using that assumption, then a flood engineer “would be in a better position to act to cease releases if circumstances require it”.

17. [253]

    The judge rejected the defendants’ submission based on the references to “likely” in the flow chart:

18. [254]

    His Honour gave specific attention to the box in the flowchart which referred to maximum flows at Lowood and Moggill, stating at [229] that “[i]t can be accepted that this requires that consideration be given to outflow levels from Wivenhoe Dam”. However, he continued:

19. [255]

    The primary judge concluded this section of his reasoning by noting five further matters.

    1. (1)

       First, his Honour recorded “evidence” adduced by the defendants to the effect that adopting a no release assumption would lead to W4 being adopted whenever inflows were expected to exceed 910,000Ml. His Honour said that many flood events had inflows of less than 1,000,000Ml, and that this did not suggest that “the adoption of no release volumes would repeatedly and unnecessarily place flood operations into W4”: at [232].

    2. (2)

       Secondly, his Honour said that “just because the Manual requires an assessment to be undertaken on a no release basis does not mean that the flood operations engineers were precluded from modelling gate operations to determine the release rate and the likely height of the dams that would result if certain release rates were adopted”: at [233]. This led to the following analysis, based on the fact that all of the higher order objectives required consideration to be given to lower order objectives, his Honour saying:

    3. (3)

       The third point was that the consequence of a no release assumption was that the predicted maximum depended on the chosen forecast period. His Honour said that while this introduced an element of subjectivity into the determination of strategy it did not involve the circular reasoning that considering outflows must, and was a matter for engineering judgment: at [234].

    4. (4)

       The fourth point was a minor criticism based on one of Dr Christensen’s simulations: at [235].

    5. (5)

       The fifth point reiterated an earlier point in response to a criticism that the no release assumption was unrealistic and incorrect, inter alia, in circumstances where releases were currently being made. His Honour said:

20. [256]

    The following exchange in the parties' written submissions encapsulated the essence of the debate. Seqwater submitted: [114]

21. [257]

    Rodriguez' response was as follows:

22. [258]

    In oral address on this ground of appeal, Mr Sexton candidly addressed the most obvious difficulty confronting acceptance of this aspect of the primary judge’s reasoning: [115]

23. [259]

    He concluded:

24. [260]

    Although styled as a question of construction, the “no release assumption” is better seen as answering the question “what is involved in the fundamental integer in the operational procedures”, namely, predicting the extent of the flood event from the perspective of the flood storage capacity of Wivenhoe Dam. Obviously, it is necessary to predict the volume of water flowing into the Dam. But does one disregard outflows (and thus simply assess the maximum volume of water likely to flow into the Dam) or does one take into account outflows so as to gain an understanding of the maximum volume of water in the dam throughout the flood event?

25. [261]

    In a sense, the inquiry is rather artificial. The Manual proceeds on the basis that there will be an initial assessment (section 8.3) followed by an iterative reassessment of strategy continuously throughout the flood event (section 8.4). The repeated application of section 8.4 reflects the fact that conditions are apt to change throughout the flood event. Predicted rain may or may not fall, the weather forecast may change, and downstream flows may rise or fall depending on conditions in Lockyer Creek and Bremer River. For those reasons alone, it was necessary constantly to review the decisions which have been made to release water in light of conditions which may have changed subsequently.

26. [262]

    However, the consequence of the procedure being iterative is that at all times, save at the commencement of the flood event, the procedure will unavoidably be derived in part from outflows that have occurred during the flood event to date. That is saying no more than that the height of the water in the Dam at say 16:00 on day 1 of a flood event is a consequence in part of the decision made at 15:00 to open the flood gates so as to release water at a certain rate. The repeated operation of section 8.4 must take as one of its inputs the result achieved by previous decisions made following the procedure in section 8.4. That is saying no more and no less than that the procedure is iterative.

27. [263]

    The correctness of the “no release assumption” therefore resolves to this question: During each iteration of the procedure in section 8.4, does one disregard all continuing and future outflows, although necessarily making a decision based on the outflows which have already occurred during the flood event? The Manual does not explicitly answer this question, and arguably better guidance as to how it would be construed by a flood engineer in the position of Messrs Ayre, Malone, Tibaldi and Ruffini may be gained from its purpose rather than the text of the Manual. However, both text and purpose point against the no release assumption, and it is convenient to commence with the text.

28. [264]

    The Manual’s text points against the no release assumption. There is nothing expressly requiring future outflows to be ignored when determining the predicted maximum height and storage level. It is natural to approach the Manual on the basis that the predictions are real, rather than counterfactual. Dam levels turn on inflows and outflows (other processes such as evaporation were ignored). Why would one look into the relatively certain future of rain on the ground and the less certain future of predicted rainfalls, but ignore something squarely within the flood engineers’ control, namely, outflows? The natural meaning of the Manual when it asks, repeatedly throughout a flood event, about the maximum level at Wivenhoe, and whether the “Wivenhoe Storage Level [is] predicted to be” less than 68.5m, or between 68.5m and 74m, or to exceed 74m, is that it is asking a question of the predicted actual level of water in the dam. This will take into account such releases as have occurred, and those which are in fact being made at that time. There is no relevant certainty to be achieved by assuming that on-going releases cease.

29. [265]

    The procedures in section 8.4 of the Manual involving predicted water heights are to be invoked iteratively and repeatedly during a flood event. That is to say, at the very time when the no release assumption is to be applied, the flood gates are apt to be open and releasing hundreds or thousands of cubic metres of water every second. The no release assumption asks one to assume that at all times during the flood event, all gates are closed. That is a profoundly counterfactual assumption.

30. [266]

    Further, the Manual requires peak flows rates at Lowood and Moggill to be predicted “excluding Wivenhoe Dam releases”. Those words are unnecessary if the modelling is to occur on the no release assumption, but understandable if the general approach is to predict the future making allowance for releases.

31. [267]

    Still further, there is the decision-making described in the flow chart. As the primary judge acknowledged, the box which determines W2 or W3 necessarily involves including releases. It is a remarkable outcome that, on the construction upheld by the primary judge, the boxes on the same table which identify the two anterior choices to be made in determining strategy, are made on a different basis, namely, assuming that no releases are made. That is to say, in order to apply the flow chart under the no release assumption, the flood engineer must ask whether the Wivenhoe level is likely to exceed 68.5m and 74m on the assumption no water whatsoever is released, and if it will not exceed 68.5m or if it will exceed 74m, then strategies W1 or W4 apply. However, if the level is between 68.5m and 74m, then the flood engineer must now have regard to releases, and ask what the likely peak flows at Lowood and Moggill will be.

32. [268]

    This is not merely a remarkable approach given the absence of textual justification on the face of the Manual; it is one which would require the flood engineer to use different model runs in order to resolve different decision points within the one flowchart. The RTFM would have to be applied without releases in order to determine predicted dam level. If the result were somewhere between 68.5m and 74m, then the model would have to be run again, this time putting in place releases, so as to determine the maximum flows at Lowood and Moggill.

33. [269]

    This leads to contradictions. Suppose during a flood event the dam is close to FSL. Section 8.4 requires one to ask whether the Wivenhoe level is likely to exceed 68.5m. This will occur if 171,000Ml of water flows in, if the no release assumption is adopted. According to the primary judge, a flood engineer is required to consider inflows over (at least) a four day period. But over the same four day period, if water is being released at, say, 380m3/s, then slightly more than 131,000Ml (4 x 24 x 3600 x 0.380Ml/s) will be released, in which case the net inflow of less than 40,000Ml would cause the dam level to rise less than 50cm (Appendix C shows there is 56,000Ml of flood capacity between 67m and 67.5m). The significance of this is that once releases are taken into account – as they concededly must be to determine between W2 and W3 – one remains firmly in W1. That is to say, making the no release assumption not only counterfactually disregards the 380m3/s which is presently being released; it is the no release assumption which forces the flood engineer to ask which of W2 or W3 is applicable, in circumstances where the actual Dam level is expected not to exceed 68.5m, and having reached that point in the decision tree, in order to determine which of W2 or W3 applies, one is forced to disapply the no release assumption.

34. [270]

    The attempted response to the contradiction was that while a higher strategy may be invoked, it does not mean that higher release rates are mandated. The response is not persuasive. While strategy W3 or W4 might be engaged, there was, according to the primary judge and the first respondent, nonetheless ample discretion to have outflows of much less than 4000m3/s. But this dilutes the effectiveness of the Manual as a tool. A sensible construction of the Manual produces a result which clearly identifies what at any time ought to be the flood engineers’ top priority.

35. [271]

    In order to explain why that is so, it is helpful to consider the Manual’s purpose.

36. [272]

    The purpose of the Manual is, first and foremost, to have “clearly defined procedures” to guide flood engineers during a flood event. A flood engineer needs to know, at any given time, what ought to be the “primary consideration” on the basis of which decisions to release water are made. Is it to protect the structural integrity of the dam? Is it to protect Brisbane and Ipswich from inundation? Or is it merely to keep open some bridges?

37. [273]

    The answers to those questions are determined by aspects of the physical structure of the dam, and in particular, the empirical fact that if the actual water level exceeds 75.7m, the first fuse plug will erode, leading to the loss of all water down to 67m. It is of the utmost significance therefore to consider whether the actual water level is likely to exceed 75.7m because of a physical event which will take place in the real world. It is of very little significance if over a 4 or 8 day period, the hypothetical level of water in the dam will exceed 75.7m if one pretended that the gates were closed, when in fact they are releasing hundreds or thousands of cubic metres of water each second.

38. [274]

    The Manual mandates that the primary consideration shifts from preventing urban inundation to protecting the structural safety of the dam when the dam is predicted to exceed 74m, because 74m is close to 75.7m. If the water level in fact (ie allowing for releases) is predicted to exceed 74m, then there is relatively little flood storage capacity before the first fuse plug erodes; in those circumstances it is natural to focus efforts on protecting the dam. It makes much less sense to ask the hypothetical question demanded by the no release assumption, namely, if we pretended the gates were closed, is the dam level predicted to exceed 74m over the next 4 days? If the answer is “yes”, then that does not say much about the real risk of the water level exceeding 74m or even 75.7m. One could not determine the real risk without regard to how much water was presently being released.

39. [275]

    On an approach which includes the no release assumption, the situation posited above is resolved by further steps, to the effect that: “Even though the no release assumption takes me into W4, I can see in truth that there is really very little risk to the dam structure, so I can continue with smaller releases of less than 3500m3/s so as to avoid inundating Brisbane and Ipswich, in accordance with the precept that regard is had to lower order objectives”. That is so. But it amounts to a material detraction from the Manual’s purpose.

40. [276]

    The primary judge proceeded on the basis that the rules dictated by the Manual were to be evaluated without regard to future outflows which were contemplated or indeed in some cases mandated by the Manual itself. That had the consequence that decisions were to be taken on a “worst case scenario”. It is inevitable that if dam outflows are ignored, then predicted dam levels will be greater. Further, if dam outflows are ignored, then the further ahead in time one predicts, the greater the dam level will be. If an 8-day PME were used and outflows ignored, the maximum level in the dam will inevitably occur on day 9 or 10 (after the rain predicted to fall in the catchment on all days including day 8 has found its way into the dam). That might be an available construction, but it is one which is contrary to the fundamental purpose. If one thing is clear, water should not be released from Wivenhoe at a rate greater than 4000m3/s unless any other course were unreasonably risky. But the construction favoured by the primary judge would lead to W4 being engaged if, say, over a five-day period dam levels were predicted to exceed 74.00m AHD without regard to outflows but over the same period dam levels would only rise to 73.5m AHD if outflows were taken into account.

41. [277]

    Take for example a flood event when the water level is at 68m, and the flood gates are releasing 500m3/s. Assuming that the predicted inflows over the next four days are of 500m3/s, followed by dry conditions, that is an inflow of some 172,000 megalitres over the four days. But while those inflows match the existing 500m3/s outflows, the water level in the dam will not move a centimetre. A sensible operation of the Manual would reach the conclusion that at all times strategy W1 should be applied, with attention being given to preserving the downstream bridges. Yet if the no release assumption is applied, the strategy is W2, and the flood engineers are required to have as their primary consideration the prevention of urban inundation. Why should that be so, in circumstances where there is no suggestion of any risk of urban inundation? It is no answer to that conclusion to point to the fact that even within W2, regard must be had to lower order objectives. An assumption which causes one to be placed into a strategy which has as its primary consideration the prevention or urban inundation when there is no risk of urban inundation is, it may be inferred, a false assumption.

42. [278]

    In short, the no release assumption inevitably pushes the flood engineers towards higher strategies. They are pushed into W2 or W3 whenever 910,000Ml is predicted to fall in the catchment over the period of time being considered. But on the approach adopted by the primary judge and defended by Rodriguez it is said that nonetheless, low levels of water release may be made if it is clear that there is no risk of inundating Brisbane or Ipswich. That approach substantially adds to the contestability of the processes incorporated in the Manual. It also undermines the heavy insistence in the Manual on the “primary consideration” which comes into play when any particular strategy is engaged.

43. [279]

    This gives rise to an element of circularity. The primary judge regarded this as inimical to an approach which took into account releases. But why is that so? The debate at trial concerning circularity appears to have proceeded on the assumption that a construction involving recursion or self-reference or “circularity” was, for that reason, not merely a factor tending against its acceptance, but fatal: the reverse is true. Such reasoning seems to put to one side the very iterative aspect of the process delineated in the Manual.

44. [280]

    It was common ground that the “Flood Operations Strategies” in section 8.4 were to be operated iteratively. That is to say, the decisions to be made during a flood event would vary depending upon the circumstances which might well change throughout that event (including rising dam levels, inflow levels and downstream flows). It is of the essence of an iterative procedure that it is circular or self-referential. The inputs of subsequent applications of the procedure include the output of an earlier application. Thus the operation of section 8.4 at 16:00 turns in part upon the effect of the decisions made following the operation of section 8.4 at 15:00.

45. [281]

    As the primary judge emphasised, this gave rise to questions of discretion and judgment. By proceeding on the basis that the future releases were relatively high, a flood engineer might remain in a lower strategy. The primary judge relied on this consideration as a factor in favour of the no release assumption. We disagree. First, if releases of, say, 1500m3/s enable strategy W1 to be maintained, why is that a bad thing? Secondly, there are important questions of judgment involved at all stages, including the length of the forecast period selected. Thirdly, if the no release assumption led to a higher level strategy being selected, but when releases of 1500m3/s were applied, it appeared that the primary considerations of avoiding urban inundation or protecting the dam structure could be met, then on the approach adopted by the primary judge, it would be open to the flood engineers to keep releases at W1. That simply shifts the element of discretion into the decision as to how wide to open the flood gates, and diminishes the effect of being in a particular strategy.

46. [282]

    Rodriguez placed weight on the fact that section 8.3 did not involve taking releases into account. Whilst the premise is correct, limited weight should be given to section 8.3, which is only ever applied once, at the commencement of a flood event when the floodgates are closed, in construing section 8.4, which is to be applied repeatedly throughout the duration of the flood event.

47. [283]

    It is also far from clear how much weight should be given to earlier drafts of the Manual. However, it will be seen that the approach adopted by the primary judge resembled an earlier version of the flow chart (in fact, the one erroneously reproduced in Chapter 3) which explicitly left lower level strategies in play even if W3 or W4 were engaged. But that form of the flowchart was rejected in the final version of the Manual.

48. [284]

    The extent to which regard may be had to the operation of flood mitigation systems in other dams for the purpose of understanding the procedures in the Manual is also unclear. But this does inform a related issue, namely, whether the flood engineers acted reasonably in applying the Manual on the basis which did not involve ignoring future releases. The evidence was clear that none of the witnesses, including Dr Christensen, had ever encountered a flood mitigation system which required engineers to disregard the effect of releases of water.

49. [285]

    Ground 8 is made out. The operation of section 8.4 of the Manual during a flood event did not require the engineers to assume that no water would be released from the Dam. Moreover, it was not unreasonable for the flood engineers to apply section 8.4 of the Manual having regard to the water which was in fact being released from the dam.

50. [286]

    This conclusion has an important consequence: Dr Christensen’s identification of strategy at various points in time turned on an application of the “no release” assumption. This assumption underlay all his simulations; while alternatives were considered in relation to other assumptions, no simulation modelled the iterative process. In particular, it underlay simulations C and F accepted by the primary judge as modelling reasonable strategies. Consequently, it underlay the findings that the engineers failed to follow the Manual and maintained strategies which were not engaged.

12. “Best available forecasts” – (grounds 6, 7)

1. [287]

   Grounds 6 and 7 in Seqwater’s appeal were expressed to be in the alternative. The drafting is awkward because there was a chapeau to all of grounds 6 to 15 which also contained alternative approaches and applied to each individual ground. The manner in which the Manual should be approached has been discussed in part 10 above. There is no need to say more about the formulation of the chapeau, beyond noting that how a flood engineer should reasonably have construed the Manual is antecedent to any question of breach. It poses a different question from how the flood engineers in fact construed the Manual, or if they construed it at all. The pleaded grounds of negligence alleged that the actions of the flood engineers were in breach of their duty of care to the plaintiff and members of the represented class. These issues will be addressed below: for present purposes, it is sufficient to note that some of the aspects of the operation of the Manual which are covered by grounds 6-15 need not be resolved unless they are directly engaged in the analysis of breach. Unlike the “no release” assumption, that is not self-evidently so with respect to grounds 6 and 7.

2. [288]

   Bearing those matters in mind, it is convenient to set out the terms in which the alleged errors are expressed in grounds 6 and 7:

3. [289]

   The reference to “them” in these grounds is a reference to the four flood engineers involved in the operations in January 2011. The reference to “RTFM modelling” in ground 6 may be identified by reference to section 5.1 of the Manual:

4. [290]

   Sections 8.4 and 9.3 referred to flood operations strategies for Wivenhoe and Somerset Dams respectively. Section 8.4 established some general propositions before setting out the relevant strategies for the operation of Wivenhoe. The relevant text is set out at [202] above. Similar language was used at section 9.3 in relation to Somerset.

5. [291]

   There is no doubt that the phrase “best … stream flow information” identified information as to rain which had fallen in the catchment and the system for measuring rainfall and stream flow referred to in the first paragraph of section 5.1 set out above. It is also clear that the phrase “best forecast rainfall” referred to rainfall forecasts as provided by the Bureau of Meteorology from time to time. Predictions as to dam levels and flow rates were to be made on the basis of that information. Further, section 5.1 referred to real time flood monitoring and modelling system used to estimate dam inflows and evaluate a range of possible inflow scenarios which, in each case, were to be “based on forecast and potential rainfall in the dam catchments.”

6. [292]

   In the two passages in the judgment referred to in ground 6, the judge concluded that the relevant modelling was required to take into account predicted rainfall based on forecasts, and was not limited to measurements derived from “rain on the ground”. That conclusion was correct.

7. [293]

   On the appeal, Seqwater contended that a distinction was to be drawn between (i) using best forecast rainfall in modelling exercises which were to be the basis of decisions as to water releases, “in a quantitative sense” and (ii) using forecasts of rainfall “only in a qualitative way” and “in a general sense”, leaving room for judgment by the flood engineers as to the quantitative determination of releases at particular times. Seqwater contended for the latter approach. However, as will be explained more fully below, while a computer program may produce a precise numerical outcome, the use to which that figure can be put will depend upon elements of uncertainty in the information used. At no stage did the primary judge deny that there was scope for judgment to be exercised on the part of the engineers. Indeed, as Seqwater acknowledged in its written submissions, the judge expressly stated in Ch 3:

8. [294]

   These conclusions were essentially beyond challenge. Indeed, in addressing ground 6, Seqwater focused upon two entirely separate questions, namely (i) whether the flood engineers in fact acted in conformity with the Manual and (ii) whether the judge was ultimately right in concluding that, assuming the relevant circumstances and strategies, any reasonably competent flood engineer would have made releases in accordance with one of Dr Christensen’s simulations C, F or H. These questions were not dictated by the findings challenged in ground 6, which must be rejected.

9. [295]

   Much attention was devoted in the course of the trial to the meaning of the phrase “best forecast rainfall” used in sections 8.4 and 9.3 of the Manual. However, arguably the only significant word in this phrase was the word “forecast”, which could not refer to rain which had already fallen. Beyond the conclusion that the flood engineers were required to take into account forecasts of rain, and not merely predictions of dam levels and inflows based on rain on the ground, there was little more to be derived from semantic analysis. The primary judge correctly found that the engineers were required to determine strategies and releases having regard to forecast rainfall; their real time modelling programs permitted this.

10. [296]

    Forecasting rainfall will involve three main parameters, namely timing, location and volume. What forecast is “best” will depend upon the purpose for which the forecast is used. Forecasts of the immediate future (24 hours) were more likely to be accurate in each respect than longer term forecasts. Broadly speaking, a 4-day forecast was more likely to be accurate than an 8-day forecast. However, the purpose of flood mitigation is to store the inflow from heavy rainfall and release it over a longer period of time, thus reducing peak flows in the river system. When dam storage is limited, there will be benefit in relying upon forecasts extending beyond 24 hours. On the other hand, there will be a time lag between the rain falling and the dam level rising, except for rain falling directly into the dam. A 24-hour rainfall forecast may give the flood engineer up to 48 hours within which to deal with the inflows. (Rain falling late in the period could take up to 15 hours to reach the dam.)

11. [297]

    The time chart for flows in the Brisbane River, agreed as a “rough guide”, [117] showed flows from the upper reaches of the catchment into Somerset or Wivenhoe as ranging from 12.5-15 hours; from Wivenhoe to Moggill gauge as 16 hours and from Moggill gauge to City gauge in Brisbane as 10 hours.

12. [298]

    Of critical importance for the flood engineers was the location at which the rain was expected to fall. As noted above, approximately half the Brisbane River catchment was below Wivenhoe. Rain falling below the dam would not provide flows into the dam, but it would increase flows into the Brisbane River above Moggill and thus limit the volume which could be released from Wivenhoe without causing inundation in Brisbane. Again, there were delays between cause and effect: a release from Wivenhoe would take 16 hours to reach Moggill, and similar delays would occur for rain falling in the Lockyer and Bremer catchments. The longer the term of the forecast, the greater the level of uncertainty with respect to how much rain would fall, precisely where it would fall and when it would fall. Degrees of uncertainty were reflected in the forecasts themselves using ranges for the amount of rainfall; the forecasts in 2011 did not provide degrees of probability.

13. [299]

    The Bureau of Meteorology provided forecasts in two main forms. The quantitative precipitation forecasts (QPFs) were rainfall predictions for a particular location or region covering a 24 hour period: Ch 2 [135]-[136]. QPFs were issued twice daily for 24 hours to 09:00 and 15:00 on the following day. They were specific to the Somerset and Wivenhoe catchments and were issued approximately one hour into the 24-hour period. They were provided by means of an email to the Flood Operations Centre and forecast the “catchment average rainfall” for the 24-hour period, in the form “30-50mm”. The Bureau also produced “Probability Matched Ensembles” (PMEs) based on an amalgamation of forecasts by national meteorological services from different parts of the world. The PME forecasts were made available on a daily basis and for a forecast period commencing at 22:00. They were issued in three tranches, the first for the period commencing at 22:00 were issued at 18:00, the second at 00:00 and the third at 06:00 on the following morning. The PMEs were identified by their “base time”, being the time at which the data used in the forecast became available. A base time represented as 00UTC (that is 00:00 Universal Coordinated Time, formerly Greenwich Mean Time) was 10:00 Australian Eastern Standard Time; the 1200UTC equated to 22:00 AEST. The primary judge further explained in Ch 2:

14. [300]

    Professor Michael Manton, an eminent Australian meteorologist, noted that the service had been established by the Bureau in August 2009 and upgraded in November 2010. The result, published on the Bureau of Meteorology website, was in the form of a colour chart for the whole of Australia. The programming provided for estimates of rainfall to be made at grid points 50km apart. Each block of colour therefore represented an area of 2,500km2. The colour blocks themselves involved varying levels of precision along a scale 0-1mm, 1-5mm, … 25-50mm, 50-100mm … 200-300mm, 300-400mm and over 400mm. The PMEs were made available for 1-day, 4-day and 8-day periods. The QPF was based on the first day of the PME forecast.

15. [301]

    The most complete analysis of rainfall records and their reliability was a report by Professor Manton, who also gave oral evidence and whose evidence was generally accepted by the primary judge. Professor Manton considered that forecasts were too unreliable to have a significant role to play in the identification of strategies and water releases. He did, however, accept that the 1-day QPFs and the 4-day PMEs constituted available rainfall forecasts which should have been taken into account. A summary of the information available from these sources was provided by the judge in Ch 6 in the following table:

16. [302]

    Ground 7 challenged the conclusion that the flood engineers were required to use the 4-day PME forecasts, rather than the QPFs. The ground was ineptly drawn: the judge did not suggest that the QPFs were excluded. Rather, he concluded that by themselves they did not have an adequate time horizon to permit proper flood mitigation operations.

17. [303]

    The passage in Ch 9 of the judgment which most clearly set out the methodology accepted by the primary judge, with which ground 7 took issue, was as follows (cross-references omitted):

18. [304]

    Although the written submissions for Seqwater took issue with the evidential basis of the judge’s approach, there was ample evidence to support the findings set out above. The challenge focused on Dr Christensen’s use of the 4-day (and indeed the 8-day) forecasts in formulating simulations C and F, albeit differently in relation to each simulation (as to which see [421] and [422] below). The primary judge having accepted those simulations as reasonable, it was seen to be necessary, first, to attack each aspect of the methodology adopted by Dr Christensen and, secondly, to deny that such forecasts could lead to “quantitative” conclusions as to the appropriate strategy or the level of releases, because it was not demonstrated that the engineers had in fact made decisions on that basis.

19. [305]

    These challenges were either misconceived or unhelpful. For reasons separately set out, the use by Dr Christensen of a “no release” assumption in devising appropriate strategies and his approach in releasing water below the FSLs for each dam were not approaches required of a reasonably competent flood operations engineer in the circumstances. In other respects, Dr Christensen’s modelling was transparent and reasonable.

20. [306]

    As to the second matter, there was evidence that the flood engineers did in fact have regard to the 4-day forecasts prepared by the Bureau. Whether they had sufficient regard to them was another question. However, as Rodriguez correctly submitted, that question could not be answered by positing a qualitative/quantitative dichotomy on the possible uses of rainfall forecasts. Rather, the proper course was to determine whether on the information available, and having regard to levels of uncertainty inherent in the forecasting, the steps taken by the flood engineers were reasonable in all the circumstances. As will be discussed shortly, the critical question was whether the engineers gave sufficient and appropriate weight to the 4-day forecasts provided on 7, 8 and 9 January 2011.

21. [307]

    Ground 7 should be rejected.

22. [308]

    Because the information derived from the rainfall forecasts informed much of the reasoning as to the breaches of duty, it is convenient to set out the basic information here.

23. [309]

    Based on the principles set by the Manual, and acceptance that best available forecasts included 4-day PMEs, it is convenient (albeit at the cost of some repetition) to set out what the engineers knew or ought to have known from 5 January 2011.

24. [310]

    Although insignificant rain fell on 2 and 3 January, from 18:00 on 2 January the 4-day PME (applicable from 22:00 hours onward) forecast 50-100mm above Wivenhoe and a similar (25-100mm) below the dam. Accordingly, that prediction related to 3-6 January. However, the 1-day PME indicated insignificant rain above the dam, from which it could be inferred that the heavier falls were due over 4, 5 and 6 January. By 18:00 on 3 January, no rain had fallen, but the 4-day PME was raised to 75-150mm. The 8-day PME had the same maximum but a range from 90-150mm. From that it could be inferred that the bulk of the rain was expected in the period of 4-7 January. Although the 1-day PME applicable to 4 January had predicted 5-15mm, in fact no significant rain fell on that day. The PME issued at 18:00 on 4 January had a similar estimate of 5-15mm for the following day, and maintained the 4-day PME at 50-110mm. At 16:00 on 5 January the 1-day PME was still 5-15mm and the 4-day PME above the dam was 50-125mm. (The 8-day PME issued on 5 January had risen to 100-200mm.)

25. [311]

    Mr Malone prepared an analysis of observed rainfall for the period from 2 December 2010–20 January 2011 based on gauges in particular catchments and averaging the recorded rainfall across those catchments. There were three major catchments, the largest being Upper Brisbane (4,244km2), the area immediately surrounding Wivenhoe Dam, known as “Middle Brisbane” (1,429km2), and the catchment for Somerset (1,328km2). Mr Malone’s observed rainfall figures were set out in column 5 (“Rainfall received”) of Table 6-1 at [301] above.

26. [312]

    The primary judge set out in Table 9-2 the range of 4-day PME estimates for the period 1-11 January 2011:

27. [313]

    The first column represented the forecast for the 4-day period from 22:00 the evening before; for example, the first entry represented the period from 22:00 on 31 December 2010 to 22:00 on 4 January 2011. The ranges used by Dr Christensen and the State varied depending upon how one read the large scale contour maps with coloured patterns, discussed above. The range given in the right-hand column is the range of figures in fact adopted by the various witnesses. With one exception (5 January), Dr Christensen’s accepted value was the highest in every case. However, in Ch 9 the judge reached a conclusion in these terms:

28. [314]

    The next stage was to estimate inflows based on those figures. The table used by the judge (Ch 9 [235]) was as follows:

29. [315]

    Taking the inflow figures for 10 January of some 1,200,000Ml, which incorporated both rain that had fallen and rain forecast to fall over the next four days from 22:00 the previous night, and noting that at 22:00 on 9 January Wivenhoe was at 69.44m, it appears that the storage capacity at that time was approximately 1,450,000Ml. The addition of a further 1,200,000Ml would lift the total to 2,650,000Ml, and the dam level to about 77.5m, assuming that no releases were made in the meantime. That level would be well beyond the trigger for the third fuse plug. To prevent the level exceeding 74m it would be necessary to release 600,000Ml; to achieve that would require releasing water at a rate of 1,750m3/s for four days, starting immediately. The effect of such releases would need to have been assessed having regard to the expected peak flows at Lockyer Creek and the Bremer River.

13. Releases below Full Supply Level (FSL) – (grounds 12, 13)

1. [316]

   The primary judge concluded, at Ch 5 [168], that the Manual did not prohibit releases of water below FSL during flood operations, and that in certain circumstances, such releases should be made. Ground 12 of Seqwater’s appeal challenged that finding. Ground 13 challenged a further finding, at Ch 10 [178], that releases should have been made in the first week of January based on predicted rainfall, rather than rain which had already fallen in the catchment.

2. [317]

   These grounds attracted extensive written submissions. However, the significance of these grounds is quite limited. They turn on simulation C, which was advanced by Dr Christensen and which formed a basis of the analysis of breach and causation in relation to the first week. Under this simulation, Dr Christensen maintained that a flood engineer in the position of those operating Wivenhoe Dam in the first week of January 2011 would have left the gates open for the whole of that week, so as to release some 300 or 400m3/s until the evening of 5 January 2011, and thereafter at considerably greater rates. Relevantly for the purposes of these grounds, the effect was to reduce dam levels well below 67m, indeed to as low as 63.79m, by 13:00 on 9 January.

3. [318]

   None of this arises if there were no breach of duty in the flood engineers determining that the December flood event concluded on the morning of 2 January 2011, when the dam level was 67.1m. Rodriguez accepted that the Manual did not permit opening the gates in a new flood event until dam levels rose to 67.25m, in accordance with the command in section 8.3 (“The spillway gates are not to be opened for flood control purposes prior to the reservoir level exceeding EL 67.25”). Accordingly, the starting point and an essential premise of simulation C was that a reasonably competent flood engineer would not have brought the December Flood Event to an end. For the reasons given below in part 19(3), that premise is not made out.

4. [319]

   In any event, even if the flood event continued, we respectfully disagree with the conclusion reached by the primary judge that the flood engineers were obliged to make substantial releases. That conclusion was expressed at Ch 5 in the following terms:

5. [320]

   The Manual conferred a discretion when dam levels were at 67.1m. It was open to the flood engineers to form the view, on 2 January 2011, that it was not necessary to return to FSL by 3 January 2011, and to shut the gates, leaving only the release of around 50m3/s from the regulators.

6. [321]

   Further, if that be wrong, and if, in accordance with Dr Christensen’s simulation C releases of some 470m3/s continued throughout 2 January 2011, according to Dr Christensen, the water level would reach 67.05m at 16:00 on 2 January 2011 and 67.0m at 02:00 on 3 January 2011. The QPF forecast available at 10:00 on 2 January 2011 was for less than 5mm of rain in the next 24 hours and that at 16:00 for 5-10mm. The 4-day PME forecast showed less than 10mm, with most of that predicted to fall on the fourth day. It is true that the 8-day PME forecast predicted heavier rain towards the conclusion of that extended period.

7. [322]

   A flood engineer was not required in those circumstances to continue the flood event, and to release water so that the level fell below 67m. Even on a literal approach which has no regard to the degrees of tolerance inherent in the measurements of dam levels in the Manual, [118] on the afternoon of 2 January 2011 it was open to an engineer to determine that the December flood event had ended sometime after 16:00, when the next day’s QPF forecast was available, and with dam levels at around 67.05m. This would accord with the statements in the Manual that, “[t]here is no reason why the dams should not be full following a Flood Event”.

8. [323]

   For those reasons, grounds 12 and 13 do not arise. However, if we are wrong about the absence of breach in determining that the December flood event ended on the morning of 2 January 2011, these grounds are made out. They preclude a finding of negligence (even on the ordinary standard) on the part of the engineers in failing to continue to reduce the level of Wivenhoe below FSL.

9. [324]

   This conclusion would remove the basis for the plaintiff’s reliance on Dr Christensen’s simulation C; it leaves available reliance on simulation F.

14. Strategy W4 – (ground 9)

1. [325]

   The error identified in ground 9 was that “Strategy W4 was required to be implemented by a predicted storage level above 74.0m AHD but as not requiring an increase in releases until the lake level at Wivenhoe Dam actually reached 74.0m AHD”.

2. [326]

   The practical significance of Strategy W4 has been addressed above in considering the “no release” assumption. If the level of the dam were still rising, despite releases occurring in accordance with Strategy W3 (so as to avoid inundation in Brisbane and Ipswich) and continuing heavy inflows were expected, then Strategy W4 was engaged because, once the actual level passed 74.0m, it was getting dangerously close to the point at which the first fuse plug would begin to erode.

3. [327]

   The challenged passage in Ch 3 of the primary judge’s reasoning was as follows:

4. [328]

   Seqwater submitted that, “[g]iven the risk to the safety of the dam, it would be entirely anomalous to suggest that the dam could be in W4 but no additional steps need be taken to decrease the level of the dam.” The point appears to have been that once the engineers were operating under W4 and not W3, they were free of the requirement of W3 that releases should be at a level below that which would result in 4,000m3/s at Moggill. (Above that level, significant inundation is expected in Brisbane and Ipswich.) However, W4 does not in terms require that a higher level of releases be made and the impact of “rapidly increasing discharge” should be considered. Given Seqwater’s insistence in other circumstances on the uncertainty of 4-day forecasting and the unpredictability of rainfall and therefore continuing inflows into the dam, it is curious that a reading of the Manual which conferred greater flexibility and discretion was resisted at this point. In any event, it is even less clear that this has any practical consequence. The sole purpose of the submission appeared to be that although, on the judge’s finding, Strategy W4 was engaged on 8-11 January, yet on each of those days simulation F “involved operations to increase the level of the dam ….” Therefore, it was submitted, simulation F did not represent action expected of a reasonably competent flood engineer operating consistently with the Manual.

5. [329]

   It follows from this submission that once Strategy W4 is engaged, releases must immediately be undertaken to prevent the level of the dam increasing. However, there is nothing in the language of the Manual which requires such a conclusion. The two substrategies under W4 distinguish between circumstances where the level is expected to exceed 75.5m (and initiate the first fuse plug), and where it is not. It is difficult to understand why a flood engineer faced with an actual dam level of 73.5m and rising, but with inflows likely to take the level to 74.5m but no further rain in sight, would be acting unreasonably in allowing the level to rise.

6. [330]

   The proposed reading of the Manual does not render the conduct in simulation F necessarily unreasonable. That question will be considered in its terms in due course. Ground 9 should be rejected.

15. “Peak inflow” – (grounds 10, 11)

1. [331]

   Under the general heading “Flood Operations Strategies”, the Manual provided, as discussed above, that a choice of strategy will depend upon “the best forecast rainfall and streamflow information available at the time”, and noted that strategies may change “in response to changing rainfall forecasts and streamflow conditions to maximise the flood mitigation benefits of the dams.” Critically for present purposes, the Manual then stated:

2. [332]

   The controversy surrounding the meaning of this last sentence turned on whether the Manual permitted a flood engineer to lower the dam level, and thus increase storage capacity, in advance of heavy rainfall. If “peak inflow” were restricted to the peak inflow experienced at a point in time prior to the heaviest predicted rainfall, it might be possible to slow the rising dam level, but, unless there had already been very heavy rainfall, a significant reduction in the dam levels would not be possible. At trial, the defendants contended that peak inflow should not depend upon predictions of rainfall which might not come to fruition. The general purpose of flood mitigation was to reduce the peak levels of flow in the river, not to exacerbate them. To rely on uncertain predictions by pre-releasing significant volumes of water might well exacerbate, rather than mitigate, the flood.

3. [333]

   The judge rejected that approach on two bases. First, the Manual, and indeed section 8.4, read as a whole, expressly required that the choice between strategies, with expected outflows, required that regard be had to rainfall forecasts. Accordingly, the risk of exacerbating flooding was to be offset by the potential benefits of increasing flood storage so as to reduce the peak flows resulting from predicted rainfall. Reliance on predicted rainfall, and therefore predicted peak inflow, could be accommodated by the language that peak outflow should “generally” not exceed peak inflow: Ch 3 [282], [284].

4. [334]

   The judge’s conclusion in Ch 3, challenged by grounds 10 and 11 of Seqwater’s appeal was as follows:

5. [335]

   The significance of the point may be illustrated by reference to charts showing the actual operations of Wivenhoe and the operation proposed by Dr Christensen under simulation F.

6. [336]

   As appears from the columns headed “Actual Operation”, the peak inflow (at 07:00) on January 8 was 2,144m3/s, and the highest level of outflow was 1,242m3/s. By contrast, the peak inflow under simulation F was 1,937m3/s, also reached briefly at 07:00; at that point the outflow started to exceed the inflow, and continued to exceed it for the rest of the day, the final reading providing an outflow some five times greater than the inflow. The result was a net outflow over 24 hours of 100,000Ml as compared with an actual increase in volume of 35,000Ml. (The inflows under the simulated operation were lower than the actual inflows because Dr Christensen reduced outflows from Somerset Dam by some 26,000Ml.)

7. [337]

   In fact, Wivenhoe recorded an inflow at 08:00 on 10 January of 10,100m3/s and a peak inflow at 13:00 on 11 January at 11,600m3/s. The peak outflow was 7,460m3/s on 11 January at 19:00.

8. [338]

   Although the calculations appear not to have been undertaken on this basis, given the purpose of mitigating flooding in the Brisbane River below Wivenhoe, the peak inflow at any point in time should have covered both the Wivenhoe and Somerset catchments. So as not to double count (outflow from Somerset being included in inflow into Wivenhoe), the calculation should have been undertaken on the basis of a net inflow into Somerset. (That figure could be negative if the outflow from Somerset into Wivenhoe exceeded the inflow into Somerset.) Further, although the focus in attempting to understand the general proposition articulated in the Manual was on prediction of inflows, little attention appears to have been given to concept of “peak outflow”, nor as to possible temporal elements. The assumption based on an actual “peak inflow” appears to have been that the releases could not rise above that peak at any time. However, having regard to temporal factors, if the peak inflow at any point permitted outflows which caused significant flooding downstream, to continue those outflows for a significantly longer period than the peak of the inflows might equally be thought inconsistent with the objective of flood mitigation.

9. [339]

   Seqwater submitted that because the Manual was at least unclear as to what was intended, a range of opinions was available. As the primary judge noted, three understandings were advanced, namely (i) peak inflow up to the point in time where releases were being made; (ii) the higher of the peak inflow up to that point in time and a predicted inflow based on rain on the ground, and (iii) peak inflow based on the best forecast of future rainfall and streamflow information. Each construction being accepted by several witnesses, other than (iii) which was supported only by Dr Christensen, Seqwater submitted that the judge had been wrong to conclude that a flood engineer could reasonably construe the Manual only as mandating the use of the best available forecast of rainfall in determining “peak inflow”.

10. [340]

    The judge accepted Dr Christensen’s view because, whilst Dr Christensen accepted the general principle or guideline that flood mitigation should not exacerbate the level of flooding by releasing water at a higher level than the peak of the natural flow of the river, he understood the Manual to allow for the risk of departure from that principle by use of predicted rainfall. There were benefits to be achieved by taking that course.

11. [341]

    The reasoning of the primary judge that the Manual permitted, and may have required, use of rainfall predictions in considering in advance what would be the “peak inflow” during the flood event should be accepted. Nevertheless to take the further step of concluding that such a reading was not only reasonable, but was the only reasonable reading of the Manual was open to challenge. The views of the flood engineers themselves, as conveyed by their evidence, were not readily dismissed as a retrospective justification of the conduct undertaken during the flood event. The actual conduct tended to confirm that they held such views at the time of the January flood. The primary judge did not find otherwise. However, whether or not one could say that the alternative constructions of the Manual were unreasonable, it is not possible to describe them as constructions which no reasonable flood engineer could have adopted in the circumstances.

12. [342]

    Grounds 10 and 11 should, to that extent, be upheld, although only as addressed in the written submissions, and not as formulated in the notice of appeal. That more limited finding is, however, sufficient in circumstances where the Court is satisfied that the standard provided by s 36(2) of the Civil Liability Act is engaged.

16. Operation of Somerset Dam – (grounds 14, 15)

1. [343]

   Somerset Dam is located on the Stanley River which flows from the ranges north of Wivenhoe in a south-westerly to southerly direction until it joins the Brisbane River. Somerset Dam long pre-dated Wivenhoe, construction having been commenced before but only completed in 1953. The dam was commissioned in 1956. Further, the structure of the dam differed in significant respects from that of Wivenhoe. It had a fixed crest at 100.45m AHD on which were situated eight radial crest gates, giving a total crest level at 107.45m with the gates shut. The crest operated as a spillway with a width of 135.33m. In a flood event, the crest gates could be raised so that the water simply flowed over the crest: unlike Wivenhoe, overtopping was an intended operation in times of flood, and did not threaten the structure of the dam.

2. [344]

   Water could also be released through eight sluice gates (and four smaller regulators) which are low down in the dam wall and well below the Full Supply Level (99m). The discharge from the regulators and the sluices increased significantly as the dam level increased. Thus, as revealed in Appendix D to the Manual, the discharge per sluice at 90.0m was 163m3/s, whereas at 105.0m, the discharge was 223m3/s. There was no significant discharge from the crest gates until the dam level reached 101.0m, at which stage the total discharge was a very light 32m3/s across the whole spillway. By the time the level reached 105m, the spillway discharge was 1,212m3/s. It became the dominant partner when the level was between 106m and 106.5m.

3. [345]

   At 00:00 on 2 January, the lake level was at 99.07m, that is some 7cm above FSL. At that stage there was a minimal outflow of 34m3/s. The inflow was, for most of 2 January, approximately twice the level of the outflow. As a result, by 00:00 on 3 January the dam level had risen to 99.17m. Subject to some variations, it rose slowly over the next four days, to reach 99.52m at 00:00 on 7 January. After significant inflows on 7 January the level reached 100.01m at 16:00 on that day, and the sluice gates were partly opened. However, the dam level continued to rise slowly to 100.46m at 08:00 on 8 January. By 09:00 on 9 January the level had fallen slightly to 100.28m, at which stage the inflows exceeded 1027m3/s.

4. [346]

   The Manual identified the broad flood operation strategies for Somerset as follows:

5. [347]

   The Somerset flood strategy flow chart was devised by reference to predicted levels in Wivenhoe. First, if Wivenhoe were not likely to exceed its FSL, Strategy S1 was to be adopted; if Wivenhoe were likely to exceed its FSL but initiation of the fuse plugs were not likely, Strategy S2 was to be adopted. If fuse plug initiation were likely, the prescribed strategy was S3.

6. [348]

   The focus of Strategy S1 was to minimise impact on rural life upstream from the dam and was to apply where Somerset was expected to exceed its FSL (99.0m) and Wivenhoe Dam was not expected to reach 67.0m (its FSL) during the course of the flood event. Because in early January Wivenhoe was always at or above 67.0m, S1 was not engaged at any stage. However, it is convenient to note the direction contained in S1, namely:

7. [349]

   Strategy S3 envisaged possible fuse plug initiation at Wivenhoe, which, for reasons discussed above, was not likely either on 2 January or 8 January 2011. Strategy S3 may, for present purposes, be disregarded. Accordingly, the relevant strategy at material times was S2. The purpose of that strategy was to “[m]inimise impacts below Wivenhoe Dam”, according to the title of the strategy. However, after setting out the conditions noted above, the strategy continued, in bold typeface:

8. [350]

   The box containing appropriate actions was divided into four parts. The three relevant parts were as follows:

9. [351]

   The third part was not engaged because at no relevant time was Wivenhoe falling, and a fourth part (also not engaged) concerned a flood event which emanated mainly from the Stanley River catchment without significant runoff from the Upper Brisbane Catchment (which fed directly into Wivenhoe). However, in each of the second, third and fourth boxes, the same constraint was noted, namely that the release rate was “generally not to exceed the peak inflow”.

10. [352]

    As the conditions disclosed, the critical level in relation to Somerset was 100.45m, which was the fixed crest of the spillway, and the bottom of the crest gates. Even with the crest gates raised, and Somerset below that level, there would be no outflow if the regulators and sluice gates were closed. Once the level exceeded 100.45m, with the crest gates raised, there would be an “uncontrolled discharge”, although the volume would be small unless the level continued to rise. Strategy S2 was silent as to whether the regulators and sluices may be opened, in contrast to the requirement that they were “generally kept closed” when the level was below 100.45m. However, the fourth part expressly stated that the valves and sluice gates were to be used to maintain the level below 102.0m, even while the crest gates were raised, the purpose being to protect the upstream bridge.

11. [353]

    Between 2 January and 6 January, as noted above, the level of Somerset was at all times below 100.45m. With the regulators and sluices closed, there would be no outflow into Wivenhoe. Further, the “operations target line” would not be engaged. However, because Somerset later rose above 100.45m, the engineers had regard to the operating target line and the relevant part of the Manual should therefore be set out.

12. [354]

    The operating target line apparently allows for the level of the dam to increase to 109.7m which is treated as the “likely failure level”. The notes state that this level is only achieved when the crest gates are fully open; it may be inferred that the top of the radial crest gates will rise to that level when open. Thus the failure level is reduced if full opening cannot be achieved.

13. [355]

    The issue in dispute between the parties turned on the fact that the flood engineers opened the sluice gates in increments after the level at Somerset reached 100.06m, before the beginning of the target line at 100.45m. Thereafter, the level of the dam steadily rose (inflows exceeding outflows) until 06:00 on 12 January when the level peaked at 105.11m. At that stage Wivenhoe was at 74.77m, a little below its peak on 11 January (74.97m).

14. [356]

    A significant issue at the trial was whether the statement in the first part of strategy S2 (with Wivenhoe rising and Somerset below 100.45) that regulators and sluices “are generally kept closed” involved a general prohibition or merely a guideline. The plaintiff’s case in this regard was somewhat obscure. Dr Christensen in simulation C would have opened the sluice gates when the lake level was at 99.07m (on 2 January) so as to release 200m3/s for approximately 24 hours from 2-3 January and then would have recommenced releases in excess of 130m3/s on 5 January. The effect of these releases would have been to lower the dam level below FSL and maintain it below FSL until 17:00 on 9 January, when the actual level was 101.14m, some 2m above FSL.

15. [357]

    For reasons set out above, there was no negligence on the part of the flood engineers in failing to reduce the levels in either dam below FSL. Accordingly, the steps taken in that regard by Dr Christensen were not those required for the reasonable operation of Somerset Dam in the interests of downstream residents.

16. [358]

    More importantly, if Dr Christensen considered it permissible under S2 to use the sluice gates to lower the dam level, it is unclear on what basis the plaintiff argued that use of the sluice gates during that period was not open to the flood engineers acting reasonably.

17. [359]

    The gravamen of Rodriguez’ case appears to have been that the flood engineers belatedly released too great a volume of water, thus increasing the level at Wivenhoe with consequences for the level of releases required from that dam. However, there were two answers to that case. The first was that higher levels of outflow were required in order to achieve a similar dam level when the heavy rains came as was achieved under Dr Christensen’s models. If there were no negligence in the flood engineers starting with a higher dam level (because they did not drop the level below FSL) there was no obvious negligence in their higher discharges at a later point in time which were a necessary consequence of the higher starting level.

18. [360]

    Secondly, that the higher discharges were achieved by opening sluice gates could not be the subject of complaint in circumstances where Dr Christensen considered that an appropriate course. No other witness supported negligence in this respect.

19. [361]

    On the assumption that, as discussed above, there was no negligence in terminating the flood event on 2 January, the question of releases from Somerset, in accordance with Strategy S2, could not arise until the further flood event was declared on 6 January 2011.

20. [362]

    Dr Christensen’s simulation F, commencing at 00:00 on 8 January, started with the actual lake level at 100.31m. His description of the strategy adopted was, “[b]egin operation under S3 strategy, constrained to keep gates open”. He presumably felt so constrained because Strategy S3 proposed steps “[i]n addition to the operating protocols used in Strategy S2” which, as noted above, required the crest gates to be raised, even where the dam level was below 100.45m. However, despite the fact that the dam level had exceeded 100.45m by 05:00 on 8 January, Dr Christensen worked on the assumption that there was no measurable outflow for the rest of the day (459Ml). With an inflow of 28,000Ml, the dam level rose to 100.82m at midnight.

21. [363]

    Grounds 14 and 15 read as follows:

22. [364]

    Although these grounds took issue with the reasoning of the primary judge in accepting Dr Christensen’s approach in respect of simulations C and F, there is substance in Rodriguez’ submission that the issue goes to causation only. There appears to be no relevant finding that the engineers were negligent in their operation of Somerset Dam on 8 January. While the judge noted the pleaded allegation that the reasonably prudent flood engineer should, on 8 January 2011, “have implemented and maintained Strategy S3 at Somerset Dam” and “would not have substantially increased the rate of inflow from Somerset Dam without implementing a corresponding increase in the rate of outflow from Wivenhoe Dam”, [120] the judge nevertheless found a breach in the broader terms that the engineers should have caused Wivenhoe to release water at rates exceeding the inflows: Ch 12 [156]-[157]. There was a finding that Mr Malone was negligent in failing to release water from Somerset at a rate exceeding inflows; [121] Somerset was 7cm above its FSL at the start of 2 January, but net inflows were a mere 30m3/s on average that day. The specific allegations in pars 211(c) and (e) of the statement of claim were rejected.

23. [365]

    In substance simulation F sought to reduce outflows on 8 January to allow the water level in Somerset Dam to rise, but the result was to require greater releases in later days, as indeed simulation F permitted. The primary judge accepted Dr Christensen’s explanation that this approach was beneficial, but it is by no means clear why that was so. By allowing the dam level to rise above 105m on 10 and 11 January, as opposed to the engineers’ height of about 103m, Dr Christensen was able to maintain outflows well below those in fact released up until 07:00 on 11 January when the engineers significantly reduced outflows and Dr Christensen started to increase them. Accepting that Dr Christensen’s approach, commencing on 8 January, was reasonable in the circumstances, it does not follow that the operations by the engineers were unreasonable. In terms of causation, no attempt was made in this Court to explain how and to what extent the adoption of simulation F would have lessened the releases from Wivenhoe, with consequential benefits for downstream flows.

24. [366]

    In broad terms, Dr Christensen’s course of action depended on the fact that, although conditions were benign on 8 January, the 4-day forecasts predicted heavy rain to come. His proposal was to lower the level of Wivenhoe as far as possible without causing downstream inundation prior to the inflows from the heavy rain, and maintain as much rain as possible in Somerset, to be released after the rain (and presumably peak inflows) had passed. This strategy appeared to be based on a degree of speculation as to when the rain would fall in the Somerset catchment as compared with the Wivenhoe catchment and when levels would peak in each dam. Avoiding hindsight, that was similar to the problem in predicting whether the heaviest rain would fall closer to the coast, and below Wivenhoe but in the Lockyer and Bremer catchments, or would fall above the dams. However, as a matter of practice, and as reflected in the pleadings, the overall question was how to flatten the releases from Wivenhoe so that they took place over a longer period and thus achieved a lower peak downstream.

25. [367]

    The primary judge found a breach of duty on the part of Mr Malone on 9 January in failing to ensure that the rate of outflow from Wivenhoe “substantially exceeded the rate of outflow from Somerset”: Ch 12 [181]. Again, the degree of particularity in the pleading of breaches created artificiality in the assessments of breach which distracted attention from the overall picture. Thus, during the high inflow periods of 9 and 10 January, 82% of the inflow to Wivenhoe came from the catchments above the dam, other than Somerset, and only 18% of the inflow came from Somerset. If there should have been further releases from Wivenhoe at or before that time, that was because other inflows caused the bulk of the rise in the dam level. While it is true to an extent that the inflow from Somerset was controllable, to treat the Wivenhoe releases as entirely referable to the Somerset inflow, in order to calculate how much should be attributed to other inflows was not an informative exercise.

26. [368]

    Further, the exercise did not have regard to the operating target line which was designed to determine the basis on which releases should be made. Its purpose was expressly stated to be to minimise the flood peaks in both dams having regard to their storage capacities, and to minimise flows downstream of Wivenhoe. [122] Although the target line was specified under Strategy S2, it was only appropriate to move into Strategy S3 when the structural safety of Wivenhoe was under threat. That permitted temporary departure from S2. Whether fuse plug initiation at Wivenhoe was ever expected depended on whether releases from Wivenhoe were being considered.

27. [369]

    At Rodriguez (22), Ch 10 [37], the judge mapped the actual levels of the dams and the levels proposed under simulation F, against the operating target line. The actual dam levels achieved by the engineers were at all stages between 8 and 12 January significantly closer to the operating target line than those proposed under simulation F. The judge addressed the criticism that this was a departure from the requirement of the Manual in the following terms:

28. [370]

    While it was true that one of the conditions in S2 was not engaged until Wivenhoe was rising, Wivenhoe had in fact been rising gently from 2 January until 8 January, with the largest inflow on 7 January. For a period on 8 January the outflows exceeded the inflows, but unless the engineers were expected to shift from one strategy to another and back again over a matter of a few hours, that was a minor point which could not entail negligence. The rest of the explanation was a justification for permitting the actual levels to move away from the target line: that course may have been justified, but it did not demonstrate that by adhering to the target line more strictly than Dr Christensen did the engineers were therefore negligent. One might have anticipated the reverse conclusion, consistently with other findings in relation to departures from the Manual. It is sufficient to conclude that there was no basis for a finding of negligence (on the ordinary standard) with respect to the manner in which the engineers operated Somerset releases between 8 and 12 January 2011.

29. [371]

    In dealing with questions of liability, it may be correct that grounds 14 and 15 are immaterial and may be disregarded. However, the findings of negligence based on them cannot be sustained; it follows that the findings cannot be upheld on the s 36(2) standard. To the extent that they form part of the underpinning to the finding of negligence on the part of Mr Malone in the operation of releases from Somerset on 9 January, grounds 14 and 15 should be upheld.

17. Role of Senior Flood Operations engineer – (ground 3)

1. [372]

   Ground 3 took issue with the judge’s conclusion that the Manual conferred on the Senior Flood Operations Engineer a position in a hierarchy above that of the Duty Flood Operations engineers for only a limited purpose. Ground 3 stated:

2. [373]

   The existence of the suggested hierarchy arose from both the use of the title “Senior” and the specific provisions in sections 2.3 and 2.4 of the Manual. Section 2.3 required the nomination of a suitably qualified person by Seqwater, to be approved by the Chief Executive, and continued:

3. [374]

   Section 2.4 dealt with the responsibilities of flood operations engineers and required that, when rostered on duty, they were to “[d]irect the operation of the dams during a flood event in accordance with the general strategy determined by the Senior Flood Operations Engineer” and were to “[f]ollow any direction from the Senior Flood Operations Engineer in relation to applying reasonable discretion in managing a flood event as described in Section 2.8.” Section 2.8, headed “Reasonable Discretion”, commenced as follows:

4. [375]

   The senior flood engineer during the January flood event was Mr Ayre, who was employed by SunWater. The judge noted that Seqwater and the State each sought to “deflect any or some responsibility they may have for their conduct of flood operations by asserting that they were acting in accordance with Mr Ayre’s direction”: Ch 3 [319].

5. [376]

   To the extent that there was some departure from the procedures under the Manual, that could only have followed from an exercise of discretion by Mr Ayre. The primary judge saw some difficulty in the suggestion that a general or overall strategy could be set at the outset of a flood event which would require flood engineers to act contrary to the requirements of the Manual as events unfolded: Ch 3 [324]-[327]. Seqwater challenged this reasoning.

6. [377]

   The primary judge returned to the question of Mr Ayre’s authority in other contexts. Thus, the judge found that Mr Malone’s rainfall analysis failed to provide “any justification for any failure to continue the flood event or declare a new flood event during the period 3 to 5 January 2011”: Ch 6 [143]. He then noted a submission that Mr Ayre had authority to “direct Mr Malone to continue the flood event from 2 January 2011 or declare a flood event himself.” However, the judge accepted the submission of SunWater that Mr Ayre had no such authority: Ch 6 [144]. The judge read the powers conferred on the senior flood engineer by section 2.3 as only available when that engineer is “rostered on duty during a Flood Event.” Thus the senior flood engineer had no power to direct Mr Malone not to terminate the December flood event whilst it was still ongoing, unless he was at that moment rostered on duty himself.

7. [378]

   That might be thought a surprising conclusion. At an earlier point, the judge had noted that cl 2.2 of the Manual stated that a senior flood operations engineer is designated “to be in charge of Flood Operations at all times during a Flood Event”: Ch 6 [145]. In fact, Mr Ayre was himself on duty on the evening of 1 January 2011 and the morning of 2 January 2011, and, as the judge observed, had “the capacity to take action to have the flood operations engineer rostered on duty comply with the Manual”: Ch 6 [67]. The finding that Mr Ayre “joined in the consensus to end flood operations”, was not a finding which denied the senior flood operations engineer the power and authority to direct a different outcome.

8. [379]

   The issue is, however, primarily relevant to the application of s 36(2) and apportionment of liability for damages, which are addressed below. Mr Ayre’s employer, SunWater, took no relevant part in the appeal and it is neither necessary nor appropriate to make findings beyond those necessary to dispose of the issues which remain alive on the appeal between Seqwater and Rodriguez.

18. Breaches of duty – Rodriguez’ application to rely on s 36(2)

1. [380]

   The important conclusions from the foregoing analysis are that:

   1. (1)

      the standard of negligence by which Seqwater’s liability should have been tested was that provided by s 36(2) of the Civil Liability Act; and

   2. (2)

      there was no breach of duty on the part of Seqwater’s flood engineers in:

2. [381]

   On the second last day of the hearing, the Court granted Rodriguez leave to contend that “the acts or omissions found by the primary judge to constitute breaches of duty for which the appellant was liable were, in the circumstances, so unreasonable that no public or other authority having the functions of [Seqwater] could properly consider those acts or omissions to be a reasonable exercise of its functions for the purposes of s 36(2)”. [125] The Court’s reasons for making that order follow. It then remains necessary to consider the acts or omissions found by the primary judge to constitute breaches of duty for which Seqwater was liable, because it was those acts and omissions to which Rodriguez’ contention 3 was directed. As will appear, they did not include Mr Malone’s alleged breaches of duty on 2 January in bringing the December Flood Event to an end and by failing on 3, 4 and 5 January to declare a further flood event and make releases to reduce the level of Wivenhoe below FSL.

3. [382]

   The primary judge determined that s 36 did not apply, and refrained from determining whether, if he were wrong about that, the flood engineers’ acts or omissions were so unreasonable that no public or other authority having the functions of Seqwater could properly consider them to be a reasonable exercise of its functions: Ch 1 [88].

4. [383]

   At the forefront of Seqwater’s appeal was its challenge to the determination that s 36 was inapplicable. This was the first ground of its notice of appeal. Seqwater’s written submissions in support of its appeal were dated 22 June 2020, 11 months before the hearing. Seqwater maintained at the outset of those submissions that not only did s 36 apply, but further that “the first respondent never sought to plead, nor run, a case against Seqwater based on this higher standard [namely, that imposed by s 36(2)]” (par 1.16). That was developed at the conclusion of Seqwater’s submissions on ground 1:

5. [384]

   Rodriguez’ written submissions of 144 pages were dated 18 December 2020. Rodriguez sought to defend the primary judge’s reasons on the question of the applicability of s 36, save that it challenged the judge’s conclusion that the section was not confined to claims for breach of statutory duty, and advanced reasons additional to those on which his Honour had relied. Rodriguez also maintained a fallback submission:

6. [385]

   That submission was developed at pars 46-50 of its written submissions, which should be reproduced in their entirety in light of the Court’s ruling on 28 May 2021 on the notice of contention: [126]

7. [386]

   As the primary judge refrained from making a finding in relation to the standard set by s 36(2), Rodriguez’ fallback submission contended that the judgment should be affirmed on a ground which was not part of his Honour’s reasons. On any view, this was a matter for a notice of contention. Rodriguez made no submission to the contrary. UCPR r 51.40(1) provides:

8. [387]

   The notice of contention should have been supplied by late July 2020. It was a point of substance, especially in an appeal of this magnitude, given that it involved further fact-finding, which in turn would potentially involve consideration of the evidence bearing upon breach. The directions governing the preparation of these appeals for hearings required the parties to include in physical appeal books those documents to which reference was made or was expected to be made in written or oral submissions, supplemented by other documents to be made available electronically. The result was that only a tiny fraction of the material tendered at trial and addressed at the hearings (the transcript was some 10,000 pages) was reproduced in the physical appeal books, and the appeal was largely conducted on the basis of the physical appeal books (although documents which had been tendered at trial but not reproduced in the physical appeal books were available electronically and from time to time relied upon). It is unrealistic for any practitioner to be able to remember in 2021 the entirety of the evidence of a very lengthy trial that stretched from 2017 to 2019, and occupied most of 2018; hence the significance of the selection of a small minority of documents for inclusion in the physical appeal books (which, even so, were somewhat more than 5,000 pages). Those considerations strengthen the importance of compliance with the rules in identifying the issues for determination on appeal.

9. [388]

   Seqwater’s submissions in reply were dated 9 April 2021, which is to say, five weeks before the hearing. Seqwater complained that two of Rodriguez’ submissions should not be entertained without a notice of contention, but did not make that complaint about what was put forward in relation to s 36(2). Instead, Seqwater made the following submission:

10. [389]

    A footnote referred to Suttor v Gundowda Pty Ltd [128] and Coulton v Holcombe. [129] The references to the opening and closing submissions were to the transcript references footnoted in Rodriguez’ submissions. They were fairly described as “fleeting” and “brief”. The point was mentioned but left undeveloped.

11. [390]

    So far as is apparent from the materials available to this Court, there were no subsequent communications between the parties concerning s 36(2).

12. [391]

    In opening the appeal on 17 May 2021, senior counsel for Seqwater maintained that reliance on the acts and omissions being unreasonable within the meaning of s 36(2) was a point which “needs to be pleaded if it is going to be run” [130] and continued after reference was made to brief references in opening and closing submissions:

13. [392]

    Ten days later, on 27 May 2021, the second last day of the hearing, at the conclusion of all other aspects of the appeal and the cross-appeal, senior counsel for Rodriguez applied for leave to rely on a notice of contention. The first two paragraphs maintained that the primary judge should have found that s 36 did not apply, because the proceeding was not based on the alleged wrongful exercise of a function of a public authority, or because the section only applied to breach of statutory duty. Seqwater consented to Rodriguez being able to advance those submissions.

14. [393]

    The third paragraph of Rodriguez’ notice of contention was:

15. [394]

    Seqwater opposed Rodriguez being granted leave to rely on the third paragraph of its notice of contention. It asked this Court to refuse the application, on two bases: [132]

16. [395]

    The parties’ submissions thereafter will be summarised below. At the commencement of the following day, this Court proposed the following ruling:

17. [396]

    Noting that that was an intermediate approach, the Court invited the parties to review the ruling during the course of the day, and granted them leave to advise of any concerns. At the conclusion of the hearing (and after the luncheon adjournment), both sides confirmed they did not wish to be heard further about the ruling on the notice of contention.

18. [397]

    Our reasons for that ruling, which involved a rejection of both of Seqwater’s submissions in opposition to the late filing of the notice, but also the imposition of a term restricting the use which Rodriguez could make of the notice, confined by reference to the written submissions, were as follows.

19. [398]

    It was strictly correct for Seqwater to submit that Rodriguez had not pleaded s 36(2). But that reflected the unusual course taken by the pleadings.

20. [399]

    Throughout its defence, Seqwater had invoked s 36, alleging not merely that it applied, but going further and alleging that “the Flood Engineers’ acts and omissions were not so unreasonable that no public authority having Seqwater’s functions could properly consider those acts or omissions to be a reasonable exercise of its functions” and accordingly by s 36 those acts and omissions were not wrongful: see pars 299(c)(i) and (ii) in answer to the allegations of breach on 2 January 2011; later paragraphs adopted the same form.

21. [400]

    It may very well have been, strictly speaking, unnecessary for Seqwater to allege the negative proposition that the acts were not sufficiently unreasonable, but for the plaintiff to contend that the case fell within the qualification or exception. Where it applies, s 36 sets the standard of care against which the defendant’s acts or omissions fall to be assessed. That is the force of providing that the authority’s acts or omissions are not a wrongful exercise or failure to exercise unless the unreasonableness standard has been made out. Accordingly, it might be thought that it was for Seqwater to plead that s 36 applied to the proceeding, but thereafter it was for Rodriguez to make out a breach of the unreasonableness standard, rather than for Seqwater to allege that its conduct had not been unreasonable within the meaning of s 36(2).

22. [401]

    This background discloses a shortcoming in the way the issue was fought at trial. Had Rodriguez positively alleged that the flood engineers had been unreasonable within the meaning of s 36(2), the allegation would likely have been accompanied by particulars. Because Seqwater asserted the negative, and Rodriguez merely denied that assertion, no such particulars were provided. As Seqwater put it: [133]

23. [402]

    Little weight should be placed on the absence of particulars of the denial in Rodriguez’ reply. Where a denial involves in substance the setting up of a positive case, particulars may be ordered, as Kitto J explained in George v Federal Commissioner of Taxation. [134] In turn, Aickin J stated that there were “many situations in which the party who gives a general denial to the pleading of the party on whom the onus rests may none the less be required to give particulars if the general denial really involves some positive allegation”, and that this general principle was well established: Bailey v Federal Commissioner of Taxation. [135] Whether or not Seqwater requested particulars of the denial is not clear from the materials available to this Court.

24. [403]

    It is unnecessary to express a concluded view as to whether, if s 36 applied, it was for Rodriguez to allege unreasonableness. What matters is that by its reply, Rodriguez denied each of Seqwater’s allegations invoking s 36(2). It follows that whether the acts or omissions of the flood engineers for which Seqwater was vicariously liable were unreasonable within the meaning of s 36(2) was in issue on the pleadings.

25. [404]

    As much was confirmed by common questions 11 and 14(b), which were the subject of orders made on 14 September 2018. These were not initially relied upon by Rodriguez, but were raised by members of the Court during the application to rely on the notice of contention. Each question asked the primary judge to answer whether, inter alia, Seqwater had acted in a way that was so unreasonable that no authority having the function or power in question could properly consider the acts or omissions to be a reasonable exercise of the function or power within the meaning of s 36(2). Towards the end of the trial, each of Seqwater and Rodriguez supplied submissions as to the answers each favoured. The primary judge, consistently with his reasons, answered those questions by noting that they did not arise. But the foregoing makes it plain that the issue was live on the pleadings and treated as such throughout the trial.

26. [405]

    For those reasons, we rejected Seqwater’s submission that the notice of contention raised a fresh point that could not be raised in this Court.

27. [406]

    Seqwater submitted that it was unfair for Rodriguez to seek to make out a case of unreasonableness falling within the exception to s 36(2) because “the factual inquiry would be quite extensive to try to understand what evidence, in the mass of material, could be relevant” to the issue. [136] The same point had been advanced in Seqwater’s written submissions in reply. Seqwater emphasised that the question of breach was “absolutely central” [137] to any plaintiff’s case, and thus the applicable standard was one to be addressed at the forefront of the case.

28. [407]

    Rodriguez pointed to a schedule of 38 extracts from the judgment entitled “Rodriguez’ Schedule of findings from the liability judgment – s 36(2)” which Seqwater’s counsel stated [138] was supplied on the morning of the ninth day of the appeal. The document included assessments by the primary judge that the flood engineers had advanced constructions of the Manual and conducted flood operations which were “completely untenable”, “unreasonable” and indeed “entirely unreasonable” and “completely unreasonable”. Seqwater complained that some of these passages were directed to events of 2-5 January, some were directed to Mr Ruffini, some were directed to points taken by Seqwater in the litigation rather than acts or omissions of flood engineers, and many were not connected (or, at least, were not obviously connected) with any particular breach of duty. The submission concluded: [139]

29. [408]

    The larger point is that the force Rodriguez sought to attribute to those passages was inapt. The primary judge had explicitly declined to determine the issue presented on the pleadings by s 36(2), formalising that in his answers to two of the common questions. The occasional rhetorical flourish in a very long judgment, made in different contexts, does not translate into a finding for the purposes of s 36(2), being an issue which the primary judge declined to determine.

30. [409]

    The parties distinguished the written submissions on s 36(2) which had been supplied months before the hearing, and the schedule provided on the ninth day of the appeal. Thus Mr Sexton said “[m]y learned friends’ contention that he shouldn't now have to deal with that schedule is an argument that he could have raised anyway, and if it’s a valid complaint, then your Honours would not receive that schedule. But that doesn’t go to whether or not a notice of contention should be filed.” [140]

31. [410]

    In those circumstances, Rodriguez was permitted to advance its contention that s 36(2) was satisfied in accordance with its written submissions on that point (which have been reproduced above). However, it would have been unfair to permit Rodriguez to advance, on the ninth day of appeal in support of a belated notice of contention, submissions which went beyond that which had been advanced in writing. Rodriguez had had some 10 months to supply a notice of contention and any further submissions. It had had 5 weeks before the appeal was heard, and the previous 8 days of the appeal, to formulate with any greater precision how it sought to advance the point. It was squarely on notice of Seqwater’s objections to the point being run at all having regard to the paucity of submissions on the point at trial.

32. [411]

    This is a very large appeal. Nonetheless ss 56 and 58 of the Civil Procedure Act 2005 (NSW) apply, and in determining whether to permit Rodriguez to rely on a very late notice of contention, and if so on what terms, this Court is required to give effect to the overriding purpose of facilitating the just, quick and cheap resolution of the real issues in the proceedings, having regard to the dictates of justice, pursuant to s 58.

33. [412]

    The order made by the Court granting leave to file the notice of contention on terms recognised that while it was not appropriate to forbid Rodriguez from pursuing a submission which had been advanced some 5 months prior to the hearing, and which had been met in writing by Seqwater on its merits, it was also not appropriate to permit Rodriguez to expand beyond that submission to findings which had not been made by the primary judge, or by reference to a schedule provided belatedly which failed to articulate the basis on which the new finding was sought to be made. We accepted Seqwater’s submission that permitting the latter course, by reference to the schedule of passages from the main judgment which was supplied on the second last day of the hearing, had real scope to give rise to prejudice, in terms of prolonging the hearing of the appeal, in circumstances where there was no explanation proffered as to why Rodriguez, being squarely on notice of the issue for months, had only raised the matter at the conclusion of the hearing. However, we did not understand that Rodriguez otherwise wished to elaborate upon its written submissions. (Its oral submissions in support of the substance of the s 36(2) point, as opposed to being granted an extension of time to file the notice of contention, did not go beyond what was put in writing. [141] ) That may well have been because the written submissions encapsulated the high point of what could be advanced on this issue.

34. [413]

    The limitations accompanying the grant of leave extended also to the legal basis on which Rodriguez sought to advance its case. During oral argument, senior counsel for Rodriguez maintained that the test was objective, and did not turn on the way the flood engineers had been cross-examined.

35. [414]

    The submissions advanced in support of Rodriguez’ notice of contention covered the entirety of the period 2-10 January 2011. As already noted, the complaint that Mr Malone had unreasonably failed to declare a new flood event was withdrawn: it was untenable. It was not unreasonable for Mr Malone to accede to Mr Ayre’s decision to bring the December flood event to an end on the morning of 2 January 2011. Once that is accepted, nothing turns on what happened over the ensuing three days. Nevertheless we propose to deal with the challenges to the findings as to breaches of Mr Malone’s duty of care as a reasonably competent flood engineer during the period 2 to 5 January. They remain relevant to the finding that any reasonably competent flood engineer commencing operations at midnight on 2 January 2011 would have, at a minimum, made flood releases substantially in accordance with simulation C. That finding is the subject of the specific challenges made by grounds 16, 17 and 18. However if the challenges made to the primary judge’s findings with respect to breaches on 2-5 January are upheld, that overarching finding (applying simulation C) would no longer be correct and accordingly cease to be relevant as a counterfactual in any causation or damages analysis.

19 Simulations C, F and H (grounds 16-21); breaches on 2-5 January 2011 (grounds 22, 23(a))

1. [415]

   Success for Rodriguez ultimately involved it establishing negligence on the part of the flood engineers (for present purposes the issues raised by s 36(2) are put to one side) which resulted in levels of inundation of significant areas of Brisbane and Ipswich which either would not have occurred or would not have occurred to the extent experienced. That in turn required that Rodriguez formulate ‘counterfactual’ flood release operations against which the outcomes of the engineers’ alleged negligent operations could be assessed. The strategies adopted at the commencement of the counterfactual had to correspond with the alleged non-negligent flood operations relied on in Rodriguez’ breach case. Eventually (by Dr Christensen’s response report of July 2017) 10 simulations (identified by the letters A to J) were proposed, adopting various start times and assumptions as to the conduct of the flood operations: Ch 8 [135]-[167].

2. [416]

   It is convenient to deal with grounds 16-21 together. They are not easy to paraphrase and may best be set out in full as follows:

3. [417]

   Grounds 16, 17 and 18 were each directed to the judge’s finding that from midnight on 2 January 2011 releases should have been made substantially in accordance with simulation C. Further, that course was said to have been required up to 9 January 2011. Grounds 19 and 20 addressed the similar, but differently qualified, finding in relation to simulations F and H. The qualification was that the releases were to be as varied by Mr Ickert’s response report. Ground 21 challenged each of those findings in the event that any of Seqwater’s challenges to the interpretation and application of the Manual were upheld which affected simulations C, F or H. It follows from our conclusion that ground 8 is made out that the release strategies and decisions based on these counterfactuals cannot be sustained because each depended on the “no release” assumption.

4. [418]

   Ground 22 challenged the finding that there was a breach of duty in terminating the flood event on 2 January 2011, and ground 23(a) the findings that Mr Malone breached his duty of care on 3, 4 and 5 January “by failing to continue or commence flood operations contrary to the Manual, failing to adopt strategy W3 and not making releases substantially [exceeding] the rate of inflows, being at a minimum of the release rates in Simulation C”.

5. [419]

   The following challenges are addressed below:

   1. (1)

      to the overarching findings made as to simulations C, F and H (grounds 16 to 20);

   2. (2)

      to the findings of breach of duty on 2 January 2011 in bringing the December Flood Event to an end and failing to make further substantial releases (ground 22); and

   3. (3)

      to the findings of breaches by Mr Malone on 3, 4 and 5 January 2011 by failing to make releases to reduce the dam level and to declare a further flood event (ground 23(a)).

6. [420]

   Simulation C commenced at 00:00 on 2 January. Strategies were selected based on 24-hour QPF forecasts and a no release assumption. Longer term rainfall forecasts were used for “situational awareness”. During the flood event the Somerset crest gates remained open and the level of each dam could drop below FSL: Ch 8 [146]-[149].

7. [421]

   The objective of simulation C was to create flood storage capacity below Wivenhoe’s FSL of 67m and to do so by releases which closed all of the downstream bridges by the end of 6 January. The differences between the simulated and actual releases measured by flow rates at Moggill are shown on the following graph. By 06:00 on 10 January the simulated releases created additional storage capacity in the two dams of about 560,000Ml. That additional storage permitted later releases at significantly lower rates than those in fact made, is shown by the higher peak of the actual operation (red line) compared with the simulated operation (green line).

8. [422]

   Simulations F and H commenced on 8 January 2011. Simulation F used eight-day forecasts to determine strategy and four-day forecasts to determine releases and assumed that the Somerset crest gates had to remain open during a flood event: Ch 8 [159], 10 [5]. Simulation H, on the other hand, used 24-hour QPF forecasts to select strategy and assumed the Somerset crest gates had to remain closed: Ch 8 [162], 10 [5]. However, in the result, each proposed the same releases: Ch 10 [6]. The differences measured at Moggill between actual releases and those provided for by simulations F and H are shown in the graph below:

9. [423]

   In the period from 8 January Rodriguez contended that the engineers should have made releases at levels which when combined with the natural flows at Moggill did not exceed 4000m3/s. That continued until the morning of 11 January when it became apparent that those flows could not be kept below 4000m3/s because of the level of water in Wivenhoe. At Somerset the sluice gates were closed to store as much water as possible until midnight on 14 January: Ch 8 [160]. By 06:00 on 10 January the simulated releases created additional storage of about 190,000Ml, again permitting later releases at lower rates.

10. [424]

    Each of grounds 16, 17 and 18 attributed significance to the finding as to simulation C which it did not have. There was no finding of breach directed to the period addressed by simulation C. With two exceptions, the allegations of breach were made by reference to each engineer’s conduct on a particular calendar day in the period from 2 to 11 January. (The exceptions were the three-day period from 3 to 5 January and the two-day period of 10 and 11 January.) Each of the pleaded allegations necessarily addressed the actual and forecast rainfall and dam levels that existed during that day. To the extent that in the first days of the simulation the modelled conditions were the same or very similar to those which existed in fact, the challenges made by grounds 22 and 23(a) indirectly address the substance of his Honour’s findings with respect to the early part of simulation C. However, the grounds do not challenge the overarching finding on that basis.

11. [425]

    The primary judge observed in Ch 12, in response to Seqwater’s submission that Rodriguez’ case on breach had to establish that on each and every day of the flood event the relevant engineer failed to act in accordance with one or more of the simulations:

12. [426]

    As the primary judge recorded in this last observation, the principal significance of each of the challenged findings was that it provided a basis for Rodriguez’ causation and damages case in the event the engineers were found to have been negligent on 2 January, and perhaps in the period to 5 January. That case was that, but for such negligence, the flood operations would have been conducted in accordance with simulation C. Dr Altinakar’s modelling based on that simulation was then relied on as estimating the level of flooding that would have occurred if the outflows from Wivenhoe were made in accordance with simulation C, as distinct from those which in fact occurred: Ch 13 [2]. It follows that ground 16 should be dismissed on the basis that the overarching finding concerning simulation C was not outside Rodriguez’ pleaded (breach) case.

13. [427]

    Grounds 17 and 18 addressed the substance of that finding as directed to a damages counterfactual contended for in Rodriguez’ case. Ground 17 does no more than point out that in a minor respect his Honour’s finding departed from the releases that the reasonably competent engineer operating in simulation C “would have made”. However, that departure did not have the consequence that the releases proposed were no longer “substantially in accordance with” simulation C. Accordingly ground 17 should be dismissed.

14. [428]

    Addressing ground 18, the judge’s conclusion that the inundation of the two remaining bridges on 6 January should be delayed by about 12 hours did not adopt “a materially different methodology [from that of Dr Christensen] which was unsupported by evidence”. Whether Dr Christensen’s methodology and opinion were justified is a different question, and, in relation to the challenged findings of breach on 6 and 7 January, is dealt with below. His Honour concluded that the one respect in which his findings, as to what a reasonably competent flood engineer “must have decided”, departed from simulation C was not material for the use of that simulation in a damages counterfactual: Ch 10 [152], [153]. That conclusion was in part justified by the analysis set out in appendix G to his reasons: Ch 10 [155]-[160].

15. [429]

    Ultimately, as the primary judge explained in Ch 10:

16. [430]

    Ground 18 is not made out.

17. [431]

    The same analysis applies to ground 19 which, like ground 16, contended that, by reason of acceptance of variations to Dr Christensen’s releases proposed by Mr Ickert, simulations F and H were outside Rodriguez’ pleaded case.

18. [432]

    The variations proposed by Mr Ickert involved accepting that on the evening of 11 January a reasonably competent flood engineer could have continued gate openings as the level of Wivenhoe dam was above 74m. The consequence of adopting Mr Ickert’s variations was that slightly more water was released from Wivenhoe than was otherwise provided by simulations F and H. The outcome was, favourably to Seqwater, that the revised simulated flood operations on 11 January were broadly similar to the flood engineers’ actual operations: Ch 9 [333]. Mr Ickert’s variations were subsequently taken into account by Dr Altinakar in his modelling; the varied counterfactual described by the primary judge was not “invalidated” for that reason. Whether the methodology and reasoning underlying it was justified was not addressed by ground 19, which should be dismissed.

19. [433]

    Ground 20 contended that these simulations were not available because the primary judge made no finding that a reasonably competent engineer would have made the Ickert variations. That is not so. It was sufficient to justify the use of the varied simulated flood operations as a damages counterfactual that the primary judge found that there were two approaches which a reasonably competent flood engineer might have taken. That is the effect of his Honour’s finding that Seqwater was entitled to the benefit of the “most favourable” conduct consistent with competent operations. Ground 20 is not made out.

20. [434]

    Ground 21 turned on favourable outcomes in relation to one or more of grounds 6 to 15 as resulting in his Honour’s findings as to these counterfactuals being rejected. This ground is made out because Dr Christensen’s strategies adopted in each of simulations C, F and H turned on a “no release” assumption.

21. [435]

    Ground 22 challenged the finding that there was a breach of duty in terminating the flood operations on 2 January 2011. Over the period 2, 3, 4 and 5 January 2011, Wivenhoe water levels ranged between 67.1m and 67.28m. They were below 67.25m until midday on 5 January 2011, when they rose from 67.25m to 67.28m over the last 12 hours of 5 January. Thus, if there were no breach of duty in declaring the December Flood Event at an end at 09:00 on 2 January 2011, then there could be no breach of duty in not making further releases until midday on 5 January. At least that is so unless it were not only permissible but necessary, in order to avoid later flood damage, to disregard the Manual’s instruction not to open the flood gates in a new flood event until the dam level exceeded 67.25m. Rodriguez made no such submission. In this circumstance it is convenient to note the judge’s findings in relation to the four-day period 2-5 January 2011.

22. [436]

    In Ch 12, the primary judge made the following findings of breach. On 2 January 2011, proper compliance with the Manual required that the flood event continue, not just until dam levels reached FSL, but so as “to continue to bring the water level below FSL such that it was less than the one day ‘no release’ rise”: Ch 12 [64], Ch 10 [80]. The relevant risk was identified thus: “absent ongoing releases from Somerset Dam and Wivenhoe Dam, there would be insufficient flood storage capacity in Lake Somerset and Lake Wivenhoe to prevent urban flooding downstream of Wivenhoe Dam should further rainfall occur in accordance with, or in excess of the BoM forecasts”. This risk was based on two materially identical formulations in the pleadings, set out at Ch 12 [53]. The judge did not accept that “if rainfall occurred ‘in accordance with’ the forecasts then the risk pleaded ... arose”. Rather, he relied on the possibility that rain might fall in excess of the forecast, in light of the La Niña seasonal conditions, the fact that summer was far from over, the widespread flooding that had already been experienced and the immediate past experience which pointed to the real possibility of higher rainfall of the necessary order falling on a saturated catchment. Based on that possibility, his Honour accepted that there was a risk, as pleaded, namely, the chance that rainfall in excess of predictions might cause there to be a need for additional flood storage capacity in the future, in order to prevent substantial releases (sufficient to inundate urban areas) made necessary because of the prospect of dam levels in excess of 74m: Ch 12 [57]. This risk was held to be foreseeable and not insubstantial.

23. [437]

    The precautions which should have been taken against that risk were, according to the primary judge, a drain-down intended to achieve a dam level of 66.5m, which would represent an additional 53,000Ml of storage being made available, and which reflected the forecast one-day inflows. His Honour’s assessment of breach in Ch 12 included the following:

24. [438]

    Against those considerations, the burden of making additional releases was relatively slight, given that the low-lying bridges were already inundated. The judge concluded:

25. [439]

    It is convenient to note the reasoning with respect to the following three days, before returning to the question of breach in terminating the December Flood Event on 2 January. On 3 January 2011, there was little if any rain; inflows were still occurring from earlier rainfall, and the 4-day and 8-day PME forecasts predicted substantial rain for 6 January (up to 150mm). The 8-day PME forecast no significant rain beyond that time. The primary judge said:

26. [440]

    Although this assessment was premised on a calculation of inflows without releases, the primary judge was alert to the capacity to make releases before the rain fell, but regarded that as not undermining the existence of a foreseeable and not insignificant risk. His Honour referred to his reasoning in Ch 6 in the following terms:

27. [441]

    The judge found that “the releases in Simulation C set the minimum bound on what was required on these days” (Ch 12 [85]) and held:

28. [442]

    The primary judge addressed 4 and 5 January more concisely and made materially identical findings in relation to 4 and 5 January, at Ch 12 [94] and [102].

29. [443]

    With respect to Ch 6 [137] set out at [440] above, and the findings of breach on the later days, two things may be noted.

    1. (1)

       Although there were four breaches found, in substance they were twofold: (i) bringing the December Flood Event to an end, and (ii) not making substantial releases so as to bring Wivenhoe sufficiently below FSL to accommodate the one-day inflow estimate. The second and third findings of breach say nothing that is not within the scope of the fourth finding of breach.

    2. (2)

       If the first finding of breach is in error, the balance falls away, because the dam level was less than 67.25m throughout this period.

30. [444]

    The primary judge found that the December Flood Event should not have been brought to an end at 09:00 on 2 January 2011. After that point, and until the afternoon of 7 January, the regulators were open, releasing some 50m3/s, while the level of water in the dam was at 67.1m, rising to 67.15m by 06:00 on 3 January, 67.2m by 10:00 on 4 January, 67.25m by midday on 5 January and 67.3m at 03:00 on 6 January.

31. [445]

    The primary judge found that “[g]iven the terms of the Manual, the flood engineers had no choice but to continue the flood event on 2 January 2011”: Ch 10 [78]. That conclusion was based on the proposition that a flood event only ends when FSL is reached, and on 2 January 2011, the dam level was 10cm above FSL. That being a fair reading is confirmed by the balance of the paragraph:

32. [446]

    On the morning of 2 January 2011, when the dam level was 67.1m, there were some 11,000Ml in excess of FSL. Another way of putting that is that around 99.5% of the total flood capacity, or more significantly 99% of the flood capacity at 75.5m (1,160,000Ml), as stated in Appendix C of the Manual, remained available. As noted in discussing degrees of tolerance in part 10(7) above, the Manual did not specify levels to the nearest centimetre. Indeed, there is difficulty in measuring levels to the nearest centimetre. The evidence disclosed that the monitoring equipment showed a variation of +/- 2cm.

33. [447]

    The level of tolerance is not determined merely by the inevitable inaccuracy of measurement. The Manual proceeds much more approximately, making decisions based on +/- 25cm. No strategy in the Manual turns on a water level increment of less than 25cm. Another way of seeing this is that if the dam is at FSL, there is not a flood event every time it rains and the engineers expect the water level to rise to 67.05m, still less to 67.02m or 67.01m.

34. [448]

    Mr Malone’s view was that Wivenhoe Dam was effectively, or for practical purposes, at FSL when, at 08:00 on 2 January 2011, it had reached 67.1m and the last gate was completely closed (leaving outflows limited to the two open regulators). The primary judge disagreed on the basis that the Manual required levels to reduce to FSL at the end of the flood event. The judgment did not suggest that any express attention was given during the trial to the notion of tolerance in relation to measurements of water levels in the dams; nor did the parties’ written submissions in this Court address that concept. Rather, the primary judge concluded that “when sections 8.5 and 9.4 are read together with section 2.2, it follows that a flood event cannot end while either dam remains above FSL.” Furthermore, he found that “[a]ny contrary view is unreasonable”: Ch 3 [142].

35. [449]

    Mr Malone’s decision to terminate the flood event on 2 January was driven by two considerations. The first was his view that the dam was effectively at FSL; the second was his concern to allow volunteers to rescue lung fish from the pond at the base of the spillway when the volunteers would be available, namely on the weekend. The second factor was undoubtedly a reasonable consideration in terms of the objectives identified in the Manual (though it was a low level objective); the first factor, regarding the reasonableness of the conclusion that releases had reduced the water level to FSL, involving as it must a notion of tolerances for small variations, will be addressed below. It is convenient first to consider what was involved in terminating a flood event.

36. [450]

    In this context, “Flood Event” does not simply refer to the objective circumstance that there is a flood, but rather is a defined term in the Manual and means “a situation where the Deputy Flood Operations Engineer expects the water level in either of the Dams to exceed the Full Supply Level”. The practical consequence of the engineer forming that view is that steps are taken to establish a flood operations centre, which has a physical location in Brisbane and involves the allocation of staff and resources. It demands no necessary or immediate action with respect to releases of water from the dams, which, as noted above, cannot commence with respect to Wivenhoe until it reaches a level of 67.25m.

37. [451]

    As the primary judge correctly observed, the Manual does not expressly identify when a flood event ceases: Ch 3 [139]. For example, although section 8.5 of the Manual states that “final gate closure should occur when the lake level has returned to Full Supply Level”, it does not state that that is the end of the flood event.

38. [452]

    The primary judge rejected the proposition that a flood event should terminate when the water level reached FSL, and the gates were closed, because the situation could involve a continuing (or new) expectation on the part of the flood engineers that the water level will exceed FSL. In that case, it would be necessary for the engineers immediately to declare a further flood event: Ch 3 [140]. The judge continued, “in such a scenario … the gates at Wivenhoe Dam might be closed but the flood event would not cease until the flood engineer no longer expected FSL to be exceeded having regard to actual and predicted weather.” On that approach, Mr Malone could properly have directed “final gate closure”, in accordance with section 8.5, to minimise the stranding of fish downstream, without terminating the flood event.

39. [453]

    One reason for adopting this construction of the Manual was that it would permit reopening of the spillway gates for flood control purposes before the level exceeded 67.25m, despite the express prohibition on opening gates contained in section 8.3. However, the prohibition was not contingent on the declaration of a flood event, but rather on the fact of the gates being closed. The administrative steps accompanying that action would have been understood by the engineers and did not need to be spelled out in the Manual.

40. [454]

    As explained above, in dealing with the proper approach to the Manual, the primary judge appears to have accepted that, being a document prepared by and addressed to flood engineers, it should be construed accordingly, and not as a legal document. On that approach, it may not matter when precisely a flood event terminates or is terminated. What matters is whether, once the gates are closed, the prohibition against reopening until the level exceeds 67.25m is met. However, in a practical sense, final gate closure may terminate the flood event, in which case it would be expected that the flood operations centre would be closed by the flood engineer then on duty.

41. [455]

    The question is whether negligence was established on the part of the flood engineers in achieving a final gate closure on 2 January 2011. For reasons elaborated on below, the answer to that question is, no. First, for practical purposes, a reasonable flood engineer was entitled to treat the level of Wivenhoe as at FSL; secondly, the relevant weather forecasts at that time did not predict heavy rainfall which would lead to an expectation that FSL would be exceeded in a practical sense. At least, it was not demonstrated that an engineer who held such a view was acting unreasonably.

42. [456]

    Relevant to the levels of tolerance in the draining to FSL, section 8.5 addressed the position which obtained on 2 January 2011:

43. [457]

    The Manual thus explicitly relaxed the requirement to empty stored flood waters within seven days, so long as the height was less than 50cm above FSL. The Manual contemplated that there are environmental benefits in water not being released through the floodgates, including dealing with stranded lung fish, and that in circumstances where the water level is within 50cm of FSL, the flood engineers had a discretion to delay releases. There was a favourable weather outlook on 2 January 2011; less than 5mm of rainfall was forecast for the 24 hours ending 09:00 Monday, and the 4-day PME forecast 2-10mm.

44. [458]

    Those considerations made it not unreasonable for a flood engineer to form the view that the dam had, by the morning of 2 January 2011, returned to FSL, although in fact it was at 67.1m, rather than 67.0m.

45. [459]

    But let it be assumed, favourably to Rodriguez, that the flood event should not have come to an end at 09:00 on 2 January 2011. Dr Christensen was of the view that rather than lowering the gates from midnight on 2 January 2011, they should have been slightly increased to 30 increments, with the result that 470m3/s was to be released from 02:00 that morning for the next 30 hours. Had that occurred, simulation C predicted that by 16:00 on 2 January 2011, the level would have reached 67.05m, with 67.02m being reached at 23:00 on 2 January 2011, and 67.0m being reached at 02:00 on 3 January 2011. Dr Christensen would then have reduced dam levels below FSL. But on the afternoon of 2 January 2011, with levels at 67.05m and falling, there is no reason why it would not have been reasonable to commence closing the gates with a view to their being completely shut over the next 6 hours with the dam levels being only a few centimetres above FSL. Certainly such conduct was not so unreasonable that no competent flood engineer could have adopted it.

46. [460]

    Thus there was no breach of duty in bringing the December Flood Event to an end at 09:00 on 2 January and closing the radial gates.

47. [461]

    It is not clear that the breaches on 2 January 2011 found by the primary judge were to be treated independently. In other words, it is not clear that if the engineers were not negligent in terminating the December Flood Event, they were nevertheless negligent in not making further releases. The purpose of keeping the December Flood Event open was to make further releases; there was no express finding that they were entitled to continue to release water below FSL except during a flood event. The Manual made no express provision for releases except during a flood event; the powers to release water were conditioned by the phrase “during a flood event”: Manual, sections 2.2, 2.3, 2.4, 5.2, 6.2, 8.2, 8.3 (first sentence). The better view may be that the finding of breach of a duty to make further releases fell away if the event were terminated without breach. That would mean the failure to make further substantial releases on 2 January was a consequence of the wrongful termination and, if it caused loss, the loss was a consequence of the breach. There was no separate breach.

48. [462]

    Against the possibilities that (i) there was power to continue to release water although the December Flood Event had been terminated without breach, and (ii) the primary judge had found an additional independent breach, the circumstances facing the engineers on 2 January will be explored further.

49. [463]

    The information available to the flood engineers on 2 January 2011 was that water was still flowing into the dam at rates which were around triple the 50m3/s being released by the regulators, and that the catchment was wet (“saturated”) and might be expected to produce greater run-off from further rain than would otherwise be the case. However, weather conditions for the next few days were relatively benign compared to the flooding of late December. The 1-day QPF was for less than 5-10mm of rain. The 4-day PME was for 1-10mm of rain. Even the 8-day PME was for 15-25mm of rain.

50. [464]

    It is one thing to make releases which would allow dam levels to go below 67m against the certainty of future inflows from rain on the ground, or the immediate prospect of heavy rain, even if that rain has not fallen and might never fall in the catchment. But we do not agree that a flood engineer acted beyond the range of reasonable discretion in not making further substantial releases of Brisbane’s drinking water when the outlook was as favourable as it was on 2 January 2011.

51. [465]

    The primary judge accepted Rodriguez’ submission that an additional 53,000Ml was “potentially material to creating sufficient storage space to address the contingency of a second flood”. We do not agree.

52. [466]

    The flood capacity of Wivenhoe Dam was stated in the Manual to be 1,980,000Ml (at a depth of 80m). Even if one puts that to one side and identifies the flood capacity of 1,160,000Ml at a depth of 75.5m, that is vastly more than the 53,000Ml which was regarded as material. 53,000Ml is less than 3% of the total flood capacity, and less than 5% of the effective flood capacity (after which one or more fuse plugs would be eroded). In many areas of measurement and in many areas of statistics, variations of less than 5% are not regarded as significant.

53. [467]

    Another way of exposing the limited significance of the additional 53,000Ml of capacity is to note that it reflected a saving of around 30cm by the time dam levels reached 74m or 75m.

54. [468]

    The question is whether the failure to release 53,000Ml on 2 January 2011 was so unreasonable that no dam operator in Seqwater's position could properly consider the failure to be reasonable. There is no substance in the contention that there was an urgent need to do so; the issue was whether possible rainfall in the order of a week in the future might give rise to a risk of dam levels exceeding 74m. In those circumstances, it is necessary to ask a question which appears not to have been addressed at trial, namely, how readily might 53,000Ml be released if it were necessary to do so? That question is not difficult to answer. A release of 3,000m3/s equates to 10,800 Ml/hr. Thus the 53,000Ml of additional capacity could be achieved by releases at 3,000m3/s for fewer than five hours; a rate of 2,500m3/s would take a little under six hours to release 53,000Ml. (In the real world it would take time to achieve 25 increments of radial gate opening. Further, if downstream rates at Lowood and Moggill were high, or if it were necessary to protect bridges, then a lower release rate might be required.)

55. [469]

    The point of the example is that the 53,000Ml of additional capacity on which the finding of the primary judge turned was simply not material in the scheme of things. On the weather predictions, the flood engineers had days ahead of them to achieve the additional water capacity, held to have been what a reasonable flood engineer would have put in place on 2 January 2011.

56. [470]

    The foregoing adopts the upper end of the finding that 15,000-53,000Ml could be significant. As 15,000Ml is less than a third of 53,000Ml, it is correspondingly less significant; it is released in fewer than 90 minutes at a rate of 3,000m3/s; is less than 1% of the flood storage capacity and corresponds to less than 10cm of depth when the dam is at 74m.

57. [471]

    None of the above is to deprecate the significance of 53,000Ml, or even 15,000Ml, at the height of a flood. That additional capacity could, as the primary judge noted, be highly significant. It could make the difference between the fuse plug being eroded and not being eroded. But the fact that 53,000Ml (or 15,000Ml) would have significance several days later could only justify releasing that amount of water on 2 January if it were difficult or impossible to do so in the days that followed, when the need to do so became apparent.

58. [472]

    We do not accept that it was unreasonable for a flood engineer to have failed to make substantial releases on 2 January 2011 so as to achieve an additional 53,000Ml of flood storage capacity against the possibility a week or more in the future that 53,000Ml might prove material. It follows that it could not be said that there was a breach of the lower standard prescribed by s 36(2) of the Civil Liability Act.

59. [473]

    There was no breach of duty by Mr Malone on 2 January 2011. Ground 22 is made out.

60. [474]

    On the basis that there was no breach of duty on 2 January, the pleaded breaches thereafter of failing to release water substantially in accordance with simulation C are not engaged. However, the point of principle was that releases had to be sufficient to reduce the dam level and therefore exceed the actual or predicted inflows over those days.

61. [475]

    These breaches cannot be sustained. First, the Manual only permitted opening the radial gates if a new flood event were declared: because, until the afternoon of 5 January, the water level was below 67.25m, these breaches depended upon the first aspect of the breach found on 2 January, namely that it was negligent to terminate the December flood event. Secondly, the same considerations identified above stand in the way of a conclusion that a reasonable flood engineer should have caused substantial releases, below FSL on those days, in the face of predicted rain and the prohibition on lowering the dam level below FSL, discussed above.

62. [476]

    There was nothing unreasonable in declining to declare a flood event where there was no expectation of opening a gate, and indeed no ability to do so because of the prohibition on opening gates unless 67.25m is exceeded. Indeed, it may well be unreasonable to declare a flood event, which entails mobilising the resources of a flood operations centre in Brisbane, until that level has been exceeded, or such a rise is imminent.

63. [477]

    It was not unreasonable for the flood engineers not to expect that further rains would cause the level to rise above FSL to an appreciable extent. Although not articulated as involving a degree of tolerance, the primary judge did address Mr Malone’s evidence that an engineer had a discretion not to declare a flood event unless he expected the water level of one of the dams to exceed FSL and to engage the power to make releases by opening the gates. The judge dismissed that view as “completely untenable”: Ch 3 [144].

64. [478]

    The evidence of Mr Malone was more nuanced than was acknowledged by this finding. It was not, as the primary judge described it, a question as to whether or not there was a “discretion” to declare a flood event; the evidence was rather that there was an element of professional judgment involved. The cross-examination of Mr Malone commenced with the proposition that a “flood event” as defined in the Manual “is a situation”. [142] The cross-examiner continued:

65. [479]

    Section 8.3 has been set out above: it contains two propositions, one commencing “[o]nce a Flood Event is declared”, followed by three bullet points, the second sentence having no such limitation in the opening words. Rather, the prohibition against opening the gates is concerned with the purpose of “flood control”. Accordingly, if the expectation held by the flood engineer was merely that Wivenhoe would rise to 67.2m, the sole purpose of declaring a flood event would be to open the flood operations centre and monitor events, in circumstances where there may be no expectation of any rainfall beyond that which was expected to give rise to a level of 67.2m. Without any releases except those small amounts through the open regulators, in fact Wivenhoe did not exceed 67.25m for the whole of 2, 3 and 4 January.

66. [480]

    It may also be noted that there was reference in the course of the cross-examination to the third bullet point in section 2.2 which, with the chapeau, read as follows:

67. [481]

    This is at best awkward drafting; what Seqwater was required to ensure was that an engineer was on call at all times during a flood event. What followed from that proposition is obscure, if it be assumed that a flood event has already been declared. Given the lack of precision in the language adopted by the Manual, its potential incoherence, and what must undoubtedly be allowed by way of levels of tolerance in respect of measurements and professional decision-making, a flood engineer must have a significant margin for error before it would be appropriate to conclude that he had acted unreasonably.

68. [482]

    There was a degree of incoherence in identifying a breach of duty where the dam level fell to 67.1m when the gates were shut, but not in permitting releases below 67.0m in the absence of any clear basis in the Manual for taking such a step. It is sufficient to say that on a strict interpretation of the Manual, no engineer could be found to act unreasonably in refusing to release water when the level of the dam was below 68.25m and a flood event had not been declared.

69. [483]

    There was in fact no case run by the plaintiff that, if the December Flood Event was properly terminated on 2 January 2011, a further flood event should have been declared over the next three days. There were, no doubt, two reasons for adopting that course. First, until the water level in Wivenhoe exceeded 67.25m, there would have been no loss caused by the failure to take that step. Secondly, a further flood event was declared on the morning of 6 January 2011: it was the steps which were not taken promptly at that time which constituted the alternative basis of Rodriguez’ case, based on simulations F and H, which commenced at 00:00 on 8 January.

70. [484]

    With respect to the breaches of duty found to have occurred on 3, 4 and 5 January, the artificiality of pleading separate breaches on consecutive days has already been noted. If a process of making releases below FSL in anticipation of further rain should have been adopted, that conduct would have provided consequential loss, but was dependent on the finding with respect to the termination of the December Flood Event.

71. [485]

    There was no breach of duty by Mr Malone on 3, 4 or 5 January by failing to make releases or to declare a further flood event. Ground 23(a) is upheld.

20 Breaches on 6-10 January – (grounds 23(b), (c), 24, 25(a), (b))

1. [486]

   In oral argument Rodriguez accepted that although it had “pleaded separate breaches on separate days” against each of the engineers, the dam operations were a “continuum” and based on a consensus between the engineers so that it was necessary to address the decision making process on that basis. [143] The question which arose during the period from 6 to 9 January was whether the strategy formulated by Mr Ayre and Mr Malone (that is, delaying releases until the peak from Lockyer Creek had passed and then until the combined flows below Lowood inundated Mt Crosby Weir and Fernvale Bridges) could not be sustained.

2. [487]

   Rodriguez’ case was that at some stage on 7 or 8 January the point was reached where the dam level was undoubtedly going to exceed 68.5m, engaging strategy W2 or W3 and requiring attention to the primary objective of protecting urban areas from inundation. In fact, no releases were made before 15:00 on 7 January and the objective of keeping those bridges open was not abandoned until the evening of 9 January. In the period from 8 January (the subject of simulations F and H) Rodriguez contended that the engineers should have made releases at levels which, when combined with the natural downstream flows, did not exceed 4000m3/s at Moggill, thereby creating storage capacity. On 11 January it was accepted that the object of the releases was to keep the combined flow at Moggill as low as possible (given that it would inevitably rise above 4000m3/s).

3. [488]

   Grounds 23(b) and (c) challenged findings of breach of duty on the part of Mr Malone in his shifts on 6 and 7 January respectively. These findings turned on a failure to implement and maintain strategy W3 throughout those two days. The primary judge found that releases in the order of 1200 to 1400m3/s should have been made on 6 January and that, on 7 January, releases sufficient to inundate Mt Crosby Weir and Fernvale Bridges were required, being releases above 2000m3/s depending upon the predicted natural flow rate at Lowood.

4. [489]

   Grounds 24, 25(a) and 25 (b) challenged findings of breach of duty on 8, 9 and 10 January respectively on the part of Mr Malone and Mr Tibaldi during their day or overnight shifts during that period.

5. [490]

   It is also necessary to address Rodriguez’ contention 3, and in particular pars 48(b)-(f) of its written submissions. Paragraph 48(b) dealt with 6 and 7 January; 48(c) addressed 7, 8 and 9 January (and perhaps also 6 January); 48(d) addressed releases on 10 January, while 48(e) and 48(f) were directed to the use of forecasts throughout that period. That contention is addressed in part 21 below.

6. [491]

   The flood engineers on duty over this period were:

7. [492]

   Because what ultimately matters is what the flood engineers did and omitted to do, it is convenient to set out the actual decisions to open or close the five radial gates over this period. What follows is taken from the tables in the January 2011 Flood Event Report at pp 154-159.

8. [493]

   The relevant directives for Wivenhoe Dam (up to 15:00 on 10 January) which resulted in those release activities were issued as follows: [144]

9. [494]

   Despite there being only one flood engineer on duty until the evening of 9 January, the Senior Flood Engineer (Mr Ayre) was in charge. The Manual designated him to be in charge, and his evidence (which appears to have been accepted by the primary judge) was that he was “at all times ... kept in the loop in relation to strategies and release rates”: Ch 6 [263]. Typically, gates were opened at the rate of one increment per hour, and it was obviously desirable to avoid sharp increases in release rates. The Manual provided:

10. [495]

    Finally, the releases made at Wivenhoe had to take account of conditions at Lowood (two hours downstream) and Moggill (16 hours downstream). The natural flow rate at Lowood included flows from Lockyer Creek, and at Moggill included flows from the Bremer River. In the case of each of those tributaries the flow times from the upper reaches of their catchments to their confluences with the Brisbane River were up to 15 or 16 hours. Each of these time frames had to be taken into account in assessing future downstream natural flow rates when making releases from Wivenhoe.

11. [496]

    What follows addresses only the findings of breach by Mr Malone or Mr Tibaldi. In the case of each shift the findings of breach are outlined and the relevant ground then addressed.

12. [497]

    Mr Malone’s shift at the Flood Operations Centre (FOC) on Thursday 6 January ran from about 07:00 to 18:30, when Mr Ayre started his overnight shift. At 07:42 Mr Malone emailed flood operations personnel, including the other flood engineers, mobilising staff for gate operations. Mr Malone’s reason for doing so was “Wednesday night’s rainfall and further totals up to 150mm expected during the next two days”. In the 24 hours to 09:00 on 6 January the average rainfall in the dam catchments was 19mm (Somerset) and 26mm (Wivenhoe). [153] The forecasts referred to were the 1200UTC PMEs available at about 06:00: Ch 6 [155]. In his situation report issued just after 08:00, Mr Malone again referred to those forecasts “for the next 24 to 48 hours being for totals of up to 150mm in SE Qld”. The reference to “SE Qld” was a reference to a “vast area” above and below the dams which included, but was not limited to, the dam catchments and downstream Lockyer Creek and Bremer River catchments.

13. [498]

    The morning and afternoon QPFs and 1-day PMEs predicted rainfall in the catchment areas above and below Wivenhoe Dam: Rodriguez (22) Ch 6. The morning QPF was for 30-50mm in the Somerset and Wivenhoe catchments for the 24-hour period from 09:00: Ch 7 [151], [152].

14. [499]

    At 10:27 Mr Malone emailed Messrs Ayre, Ruffini and Tibaldi to update them on his operational strategy. The email set out by the judge at Ch 6 [159] read:

15. [500]

    At 06:00 the dam level was 67.31m; by midday it had risen to 67.34m, with inflows averaging around 250m3/s and outflows of 50m3/s via the regulators: Ch 6 [153]. Mr Malone’s first ROG model run at 12:00 predicted a natural peak flow rate at Lowood of 447m3/s early in the morning of 7 January which would have submerged Burtons Bridge without further rain. That run continued releases of 50m3/s and predicted a dam level of 68.2m sometime on 11 January: Ch 6 [164]. At 12:14 on 6 January Mr Malone emailed the flood engineers advising that from 18:00 the gates would be opened progressively by five increments of 0.5m until 22:00 (about 250m3/s) with the aim of keeping Burtons Bridge open (its submergence flow rate being 450m3/s): Ch 6 [165]. Further openings would depend on forecast rain and Lockyer flows.

16. [501]

    His 13:30 email revised that strategy, advising the engineers and others that “further heavy falls in the Lockyer” since 09:00 meant that the flow of water from Lockyer Creek could be as high as 600m3/s peaking on 8 January with the result that the gate openings would be delayed until the “Lockyer peak passes”: Ch 6 [165]. In his afternoon situation report issued at 14:54, Mr Malone reported “isolated heavy falls [of] up to 60mm in the Somerset and Wivenhoe catchments” with forecast totals in the next 24 to 48 hours of up to 100mm to fall in “SE Qld”. He repeated his expectation that a peak flow rate of 600m3/s was expected “from the Lockyer late Friday”. Allowing for the “flood levels in the lower Lockyer” to subside, he foreshadowed releases from Wivenhoe during 8 January “as high as 1500m3/s” and continuing “for a couple of days”: Ch 6 [167].

17. [502]

    During the balance of his day shift, Mr Malone conducted further ROG modelling, two at 15:00 and one at 16:00: Ch 6 [168], [171]. Those at 15:00 did not assume any radial gate releases from Wivenhoe. The 16:00 model included releases starting at 16:00 on 7 January. The earlier model runs predicted progressively higher peak flow rates at Lowood (Lockyer Creek) and Moggill, the result of higher ROG inputs over time. The likely later of the two 15:00 models showed Wivenhoe Dam peaking at 68.51m at 11:00 on 11 January, on the basis of no releases. The 16:00 model which provided for releases from 16:00 on 7 January predicted a maximum dam level of 68.41m at 16:00 on 8 January.

18. [503]

    The afternoon QPF was for 20-30mm of rain in the dam catchments in the 24 hours from 15:00. In his third situation report issued at 17:33, Mr Malone recorded that Wivenhoe was at 67.39m and “rising slowly”: Ch 6 [173]. He confirmed that the Wivenhoe gates would be opened “after the impact of Lockyer flows on Burtons Bridge had been ascertained and flood levels in the lower Lockyer subside”. It was “not expected” that there would be any adverse impacts upon Fernvale Bridge or Mt Crosby Weir Bridge. This report noted that the rain forecast “in SE Qld” for the next 24 to 48 hours was for “totals up to 100mm”: Ch 6 [173].

19. [504]

    The rate of inflows to Wivenhoe between 07:00 and 19:00 ranged from 152m3/s to 559m3/s, and the level of the dam rose from 67.31m to 67.41m, an increase in the storage volume of 10,985Ml which could have been discharged in about three hours at a rate of 1000m3/s. During the same period the level of Somerset rose from 99.34m to 99.46m.

20. [505]

    The primary judge concluded in Ch 12:

21. [506]

    In so holding his Honour accepted Rodriguez’ submission that strategy W3 was “required” and that if the flood engineers were operating in accordance with that strategy they would have commenced releases immediately and at significant rates: Ch 12 [116]. Rodriguez submitted that the making of such releases followed from that “correct selection of strategy”. The primary judge quoted and accepted that submission: Ch 12 [111], [116]. There remained the large question as to whether a reasonably competent flood engineer operating in strategy W3 in the circumstances facing Mr Malone on the morning of 6 January “must” have made releases that “substantially exceeded the rate of inflows” and in the order of 1200m3/s to 1400m3/s.

22. [507]

    Rodriguez submitted that if strategy W3 was engaged releases could not be delayed and “should have been made at rates substantially exceeding rates of inflow to create storage”. As to why such releases should have been made, Rodriguez contended: [154]

23. [508]

    That submission did not contend that a reasonably competent flood engineer must have made releases between 1200m3/s and 1400m3/s; rather it relied on “admissions” as justifying the making of significant releases to reduce the water levels in the dams. The “admission” attributed to Mr Malone was that “if it was permissible to make releases on the rising limb of the hydrograph and to make releases below FSL, then the afternoon of 6 January 2011 would have been a ‘good time’ to do so”, without submerging Mt Crosby Weir or Fernvale Bridges.

24. [509]

    In his analysis of simulation C – the declared or target strategy being to make releases from Wivenhoe to a level 2.5m below FSL [155] – the primary judge found that by 00:00 on 8 January “a reasonably competent flood engineer conducting flood operations in SIM C but using 4-day PMEs to determine strategy and releases” would have been obliged to close Fernvale Bridge and Mt Crosby Weir Bridge by making releases of around 1400m3/s: Ch 10 [154]. The primary judge described that finding as “consistent with” acceptance of Rodriguez’ submission at par 1167: Ch 12 [117].

25. [510]

    The primary judge concluded that given “the identified risk” the reasonably competent flood engineer “would” have made releases between 1200 and 1400m3/s. That “risk” was that if releases did not commence immediately and at the rates contended for there would be insufficient flood storage capacity in Wivenhoe “to store incoming flows should further rainfall occur in accordance with, or in excess of, that forecast” and that without such capacity subsequent releases would be necessary in volumes that would cause urban flooding downstream. [156] The primary judge found that the “effect of the forecasts was that the period of time in which to make releases without the potential to cause damage when combined with downstream flows was shortening”: Ch 12 [118].

26. [511]

    The forecasts were the 1200UTC forecasts available at 06:00 on 6 January and the 00UTC forecasts available at 18:00 on 6 January. The former were for the four-day period commencing at 22:00 on the previous evening and the latter for the four-day period commencing at 22:00 on 6 January. [157] Those forecasts predicted that the highest rainfall would occur on 9, 10 and 11 January. Mr Malone’s evidence was that, having regard to the 4-day PMEs available to him, “the uncertainty over the location of where rain might fall warranted consideration of the realistic possibility that the forecast rain would fall upstream instead of, or together with, rainfall downstream”: Ch 6 [157].

27. [512]

    Mr Malone accepted that this was a “clear possibility” that he “needed” to address. The judge summarised Mr Malone’s assessment of the likelihood of that possibility coming to pass (in Ch 6) as follows:

28. [513]

    By ground 23(b) Seqwater challenged the finding that Mr Malone breached his duty of care on 6 January by “failing to continue or commence flood operations and make releases, failing to implement and maintain strategy W3, and failing to make releases at rates which substantially exceeded inflows, specifically a rate at least above that necessary to inundate Kholo Bridge in the order of 1200 to 1400m3/s”.

29. [514]

    Seqwater’s principal submission was that the judge’s conclusion that strategy W3 should have been implemented was affected by errors in construing the Manual, with the result that the finding that releases in the order of 1200 to 1400m3/s should have been made was not supported by the evidence.

30. [515]

    In oral argument, the errors in the construction of the Manual were identified as the adoption of the “no release” assumption in selecting the flood operations strategy, as well as the making of releases below FSL and use of the 4-day PME forecasts in determining the strategy. [158] Although not referred to, Mr Malone’s evidence also raised a question as to what the constraint in section 8.4 of the Manual that “peak outflow should generally not exceed peak inflow” conveyed and permitted.

31. [516]

    In response, Rodriguez maintained the correctness of the primary judge’s adoption of the “no release” assumption and the use of 4-day PME forecasts or 1-day QPFs in determining strategy and gate releases.

32. [517]

    The condition engaging strategies W2 and W3 was that the predicted level of Wivenhoe exceed 68.5m, but not 74m. If that condition were satisfied the strategy flowchart directed attention to the maximum (combined) flows at Lowood and Moggill. If those combined flows were likely to be less than 3500m3/s and 4000m3/s respectively, strategy W2 was engaged. Adopting the “no release” assumption and using 24 hour or 4-day PME forecasts, the primary judge found that condition would have been satisfied. That much was conceded by Mr Malone: Ch 6 [194]. However, as accepted above, the Manual did not require adoption of that assumption in selecting a strategy. It was appropriate for any prediction of the dam level for that purpose to take into account the releases which a reasonably competent flood engineer would consider could be made over the period to which the prediction related. Except in one respect, the prediction or predictions that might have been made on that basis are not the subject of any findings.

33. [518]

    The exception relates to his Honour’s finding that “with forecast” modelling of 24-hour rainfall would have yielded a predicted dam height greater than 68.5m “if conducted on either a no-release basis or [with] the gate releases suggested by” Mr Ayre’s model run at 21:00 on 6 January: Ch 6 [194]. That finding was based on a model run undertaken by Mr Malone, after the event, using data available at 18:00 on 7 January and otherwise making the assumptions in Mr Ayre’s 21:00 model. Those releases differed from the releases adopted by Mr Malone in his ROG model run at 16:00.

34. [519]

    However, that evidence did not justify a finding that a reasonably competent flood engineer determining which strategy to apply and taking account of the QPF forecasts and 4-day PMEs must reasonably have concluded that the level of Wivenhoe was likely to exceed 68.5m, so as to engage strategy W2 or W3. In the absence of any contention that a reasonably competent flood engineer could only have predicted such a dam level (taking into account Mr Ayre’s modelled releases), Seqwater’s challenge to the finding that Mr Malone was “required” to act in accordance with strategy W3 is made out. It also follows that the challenge to the finding of breach in not making the releases contended for is also made out. If the predicted level of the dam had exceeded 68.5m, the question remained whether strategy W2 or W3 was to be applied. The Manual’s strategy flowchart suggested that strategy W2 was engaged if the combined maximum flow at Lowood was likely to be less than 3500m3/s and at Moggill less than 4000m3/s. Dr Christensen’s methodology suggested that in determining which strategy to apply, at least each day (if not more frequently), the engineer should compute a flow hydrograph for Lowood and Moggill based on rain on the ground (best forecast stream flow) and 24-hour forecast (best forecast rainfall for this purpose).

35. [520]

    Assuming that strategy W3, rather than W2, was engaged, the remaining question was whether a reasonably competent flood engineer in Mr Malone’s position “must” have released water at rates above 1200m3/s. In addressing that question, the primary judge referred to his conclusion in the very different circumstances prevailing in simulation C.

36. [521]

    The strategy in that counterfactual was to create flood storage space in Wivenhoe by reducing the dam level to 2.5m below FSL. Dr Christensen did not consider he was constrained from doing so or from making releases in excess of the peak inflow to the point in time when the releases were being made. It is not necessary to pursue this aspect of his Honour’s finding because it only has relevance if strategy W3 were required to be implemented.

37. [522]

    At 07:00 on Friday 7 January Wivenhoe Dam was at 67.68m, having increased from 67.41m at 19:00 on the previous evening. During that period total flow rates into Wivenhoe, including from Somerset, were as low as 355m3/s at 19:00 and as high as 1342m3/s at 05:00. No releases were made (other than those from the regulator and hydro). The flood engineers eventually commenced releases at 15:00 on 7 January, those releases increasing to 1239m3/s by 14:00 on 8 January.

38. [523]

    At about 06:00 on 7 January the five 1-day 1200UTC PME forecasts for the period commencing 22:00 on 6 January became available. Mr Malone commenced his shift at 06:45, replacing Mr Ayre whose shift finished at 07:00. Mr Malone was later replaced by Mr Ruffini, whose first shift commenced at 18:45. The ROG model run by Mr Ayre at 21:00 the previous evening had assumed gate operations starting at 23:00 on 7 January, increasing to a peak release rate of about 1225m3/s (gates opened to 24 increments) at 22:00 on 8 January. Mr Ayre’s 06:06 situation report noted that there had been “significant” rain in the Lockyer Creek catchment and that the forecast for the next five days was for “totals between 100 and 200mm in SE Qld”, which, given the “saturated condition of the catchments”, was most likely to produce further runoff: Ch 6 [219]. During the handover from Mr Ayre, it was decided that Mr Malone would proceed with releases from Wivenhoe “after the peak had passed from Lockyer Creek”: Ch 6 [226].

39. [524]

    At 08:00 the Bureau of Meteorology advised that “SE Qld” could “expect some high rainfall totals over the next five days up to Tuesday with the largest predicted” for 9 and 10 January (being Sunday and Monday): Ch 6 [226]. An earlier email from the Bureau had referred to some “isolated major flooding in the Bremer River and Lockyer”. Mr Malone’s email sent to Seqwater staff at 08:04 was headed “Operating strategy over the next week” and provided:

40. [525]

    Mr Malone concluded in relation to Wivenhoe that releases would be “ramp[ed] up” to about 1200m3/s later that day, noting that “given the high likelihood of significant inflows in the next week, this may be increased to 1500m3/s in order to drain the current temporarily stored flood waters as soon as possible”. The result was said to be that all of the crossings downstream of Wivenhoe with the exception of Fernvale Bridge (submergence flow rate 2000m3/s) and Mt Crosby Weir Bridge (submergence flow rate 1900m3/s) would be adversely impacted. However at 11:13 Mr Malone advised that releases would commence at 15:00 “and be slowly increased to about 1200m3/s” by 14:00 on 8 January, and issued a directive for the opening of the radial gates by seven increments between 15:00 and 21:00 (estimated flow rate 400m3/s): Ch 6 [228].

41. [526]

    By midday on 7 January the dam level was 67.88m. By 19:00 that level was 68.17m, the rates of inflow rising as high as 2225m3/s at 11:00 and fluctuating below that level and as low as 958m3/s at 19:00. Mr Malone ran at least two further ROG models, the first at midday and the second at 18:00. The gate operations adopted in the first were the same as those in Mr Ayre’s 21:00 run on 6 January, except that they started eight hours earlier at 15:00 on 7 January. This model predicted a maximum dam level of 68.32m and a combined maximum flow rate at Moggill of 1615m3/s at 14:00 on 8 January. The second ROG model run, at 18:00, extended the 24-increment release period from midday on 9 January to 00:00 on 11 January, a 36-hour period in which the PMEs forecast heavy rainfall: Ch 6 [238]. The predicted maximum level of Wivenhoe was 68.51m at 14:00 on 8 January.

42. [527]

    At 18:00 on 7 January the level of Somerset was 100.11m. The inflow rate to Somerset (excluding outflows to Wivenhoe) since 07:00 had varied between 169m3/s and 879m3/s. At that time Mr Malone directed that the regulator be closed and one sluice gate opened. As a result the outflow from Somerset to Wivenhoe increased from 34m3/s to around 206m3/s.

43. [528]

    Mr Malone’s situation report, also issued around 18:00, reported widespread rain since 09:00 throughout the Somerset and Wivenhoe catchments (20-40mm). Referring to the five 1-day PME forecasts received in the morning, he again noted that “significant inflows to Seqwater dams will be generated”, especially following the forecast rainfall on 9 and 10 January: Ch 6 [233]. In cross-examination Mr Malone accepted that if the rain forecast for those days (100mm, the question in cross-examination referring to that rainfall depth but not identifying where that rain might fall) eventuated “it might not be possible for Wivenhoe Dam to make releases on those days without combined flows at Moggill exceeding 4000m3/s”: Ch 6 [230].

44. [529]

    At around 18:00 the 1-day 00UTC PMEs for the five days commencing 22:00 on 7 January became available. The four-day estimates issued at midnight and based on those PMEs were extracted by the primary judge at Table 9-2 which set out the forecast rainfall estimates of Rodriguez (Dr Christensen) and the State of Queensland: see [312] above. As to the rain forecast to fall on 8, 9 and 10 January, the primary judge recorded the different predictions proposed by Rodriguez and the State:

45. [530]

    The primary judge did not find that these PME forecasts, only available at around 18:00, should have been considered and acted upon by Mr Malone before the end of his shift at 19:00: Ch 6 [240].

46. [531]

    Mr Ruffini issued directives during his overnight shift which progressively resulted in the gates being opened to 18 increments by 08:00 on 8 January (927m3/s). At 22:00 he ran a forecast rainfall (72-hour time period) model which used the same gate operations as Mr Malone’s 18:00 ROG run. That model predicted a dam level of 68.9m at 21:00 on 8 January: Ch 6 [247]. There were issues about the “competency” of this modelling run. The primary judge considered that “on the information available at the time it was said to be conducted” the model “yielded a serious underestimate of the likely inflows into Wivenhoe Dam from existing rainfall and rain forecast to fall over the following 72 hours”: Ch 6 [248]. He also found that by 00:00 on 8 January the likelihood of the storage level of 68.5m being exceeded was “overwhelming” given the height of the dam at that time (68.32m), the rainfall to date and the forecasts which had been received: Ch 6 [257]. However, this finding is of only indirect significance as Seqwater was not held to be responsible for Mr Ruffini’s conduct.

47. [532]

    In Ch 12 the primary judge found that:

48. [533]

    His Honour concluded that at some stage on 7 January the selection of strategy W3 was “required” regardless of whether the relevant prediction was made by reference to the 4-day PMEs or 1-day QPFs: Ch 12 [131]. That was undoubtedly correct if a “no release” calculation were made. At 07:00 on 7 January the actual dam level was 67.68m. The estimated maximum inflow volume based on the 4-day 00UTC PMEs issued at 00:00 on 7 January was between 608,000Ml (Dr Christensen) and 547,000Ml (Mr Giles as corrected). [159] Looking at the position at 07:00 and assuming no releases, that inflow would have resulted in a dam level higher than 71m. A dam level exceeding 68.5m was also predicted (on a “no release” basis) if Mr Pokarier’s corrected 7 January morning QPF estimated inflow volume were used. [160]

49. [534]

    The primary judge also found that the rain forecasts available on 7 January “either did demonstrate, or should have demonstrated, to each of Messrs Ayre, Malone and Ruffini the strong likelihood, bordering on certainty as the day progressed, that the storage level of Wivenhoe Dam” would exceed 68.5m: Ch 6 [256]. That finding was not in terms challenged. Nor could it be in the face of the 18:00 ROG model run by Mr Malone which predicted a maximum dam level of 68.5m. Whilst that run took into account ROG as at 18:00, it did not take account of the afternoon QPF which forecast 20 to 30mm of rain in the dam catchments in the 24 hours to 16:00 on 8 January: Ch 6 [214].

50. [535]

    As to the rate at which water was “required” to be released, his Honour said in Ch 12:

51. [536]

    The finding in Chapter 10 concerning simulation C is in terms a finding that “a reasonably competent flood engineer conducting flood operations in SIM C but using 4-day PMEs to determine strategy and releases … and acting in accordance with the Manual … would have been obliged” to close the remaining two bridges by 00:00 on 8 January 2011: Ch 10 [154] (emphasis in original). The qualifier, “conducting flood operations in SIM C”, should be noted. As at 00:00 on 7 January, the actual dam level was 1.77m higher than in simulation C (equivalent to additional storage volume in the simulation of 184,471Ml). The judge separately (and more relevantly in the present context) held that a reasonably competent flood engineer operating in simulation F would have submerged the two bridges in the early morning of 8 January: Ch 10, [18], [144]. The factor identified as justifying the “substantial step” of inundating those bridges was “the likelihood of forced releases above 74m and downstream flows above 4000m3/s”, which his Honour described as “very real” and as far outweighing “the inconvenience arising from their closure”: Ch 12 [132]. That prospect had two elements. The first was the predicted level of the dam having regard to ROG and forecast rain, which in turn directed attention to inflow rates during the forecast period. The second directed attention to the predicted peak natural flows at Moggill, having regard to ROG and rain forecast in the Lockyer Creek and Bremer River catchments.

52. [537]

    The primary judge considered the predicted level of the dam on the basis of information available to Mr Malone at 18:00 on 7 January, when the level of Wivenhoe was 68.12m: Ch 6 [265]. Mr Malone’s situation report estimated the volume of ROG still to flow into the dam as 160,000Ml. The remaining storage capacity between the actual dam level and 74m was 790,098Ml, which Mr Malone estimated to be equivalent to 140mm excess or runoff rainfall. After allowing for the 160,000Ml ROG, the depth of rainfall necessary to produce the remaining inflow of 630,098Ml was estimated to be between 145mm or 130mm, the former adopting Mr Malone’s runoff ratio of 77%, and the latter adopting Dr Christensen’s runoff ratio of 86%: Ch 6 [230], [265]. Taking the 4-day 00UTC PME forecast available at 00:00 on 8 January, the estimated average rainfall depth was 200mm taking Dr Christensen’s range, and 175mm taking the State of Queensland’s range: Table 9-2 at [312] above.

53. [538]

    The primary judge held that “given the forecasts and the saturated catchment, the prospect of 140mm excess or runoff rainfall was very likely. In those circumstances there subsisted a serious and significant risk that if sufficient releases were not made at that time, releases would have to be made from above 74m later” (emphasis added): Ch 6 [266]. The finding as to the “likelihood … [of] downstream flows above 4000m3/s” [161] appears to have been based, at least in part, on Mr Malone’s acceptance in cross-examination that if the forecasts for 9 and 10 January “came to pass, it might not be possible for Wivenhoe dam to make releases on those days without combined flows at Moggill exceeding 4000m3/s”.

54. [539]

    Thus the judge’s conclusion that, on the evening of 7 January, there was a “very real” likelihood of forced releases above 74m and downstream flows above 4000m3/s [162] depended on a number of possibilities coming to pass. They included the average of the forecast rain falling in the dam catchments over the four days, sufficient releases not being made from Wivenhoe in the three or four day period of that prediction, [163] and heavy rain in the downstream Lockyer Creek and Bremer River catchments resulting in significant natural flow rates at Lowood and Moggill.

55. [540]

    From Mr Malone’s perspective the forecasts on which the primary judge’s analysis was based did not become available until 18:00. The 1200UTC PME forecasts which became available at 06:00 had been considered in Mr Malone’s email of 08:04. The average forecast rainfall over the four-days was 100mm (Christensen) and 87.5mm (the State of Queensland), as set out in Table 9-2 at [312] above. Mr Malone’s assessment on the morning of 7 January was that the coloured contour maps available on the BoM website suggested the highest falls were likely to be “very coastal” and downstream of Wivenhoe Dam: Ch 6 [221], [222], [226].

56. [541]

    His Honour’s findings in relation to this shift are challenged by ground 23(c). Seqwater made three submissions. First, it said that the primary judge’s finding that strategy W3 was required to be implemented was based on a “no release” prediction as to the dam level. Secondly, it said that the finding that the flood engineers had to make releases which submerged the remaining two bridges presupposed that strategy W3 was engaged. Thirdly, it said that the evidence did not support the primary judge’s conclusion as to the “likelihood” of forced releases above 74m and combined downstream flows above 4000m3/s, so as to justify the submerging of those bridges. This last submission challenged the judge’s reliance on the 4-day PME forecasts which became available at 18:00 on 7 January (at the end of Mr Malone’s shift). The relevant forecasts available to Mr Malone were those issued at 00:00 on 7 January, the averages of which were not sufficient to produce the further runoff rainfall required to cause the dam level to exceed 74m: see Table 9-2.

57. [542]

    In response Rodriguez maintained that the selection of strategy was to be undertaken on a “no release” basis and supported the primary judge’s findings as to the predicted level of the dam and rate of downstream flows by reference to the 4-day PMEs that became available at 18:00. That the upper range of those forecasts predicted twice as much rain as would have been required to take the dam above 74m on a “no release” basis, was said to support the conclusion that releases would not have prevented the dam from rising above 74m.

58. [543]

    As to Seqwater’s first submission, the primary judge’s finding that strategy W3 was “required” on 7 January proceeded from the premise that the position on 7 January was “no different from that stated” for 6 January: Ch 12 [121]. For the reasons given in relation to 6 January, a finding that strategy W3 was engaged assumed “no releases” were made and was not required by the Manual.

59. [544]

    However, his Honour’s finding that “as the day progressed” the engineers should have chosen strategy W3 did not depend solely on a predicted dam level based on a “no release” assumption. Mr Malone’s ROG model run at 18:00 predicted a level of Wivenhoe of 68.51m at 14:00 on 8 January. On the basis of that modelling and taking into account the five 1-day 00UTC PMEs available at 18:00, the evidence established by this time that the level of Wivenhoe was likely to exceed 68.5m, engaging strategy W2 or W3. However, those forecasts did not become available to Mr Malone before 18:00, or in sufficient time for him to undertake the analysis necessary to justify and require that decision.

60. [545]

    It follows that whilst the evidence justified a finding that by the end of 7 January strategy W2 or W3 was engaged, it did not support the conclusion that a reasonably competent flood engineer in Mr Malone’s position “must” have determined by the close of his shift that strategy W2 or W3 was engaged. It follows that the challenge to the findings of breach which are the subject of ground 23(c) should be upheld.

61. [546]

    In reaching that conclusion it has been unnecessary to consider Seqwater’s challenge to the primary judge’s finding that by the evening of 7 January there was a “very real” likelihood of forced releases above 74m that would combine with downstream flows to produce flow rates above 4000m3/s, being a likelihood that should have been appreciated by any reasonably competent flood engineer.

62. [547]

    The primary judge found that on the evening of Saturday 8 January, during the first part of his overnight shift, Mr Tibaldi failed to take precautions that a reasonably competent flood engineer would have taken. As pleaded, those precautions were complying with the Manual, implementing and maintaining strategy W4 and causing Wivenhoe to release "at rates exceeding the rate of inflow": Ch 12 [157]. In the early morning part of his shift, Mr Tibaldi was found to be in breach in failing to comply with the Manual and implementing and maintaining strategy W4: Ch 12 [182]. No separate allegation was made that "on the morning of 9 January" he failed to make releases at rates exceeding the rate of inflow. Ground 24 addressed the separate findings made in relation to the evening and early morning parts of Mr Tibaldi's shift. However, before addressing Mr Tibaldi's shift, it is necessary to record, briefly, what happened during Mr Ayre's day shift.

63. [548]

    In the 24 hours to 09:00 on 8 January, the average rainfall in the dam catchments was about 26mm. [164] Mr Ruffini's situation report issued at about 06:30 on 8 January noted that "no significant rain had fallen in the past 12 hours" and that the 4-day PME forecasts (available at 00:00 on 8 January) indicated "that SE Qld can expect further high rainfall totals over the next four days". The estimated rain forecast for the four days commencing 22:00 on 7 January was identical to the 5-day forecast from the previous day (set out at [524] above) except that for 8 January 5-50mm was forecast instead of 15-50mm: Ch 7 [5].

64. [549]

    The report stated that by midday on 8 January it was proposed to "ramp up" releases to 1200m3/s and that this may have to be increased "given the high likelihood of significant inflows in the next week". This strategy was essentially the same as that recorded by Mr Malone in his situation report issued on the previous evening: Ch 6 [234].

65. [550]

    Mr Ayre signed on at 06:30. The level of Wivenhoe at 07:00 was 68.48m and by 08:00 it exceeded 68.50m (albeit by only 2cm). That occurred earlier (by 6 hours) than the maximum height (68.51m) predicted by Mr Malone's second ROG model run at 18:00 on 7 January. At 08:15 Mr Ayre directed that the gate openings be increased to 24 increments by 14:00, with an expected outflow of around 1200m3/s: Ch 7 [15].

66. [551]

    Mr Ayre ran an ROG model at 09:00 which predicted a dam level of 68.64m at 21:00 that evening, assuming releases of 1252m3/s (24 increments) from 14:00 on 8 January which peaked at 1472m3/s (29 increments) at 01:00 on 10 January: Ch 7 [39]. Those releases were expected to result in combined flow rates at Moggill of around 1600m3/s, which would keep open the Mt Crosby Weir Bridge and Fernvale bridges: Ch 7 [39]. The natural peaks at Lowood and Moggill were shown as having already occurred.

67. [552]

    Before 11:35 Mr Ayre directed that a second sluice gate at Somerset be opened. At that time the level of Somerset was 100.45m. As a result the outflow rate from Somerset to Wivenhoe increased from 207m3/s to 413m3/s. Over the period from midday on 8 January to 14:00 on 9 January the level of Somerset fell to 100.27m and then slowly increased to 100.47m. Mr Ayre's directive noted that Somerset was expected to peak at around 100.48m but was "still rising", requiring the implementation of strategy S2: Ch 7 [47]. In fact the inflow to Somerset increased dramatically between 09:00 on 9 January (1027m3/s) and 15:00 on that day (5352m3/s) after which the inflow fluctuated, but not below 2128m3/s until 15:00 on 10 January. At that time the dam level was 103.43m.

68. [553]

    At 12:15 Mr Ayre issued an FOC status report. He noted that no significant rain had fallen in the dam catchments in the previous 18 hours, [165] and repeated the four-day forecasts summarised in Mr Ruffini's earlier situation report, noting that "SE Qld can expect further high rainfall totals over the next four days". His report continued in relation to Wivenhoe:

69. [554]

    In relation to "Impacts downstream of Wivenhoe", Mr Ayre stated:

70. [555]

    At 15:00 Mr Malone emailed Mr Ayre referring to the four-day forecasts for "SE Qld" and noting that the "BoM estimates are now double these estimates for the next few days". No finding was made by the primary judge as to the forecasts to which Mr Malone may have been referring. Mr Ruffini's estimates were based on the 00UTC PME forecasts issued at 00:00 on 8 January and did not understate what they showed: Table 9-2 at [312] above.

71. [556]

    Mr Ayre undertook two further model runs at around 15:00, one using ROG and the other the rainfall depths and loss rates for upstream catchments used by Mr Ruffini in his 72 hour model run at 22:00 on 7 January. As with Mr Ruffini's model, the primary judge concluded that Mr Ayre's modelling yielded a serious underestimate of the likely inflows based on ROG and forecast rain: Ch 7 [64]. The ROG model deferred the commencement of the gate opening sequence from 24 to 29 increments by three hours, thereby extending the maximum discharge (1438m3/s) to 19:00 on 10 January: Ch 7 [66]. That model predicted a peak dam level of 68.66m at 20:00 on 8 January. In doing so it took no account of the forecasts of heavy rain continuing on 9 and 10 January.

72. [557]

    Mr Ayre's evening situation report was issued shortly before 18:00: Ch 7 [68]. The afternoon QPF was for between 30 and 50mm of rain in the dam catchments for the 24 hours from 15:00 on 8 January. The report also referred to the PME forecasts for the next three days – 9 January (50 to 100mm), 10 January (50 to 150mm) and 11 January (25 to 50mm). The level of Wivenhoe was 68.65m “and rising slowly” and the “current gate operation strategy” was to maintain “flows of up to 1600m3/s in the mid-Brisbane River throughout the evening”.

73. [558]

    Mr Ayre repeated the following comments from his earlier status report in his evening report – that river levels upstream of Wivenhoe "have peaked and are now receding"; that it was intended to maintain releases at 1250m3/s whilst keeping flows at no more than 1600m3/s at Moggill; that further assessments were required to determine releases above that level “given the high likelihood of significant inflows in the next few days”; and that this would in turn require an assessment of the runoff from the Bremer River to determine an appropriate release strategy. Finally, he referred to projections based on forecast rainfall which suggested that flows of up to 1200m3/s would “emanate” from the Bremer River catchment; and recorded that even if releases from Wivenhoe “may need to adversely impact” the remaining two bridges, they “will be maintained below 3500m3/s”: Ch 7 [71].

74. [559]

    In fact, in the 48 hours from 09:00 on 9 January to 09:00 on 11 January the average rainfall in the catchments to the dams was 313mm in the Somerset catchment, and 238mm in the Wivenhoe catchment. During the same period the average rainfall in the Lockyer Creek and the Bremer River catchments was 169mm and 117mm respectively. [166]

75. [560]

    Mr Tibaldi had been on leave between 2 and 7 January and had only returned to Brisbane on 8 January. His overnight shift commenced at 19:00 that evening and concluded at 07:00 on 9 January. The findings did not suggest he did any model runs in the evening. He did however review the earlier 72-hour model run undertaken by Mr Ayre.

76. [561]

    In the period from 19:00 until midnight on 8 January, the level of Wivenhoe remained around 68.65m, there being no significant difference between the flow rates into and out of the dam. In the period from 00:00 to 07:00 on 9 January, the dam level dropped from 68.64m to 68.57m. Directives were given to open the gates by two increments with the result that the Wivenhoe outflow rate increased from 1241m3/s (24 increments) at 00:00 to 1334m3/s (26 increments) at 07:00.

77. [562]

    The 4-day 00UTC PME forecasts issued at 00:00 on 9 January estimated rainfall in the catchments above the dams in the range of 75-300mm (Dr Christensen) or 50-300mm (State of Queensland): Ch 9 [138], Table 9-2.

78. [563]

    Mr Tibaldi's situation report issued at 06:15 on 9 January noted that some areas in the Somerset catchment had recorded falls exceeding 60mm “over the last two hours”. The current gate strategy continued to be maintaining “flows of around 1600m3/s in the mid-Brisbane River”: Ch 9 [138], Table 9-2.

79. [564]

    At the end of his shift, Mr Tibaldi undertook an ROG model run which used the same gate openings as Mr Ayre's 8 January 15:00 run, except that it extended the maximum gate opening period (29 increments) to 00:00 on 12 January. That model predicted a dam level of 68.66m at 20:00 on 8 January, a peak release rate of 1493m3/s, and a maximum combined flow rate at Moggill during that peak release period of 1639m3/s: Ch 7 [151]. As in the earlier models, only one natural peak was predicted at Lowood and Moggill and in each case was shown as having occurred on the evening of 7 January.

80. [565]

    In the period from 19:00 on 8 January to 00:00 on 9 January the primary judge found in Ch 12 in relation to Mr Tibaldi:

81. [566]

    Addressing the remaining period from 00:00 to 07:00 on 9 January, the primary judge found in relation to Mr Tibaldi:

82. [567]

    The finding that the flood engineers were obliged to adopt strategy W4 was based on a prediction of the likely level of Wivenhoe made by reference to the 4-day 00UTC PME forecast available at 00:00 on 8 January. As has been noted, the average of that forecast was 200mm (Dr Christensen) and 175mm (the State of Queensland). The projected levels of the dam, assuming no releases during the relevant four-day period, were 75.82m (Dr Christensen) and 75.09m (Mr Giles (corrected)). [167]

83. [568]

    The primary judge considered whether “proper modelling” using forecast rainfall depths would have predicted a dam level exceeding 74m. He first did so by reference to Mr Tibaldi's evidence as to his re-creation of the 72-hour model run of Mr Ayre undertaken at 15:00 on 8 January: Ch 7 [112]-[119]. The judge concluded that if modelling had been done using different forecasts of rain, “it is overwhelmingly likely that either the predicted level of Wivenhoe dam would have exceeded 74m or the strategy of maintaining Fernvale Bridge and Mt Crosby Weir Bridge open would have had to have been abandoned” (emphasis added). That conclusion acknowledged two possible outcomes from the “proper modelling” exercise, one of which was that the reasonably competent flood engineer would (or must) have adopted release rates sufficient to submerge the two bridges. That would have avoided, at least at that point in time, a predicted dam level exceeding 74m. Accordingly his Honour’s observation is not consistent with a conclusion that, taking account of releases, any reasonably competent flood engineer must have predicted that the dam level would exceed 74m: Ch 7 [119].

84. [569]

    The same reasoning explains the judge's qualified finding that Mr Ayre was obliged to select strategy W4 if his prediction as to the level of the dam involved “proper modelling” undertaken on the basis that releases would be kept below a level that kept the two remaining bridges open: Ch 7 [124]. In other words, in undertaking “with release” modelling, a reasonably competent flood engineer (at this time necessarily operating in strategy W3) would have submerged the two bridges, and on that basis not selected strategy W4 at that time. That conclusion is consistent with his Honour's observation that “even if strategy W4 was engaged, as opposed to W3” that would not have led to any immediate substantial difference in gate operations compared to the “proper implementation of strategy W3” which also required “the immediate inundation of the remaining bridges”: Ch 7 [124], Ch 12 [155].

85. [570]

    In the result, the primary judge found that the “implementation” of strategy W4 required the making of releases that submerged the remaining bridges and “as high as possible” without the combined flows exceeding 4000m3/s at Moggill. In not making those releases in the period to 00:00 on 9 January, Mr Tibaldi was held to be in breach of his obligation to “implement” strategy W4, as well as his obligation to make releases “above the rate of inflows”: Ch 12 [155]. His Honour did not separately consider whether that obligation was also shown to have been breached by a comparison of the actual inflows and outflows from hour to hour during the period from 19:00 to midnight.

86. [571]

    The primary judge's reasoning as to the application of strategy W4 and its implementation (requiring a substantial increase in releases) applied equally to the period of Mr Tibaldi's shift from 00:00 to 07:00 on 9 January: Ch 12 [165]. That sufficiently disposed of Rodriguez’ pleaded claim for that period.

87. [572]

    In its written submissions Rodriguez separately contended that Mr Tibaldi was “obliged to make releases at rates exceeding the rate of inflow”. In response Seqwater submitted that during this part of his shift “Mr Tibaldi made releases that were in excess of inflows”: Ch 12 [168]. Having first noted that there was no pleaded allegation of any failure to that effect on 9 January, the primary judge accepted that had such an allegation been made it would not have been separately established because in the early morning of 9 January Mr Tibaldi made releases that were in excess of inflows: Ch 12 [168]. Before this Court, in challenging the finding of breach during the evening shift of the pleaded obligation to make releases “above the rate of inflows”, Seqwater relied on a similar “hourly” analysis of inflows and outflows.

88. [573]

    Ground 24 challenged the findings that Mr Tibaldi breached his duty of care in failing to implement strategy W4 in his overnight shift on 8/9 January and in failing to make releases that exceeded the rate of inflow in the period to midnight. The primary judge’s finding that Mr Tibaldi was required to implement strategy W4 must be set aside. The analysis supporting that finding depended on a prediction which did not take into account the making of releases during the period of the forecasts. The evidence did not establish that a reasonably competent flood engineer, taking into account releases at rates and for periods assessed as necessary, must have predicted the dam level would exceed 74m.

89. [574]

    It follows from the rejection of the finding that the implementation of strategy W4 was “required” that the finding that Mr Tibaldi was in breach by failing to make releases at levels which exceeded those necessary to submerge the two bridges must also be set aside. The primary judge did not separately address the alleged failure to make releases that exceeded the rate of inflow, treating that obligation as breached by the failure to make releases at levels which would submerge the bridges. To that extent, that finding must also be set aside. Doing so makes it unnecessary to consider Seqwater's “hourly” analysis argument.

90. [575]

    It may seem curious that these quite specific findings of breach in relation to the rates at which releases were being made from Wivenhoe by the evening of 8 January are to be set aside. That is particularly so in circumstances where the judge found that the implementation of strategy W4 on 8 January would not have yielded a different outcome to the proper application of strategy W3. His Honour considered each required, by the end of 8 January, the submerging of the two bridges. However, the primary judge's findings addressed Rodriguez’ pleaded case which, as at 8 January, sought to engage simulation F, which in turn required the adoption of strategy W4 when it commenced at 00:00 on 8 January.

91. [576]

    On Sunday 9 January, Mr Malone signed on at about 06:30 and finished his shift at 21:30. In the 24 hours to 09:00 the average rainfall in the dam catchments had been 53mm for Somerset and 19mm for Wivenhoe. The morning QPF forecast rain in the dam catchments in the 24 hours from 09:00 of 40-60mm. In fact, at least 140mm of rain fell in those catchments during this period. The 4-day 00UTC PMEs received at 00:00 on 9 January forecast rain in the catchments above the dams of between 75-300mm (Dr Christensen) and 50-300mm (State of Queensland): see Table 9-2 at [312] above. Following a discussion in which Mr Malone expressed concerns in relation to downstream flows and “never before seen Wivenhoe releases” which would submerge the remaining two bridges, Mr Ayre called a meeting of the flood engineers at 15:00: Ch 7 [166].

92. [577]

    Mr Malone ran several ROG models in the course of his shift. The first, at 09:00, adopted the same gate openings as Mr Tibaldi’s 9 January 07:00 ROG run and accordingly sought to limit combined downstream flows to 1600m3/s. At 11:01 Mr Malone sent an email to the flood engineers and others entitled “Forecast Rainfall and Possible Runoff”. He noted that “heavy rainfall” was forecast, especially in the 24 hours from 22:00 on 9 January, with totals of between 200mm and 300mm predicted. Noting that “presently” the ratio of runoff to rainfall was “about 0.45 for Wivenhoe, … and 0.75 for Somerset”, his memo relevantly continued: [168]

93. [578]

    At 11:00 Wivenhoe Dam level was 68.54m, leaving storage capacity up to 74m of about 742,000Ml. Mr Malone expected that only “about a third” of that forecast runoff would enter Wivenhoe or Somerset during the three days if the forecast rain fell: Ch 7 [174]. The 4-day 00UTC PMEs available at 00:00 on 9 January also predicted substantial rainfall in the Lockyer Creek and Bremer River catchments: Ch 7 [177]. In the 24 hours to 09:00 on 9 January, very little rain had fallen in those catchments. [169]

94. [579]

    Mr Malone ran further ROG models at midday, 14:00, 16:00, and 18:00 on 9 January: Ch 7 [181], [199], [204]. Whereas the midday run predicted a dam level of 69.21m, the 16:00 run predicted a level of 72.15m and the 18:00 run a level of 72.69m. This was due to continuing rainfall in the dam catchments resulting in higher volumes of ROG inflows being taken into account in the later model runs. The ROG inflows in the midday model were 679,815Ml, whereas the ROG recorded in the 18:00 model was 1,346,488Ml, an increase of 666,673Ml. Later records showed that in the 24 hours from 09:00 on 9 January over 140mm of rain fell in the dam catchments.

95. [580]

    Between 09:00 and 14:00, three further sluice gates were opened at Somerset, increasing the rate of outflow from Somerset to Wivenhoe (the dam level being at or below 100.47m) from 412m3/s to 1034m3/s at 14:00. Those sluice gates remained open until 15:00 on 11 January when they were gradually closed. At 14:00 on 9 January the rate of inflow to Somerset was 2,744m3/s and rising, and the outflow to Wivenhoe was 1,034m3/s. Net inflows to Somerset rising to 4,056m3/s at 16:00 on 9 January resulted in that dam’s level increasing to 102.38m at 00:00 on 10 January. At a dam level of 102m, the discharge from the spillway was approximately 200m3/s. [170] By 00:00 on 10 January the rate of outflow from Somerset to Wivenhoe was 1,359m3/s, resulting in a net inflow to Somerset of 1,924m3/s. At the same time, the inflow to Wivenhoe (excluding Somerset) was 6,577m3/s and Wivenhoe was releasing 1462m3/s: Ch 7 [107].

96. [581]

    Following the meeting of flood engineers at about 15:30, the following entry was made in the Event Log (set out at Ch 7 [190]):

97. [582]

    It was agreed that starting that evening two flood engineers would work together on each 12-hour shift. The afternoon QPF available at about 16:00 forecast a further 50-80mm rainfall in the dam catchment areas for the 24 hours from 15:00: Ch 7 [138].

98. [583]

    In his situation report distributed before 18:00, Mr Malone reported that catchment average rainfall for the past 12 hours for Somerset Dam was 150mm and for Wivenhoe Dam 80mm, and that the “bulk of the rain … has fallen in the upper reaches of the Stanley and Brisbane Rivers”. [171] Wivenhoe was at 68.7m with an estimated peak inflow of about 5,000m3/s and water level of at least 72.5m by 22:00 on 11 January. Mr Malone advised that the current gate operation strategy would maintain flows of around 1600m3/s for the next 24 hours and that those releases might have to be reduced “as Lockyer flows increase”. He added that “releases may have to be increased significantly [on 10 January] depending on the rain in the next 12 to 24 hours”. In cross-examination Mr Malone accepted that there was an “increasing likelihood” by this time that Wivenhoe would reach 74m: Ch 7 [203].

99. [584]

    At around 19:00 Mr Ruffini undertook an ROG run and a 24-hour QPF run. The former predicted a second set of natural peak flows in Lockyer Creek during the evening of 10 January and in the Bremer River during the evening of 9 January. The ROG run also predicted a maximum height of Wivenhoe of 71.69m and combined peak flows at Lowood and Moggill of around 2550m3/s at 09:00 on 12 January. The “with forecast” run predicted a maximum dam level of 73.16m on 11 January at 16:00 and combined flow rates at Lowood and Moggill of around 3300m3/s at 08:00 on 11 January. Neither of these runs contemplated more than one additional gate opening until the remaining two bridges were submerged: Ch 7 [214].

100. [585]

     In the face of that modelling and the following two days of forecast heavy rain, the primary judge held that by the evening of 9 January it should have been “obvious to each of Mr Ruffini, Mr Ayre and Mr Malone that there was a likelihood of Wivenhoe dam exceeding 74m”, so as to require the implementation of strategy W4: Ch 7 [215], [216].

101. [586]

     Shortly after 19:00, senior Seqwater and Brisbane City Council staff were advised that higher releases in the order of 3,000m3/s were expected to be necessary in view of the heavy rain and that releases “causing damaging flooding are likely to be necessary”: Ch 7 [217].

102. [587]

     Mr Malone issued another situation report at about 21:00: Ch 7 [223]. Having referred to the heavy rainfall (totals up to 100 to 140mm in the Wivenhoe and Somerset catchments in the last six hours) and the level of Wivenhoe (69.1m and rising), the report said that “at this stage, the dam will reach at least 73m” during 11 January. The report continued:

103. [588]

     Mr Malone’s report also advised that the Wivenhoe releases combined with Lockyer flows and local runoff would mean that all bridges downstream were adversely affected. At 21:10, Mr Ayre advised Seqwater’s dam operations manager that “releases will need to be ramped up from current 1400m3/s to 2500m3/s which will cause flooding in low lying areas of Brisbane.”

104. [589]

     After the conclusion of Mr Malone’s shift, further ROG and “with forecast” models were run at 22:00: Ch 7 [231]-[235]. The former predicted further natural peak flows at Lowood (613m3/s) and Moggill (830m3/s) in the morning of 10 January. The latter predicted a maximum dam level of 75.11m and significantly increased flow rates at Lockyer Creek (1338m3/s) and Bremer River (1839m3/s) in the evening of 10 January. Those flows, when combined with releases from Wivenhoe, produced peak rates at Lowood (4222m3/s) and Moggill (5652m3/s) late on 10 January or early on 11 January.

105. [590]

     Mt Crosby Weir Bridge and Fernvale Bridge were submerged at around 22:40 on 9 January. At 02:00 on 10 January the gates at Wivenhoe were opened a further seven increments, increasing releases to 2,015m3/s at 09:00.

106. [591]

     In Ch 12 the primary judge made the following findings of breach:

107. [592]

     As to the first finding of breach, the primary judge held that on 9 January each of Mr Malone, Mr Ayre and Mr Ruffini was obliged to “implement” strategy W4, which meant “an immediate increase in releases to a level above that necessary to inundate the remaining bridges”: Ch 12 [173]. By the early evening on 9 January it was reasonably clear that the level of Wivenhoe would exceed 74m. Mr Malone accepted that by 18:00 there was an “increasing likelihood” that would occur. His situation report at 21:00 said that the dam level would reach “at least 73m”. The 22:00 “with forecast” modelling predicted a maximum dam level of 75.11m and Mr Ayre accepted that, looking at the position at around 22:30, it was “very likely” the 74m level would be exceeded: Ch 7 [242]. Whether that was the position at the commencement of Mr Malone’s shift at 06:30 on 9 January, viewed from the perspective of any reasonably competent flood engineer and accepting that prediction was to take account of releases, is not obvious.

108. [593]

     The primary judge found that the “flood engineers were obliged to operate in W4 throughout the day”: Ch 7 [260]. His Honour’s reasoning supporting that holding was the same as that supporting the similar finding made in relation to operations on 8 January, which addressed the position at 00:00 on 8 January: Ch 10 [10], [19]. That reasoning started with four-day inflow volume estimates based on the 00UTC PMEs available at 00:00 on 9 January to give a predicted height of the dam on the assumption that no releases were made during the forecast period. The use of Dr Christensen’s or Mr Giles’ (corrected) four-day inflow estimates (respectively 886,000Ml and 782,000Ml), based on those forecasts, gave a predicted dam level of approximately 75.21m (Dr Christensen) and 74.64m (Mr Giles corrected): Table 9-6 at [314] above.

109. [594]

     In finding that strategy W4 was engaged, the primary judge also referred to Mr Ayre’s 72-hour modelling exercise conducted on 8 January at 15:00 which supported that conclusion if “proper” modelling were undertaken and releases were kept below the level that submerged the two bridges: Ch 12 [173], Ch 10 [19], Ch 7 [124].

110. [595]

     In relation to Somerset flood operations, the primary judge concluded that a reasonably competent flood engineer would not have “substantially” increased releases from Somerset into Wivenhoe without ensuring that the rate of outflow from Wivenhoe substantially exceeded the rate of inflow from Somerset. Taking that step was said to reduce the risk of Wivenhoe levels rising above 74m and forcing releases that could cause significant downstream flooding. The background to the operation of Somerset, and certain findings in issue on the appeal have been addressed previously in discussing grounds 14 and 15 in part 16 above. It is convenient to reiterate and expand on that discussion in order to deal coherently with the issues raised by ground 25(a).

111. [596]

     As at 14:00, five sluice gates were open at Somerset with an outflow of 1034m3/s. Thereafter, the overall rate of outflow increased as the dam level rose because the discharge rate via the spillway increased. By 00:00 on 10 January the outflow from Somerset was 1403m3/s, increasing to 1577m3/s at 10:00 on 10 January and 1665m3/s at 00:00 on 11 January. The sluice gates were closed between 04:00 and 09:00 on 11 January. During the period from 14:00 on 9 January to 00:00 on 10 January, the total outflows from Wivenhoe were between 1386 and 1462m3/s. By 00:00 on 11 January those outflows had progressively increased to 2713m3/s.

112. [597]

     Rodriguez’ case in relation to the operation of Somerset was put in two ways. First, it was said that strategy S3 was engaged and that there was a failure to implement that strategy, by either closing the eight crest gates or closing the sluice gates: Ch 12 [177]. As the Manual provided that during flood operations the crest gates should not be closed, the primary judge did not accept that a reasonably competent flood engineer would or must have closed those gates. It followed that the only precaution that could have been taken to “implement” strategy S3 was to close the sluice gates: Ch 9 [345], Ch 12 [177]. The primary judge was not satisfied that a reasonably competent flood engineer must have expected Wivenhoe to exceed 75.5m, so as to engage strategy S3, (on a “no release” basis) before 18:00 on 9 January when the daily 00UTC PME forecasts became available. It followed that there was no breach in failing to implement strategy S3 and close the sluice gates before the evening of 9 January, and after Mr Malone’s shift finished: Ch 12 [178].

113. [598]

     The judge then considered the second element in Rodriguez’ case, that a reasonably competent flood engineer would have ensured that the rate of outflow from Wivenhoe substantially exceeded the rate of outflow from Somerset, including by closing the sluice gates: Ch 12 [179]. The risk which was held to require the taking of that precaution was the “likelihood bordering on a certainty, of Wivenhoe Dam levels rising above 74m and forcing releases that could cause significant downstream flooding”: Ch 12 [181]. Because one consequence of closing the sluice gates was to change the dam storage levels relative to each other, it was necessary to have regard to the provisions in strategies S2 and S3 which required that the two dams be operated in tandem. The object was to produce a specified correlation between the predicted maximum levels of the dams. That correlation was shown by the Operating Target Line (OTL) graph shown at [353] above.

114. [599]

     The primary judge accepted that one of the countervailing risks which had to be considered was the “real possibility that the risk of overtopping both dams could not be equalised before either of them was overtopped”: Ch 12 [181]. In doing so, and referring to his earlier consideration of Dr Christensen’s simulations, the primary judge acknowledged that this required attention to the OTL: Ch 9 [347]ff, Ch 10 [44]-[49]. In the conditions facing Mr Malone by the afternoon of 9 January – the level of Wivenhoe rising and Somerset level being above 100.45m – strategy S2 required that the OTL was “generally” to be followed as the flood event proceeded. One of the objectives of this requirement was to minimise equally the expected flood level peak of each dam relative to its ultimate failure level (80m for Wivenhoe and 109.70m for Somerset). That was to be achieved by undertaking gate operations which enabled the movement of the “duty point” (the point on the graph recording the current actual level of both dams) progressively towards the “target point” (the point on the OTL based on predicted maximum levels of the dam “using the best forecast rainfall and stream flow information” [172] ).

115. [600]

     This ground challenged the findings of breach by failing to implement strategy W4 and by substantially increasing releases from Somerset without ensuring that the rate of outflow from Wivenhoe substantially exceeded the rate of inflow from Somerset. The latter breach resulted from the opening of three further sluice gates from 09:00, rather than the closing of all of them either then or later on 9 January.

116. [601]

     Seqwater submitted that the finding that strategy W4 was engaged on the morning of 9 January could not be sustained as it was based on the “no release” assumption. In response Rodriguez noted that the challenges to the finding that strategy W4 was engaged were “repetitious” of arguments made in respect of earlier days, and principally with respect to the “no release” assumption.

117. [602]

     As the above analysis of the primary judge’s findings shows, it was not until the early evening on 9 January that a reasonably competent flood engineer taking account of proposed releases ought to have expected that the level of Wivenhoe was likely to exceed 74m. For that reason Seqwater’s challenge to the primary judge’s finding of breach in failing to implement strategy W4 must be upheld.

118. [603]

     With respect to the closure of the sluice gates at Somerset, Seqwater submitted that Mr Malone’s releases resulted in the water levels of the two dams “tracking upwards towards a point on the OTL”. On the other hand, the primary judge’s finding had the consequence that the dam levels would move away from, rather than in the direction of, the OTL. The primary judge considered that the Manual allowed for “temporary movements away from the target line”: Ch 10 [40], Ch 3 [89], [91]. Seqwater challenged that interpretation of the Manual and contended that even if such temporary departures were permitted, the primary judge’s reasons did not explain why a reasonably competent flood engineer was “required” to depart from the Manual’s general instruction to follow the target line. In response, Rodriguez contended that the Manual did not require that the dams be operated in a manner that moved towards the OTL “as quickly as possible”.

119. [604]

     Contrary to Seqwater’s submission, the judge’s reasons did explain why he considered Mr Malone was required to depart from the OTL. That reason was his Honour’s conclusion that it was likely “bordering on a certainty” that the level of Wivenhoe would rise above 74m, requiring forced releases that could cause significant downstream flooding. That conclusion proceeded from the primary judge’s finding that strategy W4 was engaged “throughout the day” on 9 January. For the reasons given above, his Honour’s findings did not establish that a reasonably competent flood engineer, taking into account proposed releases, must have expected, before the early evening, that the level of Wivenhoe would exceed 74m. In the absence of a finding to that effect the challenge to the breach with respect to the making of releases from Somerset also must be upheld. Absent that justification, however, it is unclear how compliance with the Manual’s guidance as to the coordinated operation of the two dams could constitute negligence. That issue was raised by Seqwater in relation to the breach on 10 January, but, as noted below need not be resolved.

120. [605]

     In the 24 hours ending at 09:00 on Monday, 10 January, the average rainfall in the dam catchments was 210mm (Somerset) and 124mm (Wivenhoe). The average rainfall in the Lockyer Creek and Bremer River catchments was 66mm and 44mm respectively. [173] The morning QPF forecast rainfall of 50-100mm during the 24 hours from 10:00.

121. [606]

     Mr Ayre and Mr Ruffini were on duty to 07:00 when Messrs Malone and Tibaldi’s day shift commenced. At that time the level of Wivenhoe was 71.16m and the level of Somerset was 102.98m. In the previous 18 hours the inflows to Somerset ranged between 4,011m3/s at 18:00 on 9 January and 2,403m3/s at 07:00 on 10 January. The releases from Somerset to Wivenhoe during that same period rose from 1,121m3/s to 1,535m3/s, reflecting increases in the level of that dam with consequences for the spillway discharges. The releases from Wivenhoe between 18:00 on 9 January and 07:00 on 10 January increased from 1,404m3/s to 1,875m3/s.

122. [607]

     At trial, Rodriguez addressed only one complaint of breach relating to the actions of the flood engineers in the period to the late afternoon of 10 January: Ch 12 [192]. From that point no further allegations of breach were advanced.

123. [608]

     There was no specific allegation of breach by failing to implement strategy W4 in relation to releases from Wivenhoe. Specifically it was not said that the conduct of Messrs Malone and Tibaldi in suspending gate openings from 09:00 to keep the combined flows downstream below 3500m3/s was a breach of duty: Ch 12 [194], [202]. As the primary judge observed, the releases made after 08:00 on 10 January exceeded 2000m3/s “which was consistent with engaging strategy W3 and arguably strategy W4”: Ch 12 [193].

124. [609]

     The relevant allegations of breach in this period related to the operation of Somerset. At 07:00 the crest gates and five of the sluice gates remained open. Rodriguez alleged that any reasonably competent engineer would have reduced “significantly” the releases from Somerset. The primary judge found a breach in those terms, that failure continuing “up to late on the afternoon of 10 January”: Ch 12 [202].

125. [610]

     Addressing the flood engineers’ approach to Somerset Dam operations “on (9 and) 10 January”, the primary judge concluded that the “perceived necessity to move towards the [OTL] as directly as possible took precedence over the concerns in the Manual about avoiding damaging downstream flows and around a time when there was no appreciable risk of the failure level of Somerset being reached if the crest gates at Somerset remained open”: Ch 12 [199]. The finding as to there being no “appreciable risk” was based on Dr Christensen’s assessment of the likelihood of Somerset being overtopped in the circumstances of simulation F. That assessment took into account releases that Dr Christensen considered could be made from Somerset over the relevant three or four day forecast period, provided the crest gates remained open: Ch 10 [44]-[48].

126. [611]

     This ground challenged the finding that Mr Malone and Mr Tibaldi breached their duty of care by failing to close the sluice gates at Somerset. The relevant period of that breach was from 07:00 to about 17:00 on 10 January. Assuming that the sluice gates were closed by 08:00, the volume of water that would have been retained in Somerset was no more than 36,000Ml; that volume was inconsequential.

127. [612]

     Seqwater challenged this finding on two bases. First, it contended that Mr Malone and Mr Tibaldi were not negligent in seeking to follow the OTL. Secondly, it submitted that there was no evidentiary basis for the conclusion that the four-day rainfall forecasts did not present a risk of Somerset overtopping, or a position being reached where the risk of overtopping both dams could not be equalised before one overtopped.

128. [613]

     In response, Rodriguez supported the primary judge’s construction of the Manual in relation to flood operations at Somerset and relied on Dr Christensen’s evidence as demonstrating that the four-day forecast inflows to Somerset could have been discharged by releases over that period.

129. [614]

     Given the small amount of water involved, there is no utility in resolving these arguments, as the challenged breach could have no relevance to the causation analysis relied on by Rodriguez. It is sufficient to record that they turn in part on the construction of the Manual and in part on assessments as to where and when, if at all, the four-day forecast rain as at the morning of 10 January might fall in the dam catchment areas, as shown by Dr Christensen’s assessment of the risk of Somerset being overtopped in the circumstances of simulation F: Ch 9 [45]-[47]. On any view, breach was not demonstrated on the reduced standard in s 36(2).

21 Rodriguez’ contention par 3

1. [615]

   For convenience pars 46, 47 and 48(b)-(f) of Rodriguez’ written submissions are repeated below:

2. [616]

   Some of what follows includes factual material which has already been summarised or referred to in the breach analysis addressing grounds 23(b), 23(c), 24 and 25(a). Again that has been done for ease of reference.

3. [617]

   It is convenient to start with the contention in par 48(d) which is addressed to acts or omissions of Mr Malone and Mr Tibaldi during their day shift on 10 January. Rodriguez’ contention was that releases from Wivenhoe were suspended from 09:00 to 15:00 on 10 January in an attempt to keep the combined flow at Moggill at or below 3500m3/s; that being the combined flow at Moggill 16 hours after the release of water from Wivenhoe. At that time, it was contended that, if the flood engineers were operating in strategy W3, the intent should have been to limit that combined flow to less than 4000m3/s, noting that that was “the upper limit of non-damaging floods downstream”.

4. [618]

   As a result, between 09:00 and 15:00 on 10 January actual releases ranged between 2015m3/s and 2155m3/s. The releases contended for by Rodriguez depended on which simulation was adopted. In simulation F, during this period the proposed releases ranged from 1825m3/s to 1658m3/s. That was because in that simulation, whilst the objective was to keep flow rates at Moggill below 4000m3/s, looking forward the peak rate at Moggill was predicted to be between 2000 and 2090m3/s. According to Dr Christensen’s methodology, those predictions were made using ROG and 24-hour QPF forecasts: Ch 10, [255]. Only simulation G modelled different releases during this period. That simulation commenced on 10 January and, in view of the actual dam level, adopted a much more aggressive target for combined downstream flows of 5300m3/s on the basis that “releases and downstream flows could not be held below the threshold” of 4000m3/s: Ch 10 [256].

5. [619]

   The primary judge described the decision to delay the increase in gate openings as “unreasonable” given that the prevailing conditions required that strategy W4 be engaged: Ch 7 [336]. However, as his Honour later observed, it was not the subject of any separate allegation of breach: Ch 12 [194]. That was no doubt because the releases proposed in simulation F were less than those in fact made [174] and simulation G was only relied on if the Court found that the reasonably prudent flood operations ought to have first commenced on 10 January.

6. [620]

   Notwithstanding that the conduct to which this contention was directed was not alleged to be a breach, or relied on as resulting in different releases from those that would otherwise have been made, Rodriguez’ submission as to that conduct involving a breach of the standard in s 36(2) may be addressed.

7. [621]

   In his situation report issued shortly before 18:00 on 8 January, Mr Ayre had said:

8. [622]

   Subsequently, the evidence suggested there had been a debate about whether to keep downstream flows to 3,500m3/s or 4,000m3/s. In particular, the primary judge referred to a note of a meeting at 08:30 on 10 January which “indicates that one of the topics was ‘3.5 and 4’, which I infer was a discussion about the possibility of urban flooding resulting from an outflow rate of 3,500m3/s or 4000m3/s”: Ch 7 [298]. The fact that there appear to have been discussions about which outflow rate to choose suggests that this was not a decision which was so unreasonable that no dam operator could properly have considered the selection of the lower rate to be reasonable.

9. [623]

   Most relevantly for present purposes, the Event Log for 10 January recorded four requests from Brisbane City Council (BCC) to restrict flows to 3,500m3/s. They were:

10. [624]

    There may be doubt about the force of all aspects of the notes. But they confirm that active thought was being given to the Council’s view that combined flow rates should be kept to 3,500m3/s, notwithstanding that operating in strategy W3, the Manual authorised flow rates that “should not exceed 4,000m3/s”.

11. [625]

    This changed in the early afternoon. The Event Log records that at 12:36 on 10 January Ipswich City Council was called to advise “we are moving strategy from urban damage control to dam safety priority”. Brisbane City Council was not in fact contacted until 14:30 (seemingly the explanation for the delay in contacting Brisbane City Council is reflected in the note at 12:33 – “Made two calls to BCC – no answer – left messages” and a similar entry a couple of hours later). It is plain that the decision to move to strategy W4 was made no later than shortly after noon on 10 January. Mr Malone issued a directive at 15:00 to open Wivenhoe Dam a further 10 increments, one every half hour from 15:00 (as it turned out, a technical issue delayed the commencement of those increments until 16:00).

12. [626]

    The issue is whether it was unreasonable, in the sense required by s 36(2), to attempt to maintain a flow at Moggill of 3,500m3/s, as opposed to 4,000m3/s, until around noon on 10 January.

13. [627]

    The Manual stated that 4,000m3/s at Moggill was the upper limit of non-damaging flows. But a senior officer at Brisbane City Council called the FOC at 00:45 to advise that in the circumstances which actually prevailed, flow rates at Moggill above 3,500m3/s were damaging. There was no evidence to suggest that taking account of and acting consistently with Mr Morris’ advice was so unreasonable that no authority in Seqwater’s position could properly consider that doing so was a reasonable exercise of its flood mitigation function. Such a conclusion is far from self-evidently correct. The 4,000m3/s rate in the Manual is scarcely precise. The flood engineers were being told, by a person whose views might reasonably be regarded as carrying weight, that making releases which would result in a flow rate of 4,000m3/s at Moggill would cause millions of dollars of damage. Conversely, the reduction of outflows from Wivenhoe by the 500m3/s requested by Mr Morris was, in the scheme of things, a relatively minor proportion of the releases being made. [175] Adherence to this element of the strategy reduced the releases so that they were around 80% of what they might otherwise have been (they were slightly above 2,000m3/s, rather than slightly above 2,500m3/s).

14. [628]

    Still further, from the perspective of Messrs Malone and Tibaldi on 10 January, the strategy to limit flows at Moggill to 3,500m3/s had been confirmed by Mr Ayre (including in his situation report shortly before 18:00 on 8 January reproduced above: “releases from Wivenhoe Dam ... will be maintained below 3,500m3/s”). Thus Mr Malone wrote in an email at 09:55 on 10 January that “[t]he current operational strategy is to aim for a flow of no greater than 3,500[m3/s] in the lower Brisbane River”. That email, and others, [176] proceeded on the uncontroversial assumption that there could only be a single operational strategy throughout a flood event; the Manual does not support the implausible conclusion that, absent a material change in circumstances, there would be a change in strategy when each flood engineer began his shift. Although Messrs Malone and Tibaldi acceded to the decision, that does not deny that it would have been a large thing for either or both of them to disregard it unilaterally.

15. [629]

    True it is that on 10 January dam levels were rising sharply, in the order of 20cm per hour even with releases at just over 2,000m3/s. At 01:00 the level was 69.97m, by 09:00 it was 71.56m and by noon (shortly before Ipswich Council was told that operations were moving from urban damage control to dam safety priority) it was 72.07m. But if it turned out to be necessary to increase releases such that urban areas would be inundated (as of course proved to be the case), it is easy to see how on 10 January it could be thought that the flood storage capacity reflecting the 500m3/s which had been forgone could be replaced over time.

16. [630]

    During the early morning of 10 January there were unprecedented inflows into Wivenhoe, rising to 10,095m3/s at 08:00. However, it was not known that at the same time on the following day inflows would be 8,060m3/s and would then rise to 11,561m3/s at 13:00, before slowly falling to 8,196m3/s at 17:00.

17. [631]

    The act or omission was outside the case pleaded by Rodriguez and not the subject of any finding of breach by the primary judge. Rodriguez has failed to establish that attempting for some five hours or so on 10 January 2011 to maintain expected flows at Moggill (16 hours later) below 3,500m3/s, as requested by Brisbane City Council in order to avoid urban inundation, was so unreasonable that no dam operator could properly consider it to be reasonable. That applies to the conduct of all of the flood engineers, but in the case of those for whom Seqwater might be vicariously liable, the conclusion is fortified by the fact that Mr Ayre had endorsed that decision.

18. [632]

    It follows that this submission must be rejected.

19. [633]

    These were generic challenges which particularly informed the more specific contentions, including pars 48(b) and (c).

20. [634]

    The analysis of the primary judge’s findings of breach between 6 and 10 January showed that, whilst the flood engineers made decisions about gate operations using the RTFM, they did not do so using forecast rainfall other than in a few cases. They did however have regard to QPF and PME forecasts in assessing possible inflows to the dams, as well as downstream flows from the Lockyer Creek and Bremer River catchments. In doing so they made assessments as to where forecast rain might fall and when. For example, in his 11:01 email on the morning of 9 January, [177] Mr Malone sought to quantify possible inflows to the dams over the ensuing three days based on PME forecasts and historical runoff rates. In doing so he concluded that the runoff into Wivenhoe could range between 500,000 and 1,000,000ML, whilst noting that “these estimates are based upon forecast rainfall which may or may not eventuate”.

21. [635]

    Dr Christensen’s methodology in simulation C used 4-day PMEs to select strategy, and in simulations F and H used 24 hour QPF forecasts to select strategies and relied on the four and eight day forecasts for what was referred to as “situational awareness”: Ch 10 [3], [5]. In predicting likely dam levels in these simulations he adopted the “no release” assumption. Doing so invariably resulted in a higher predicted dam level over a period of up to four days in turn resulting in the adoption of a higher strategy.

22. [636]

    In response to this use of the four-day forecasts, Seqwater submitted that there was no evidence that a dam operator would consider that any operations which did not use 4-day PME forecasts in these ways was so unreasonable that it could not be considered a proper exercise of its functions. That was because (i) Rodriguez did not seek to make out such a case based on s 36(2) at trial and (ii) Rodriguez did not suggest that such uses of the 4-day PMEs was required by the Manual.

23. [637]

    The asserted force of par 48(f) derives from the following observations of the primary judge in Ch 7:

24. [638]

    The judge’s statement on which par 48(f) is based is not a finding of primary fact. Rather, it is a conclusion about the flood engineers’ assumptions and operating practices which principally involved ROG modelling that was regularly updated to account for changes in rainfall, above and below the dams. The period of “12 to 15 hours” referred to the time rain in the upper reaches of the catchments took to flow into the dams or into the downstream Brisbane River. The statement that the flood engineers overestimated “the capacity to release water beyond” that period implied that in making decisions about releases based on modelling they did not have regard to possible future dam inflows and downstream conditions. There was contemporaneous evidence suggesting otherwise. It included, in addition to Mr Malone’s 11:01 email of 9 January, [178] Mr Ayre’s comments in his status report issued at 12:15 on 8 January and repeated in his evening situation report issued shortly before 18:00: [179]

25. [639]

    At par 48(b) Rodriguez contended that the flood engineers sought to avoid closing Mt Crosby Weir and Fernvale bridges at the “expense of guarding against the risk of urban inundation”, relying on findings at Ch 1 [62] and Ch 6 [211] and [255]-[267]. Those three passages in the judgment all related to 6 and 7 January 2011, immediately after the declaration of the flood event. The primary judge said in the first two paragraphs:

26. [640]

    The second paragraph asserted that the initial assessment by Mr Ayre did not comply with section 8.3 of the Manual. That may be put to one side for present purposes on the basis that Seqwater cannot be made liable for Mr Ayre’s acts or omissions.

27. [641]

    Much of the second paragraph is premised on modelling using the “no release” assumption. For the reasons given in response to ground 8, it was not unreasonable to model during a flood event making allowances for releases.

28. [642]

    Both paragraphs reproduced above, as well as parts of Ch 6 [255]-[256], [257], [258], [262] and [264], held that releases should have been made in accordance with strategy W3, not W1. Insofar as this submission was directed to the decisions to keep the downstream bridges open, rather than protecting the urban areas from inundation, it is best addressed immediately below, with par 48(c).

29. [643]

    However for the reasons upholding grounds 23(b) and (c), in the absence of the “no release” assumption the evidence did not justify a finding that a reasonably competent flood engineer using four-day forecasts must have concluded by the end of Mr Malone’s day-shifts on 6 and 7 January that the level of Wivenhoe was likely to exceed 68.5m so as to engage strategy W2 or W3. [180] It followed that their conduct did not subvert the policy of protecting against urban inundation.

30. [644]

    Rodriguez’ par 48(c) contended that the flood engineers were following strategy W1 when they should have been following strategy W3. That is to say, it was directed to the period during which the flood engineers made releases designed to leave certain downstream bridges open, when it was said their focus should have been protecting against urban inundation.

31. [645]

    As formulated, the submission has the potential to confuse. It does not matter for present purposes whether the flood engineers believed they were adhering to strategy W1 or W3. On Rodriguez’ case, resting as it did upon the no release assumption, a flood engineer could be operating in a higher strategy, but only releasing water in a way which would be consistent with strategy W1. The submission in par 48(c) is to be approached according to what the flood engineers did, rather than what they believed. The issue is, in accordance with Rodriguez’ submission, “entirely about the acts and omissions of the flood engineers”.

32. [646]

    That said, Rodriguez’ pleaded case was that on 7 January any reasonably competent flood engineer would have concluded that strategy W3 was engaged and proceeded on the basis that the primary consideration was protecting urban areas from inundation by limiting the flow at Moggill to less than 4000m3/s. The pleaded case on 8 and 9 January was that any reasonably competent flood engineer would have concluded that strategy W4 was engaged, the primary consideration then being to protect the structural safety of the dam. [181] These pleaded strategies were the same as those adopted in simulation C on 7 January and simulations F and H on 8 and 9 January. In dealing with that pleaded case, the breaches found included failing to implement the pleaded strategy. [182] With respect to the submerging of the last of the downstream bridges, the primary judge’s reasoning was that the “implementation” of strategy W3 “required” a substantial increase in releases sufficient to inundate those bridges by midnight or thereabouts on 7 January; [183] and that the “implementation” of strategy W4 on 8 January similarly required releases that submerged those bridges and were “as high as possible” without the combined flows exceeding 4000m3/s at Moggill. [184] The same finding was made as to the implementation of strategy W4 on 9 January in relation to releases. [185]

33. [647]

    The upholding of grounds 23(b), 23(c), 24 and 25(a) means that there were no breaches of duty on the part of the flood engineers in failing to make sufficient releases on 7, 8 and 9 January for which Seqwater could be liable, making it unnecessary to determine whether that conduct breached the standard imposed by s 36(2). Nevertheless, the question may be addressed on the basis that on 7, 8 and 9 January (during shifts undertaken by Mr Malone or Mr Tibaldi) greater releases should have been made, submerging all of the downstream bridges, but not so as to inundate urban areas of Ipswich and Brisbane.

34. [648]

    Only at 16:00 on 7 January, the second day of the flood event, and with dam levels at 68.06m, were the radial gates opened at all. By 08:00 on 8 January, dam levels were at 68.52m, yet releases were restricted to around 1200-1300m3/s until the evening of the following day. The purpose was to keep the Mt Crosby Weir and Fernvale bridges open. The Manual provided that when dam levels were likely to exceed 68.5m, the “primary consideration” was to protect urban areas from inundation. With substantial rain forecast, it was a breach of that direction not to make releases which would at least submerge all of the bridges and limit later damage if greater releases became necessary. That is to say, the releases over this period should have been closer to 2,000m3/s. Eventually, keeping the bridges open proved impossible, and the gates were opened further from the evening of 9 January. By 08:00 on 10 January releases exceeded 1900m3/s. The judge found that the releases made by the flood engineers after that time were consistent with engaging strategy W3, and arguably strategy W4: Ch 12 [193].

35. [649]

    Three points should be made in relation to this scenario. First, this aspect of Rodriguez’ case does not turn solely on the “no release” assumption, because by 08:00 on 8 January the level of Wivenhoe exceeded 68.5m. However the factor identified by the primary judge as justifying the “substantial step” of inundating the two bridges was “the likelihood of forced releases above 74m and downstream flows above 4000m3/s”, which his Honour described as “very real” by 00:00 on 7 January: Ch 12 [132]. Secondly, the primary judge’s assessment of that likelihood depended on a number of possibilities coming to pass. Those possibilities included the average of the forecast rain falling in the dam catchments over the following four days, insufficient releases being made over that period and heavy rain downstream resulting in significant natural flow rates at Lowood and Moggill. [186] And thirdly, because the lower releases which kept those bridges open were sustained for a relatively lengthy period of time, there was a significant impact on Wivenhoe’s flood storage capacity as a result of that approach.

36. [650]

    There are a number of matters to be taken into account in assessing the contention. First, the fact that the relevant acts and omissions may have involved a departure from the Manual, even in a serious way, does not of itself entail breach of the standard in s 36(2). We do not accept that a mere breach of the Manual has that consequence. Indeed, breach of the Manual may not entail a breach of the standard of care under s 9. In any event, for the reasons earlier given, it is necessary to apply the language of the statute.

37. [651]

    Secondly, in applying the test mandated by s 36(2), it is important to keep in mind that the decision for the flood engineers was not the binary choice between keeping the bridges at Fernvale and Mr Crosby Weir open, at the expense of inundating urban areas, and saving urban areas from inundation. The decision for the flood engineers was between making releases which would inevitably inundate the major downstream bridges, thereby reducing the risk of urban inundation, or keeping those bridges open with a consequential increased risk of urban inundation.

38. [652]

    Thirdly, it is important to put to one side reasoning based on hindsight. With the benefit of hindsight, it is known that not merely was there urban inundation, but also that even if Dr Christensen’s simulations had been adopted, hundreds if not thousands of properties would have been inundated in any event. Indeed, it is known that even if there could have been no releases from Wivenhoe whatsoever, there would have been damaging floods. In Ch 7 the primary judge stated: [187]

39. [653]

    Unquestionably, the damage to urban areas was exacerbated by the releases from Wivenhoe. But it will be seen from the above passages that slightly less than half of the peak flow at Moggill was attributable to releases from Wivenhoe. Rodriguez’ case was, of course, that insofar as there should have been greater releases on 6, 7, 8 and 9 January, inundating all of the downstream bridges, there would have been greater flood storage capacity; and correspondingly less need to make the critical releases, which in fact combined (16 hours later) with the period of peak natural flows at Moggill.

40. [654]

    But that was not known on 6, 7, 8 or 9 January. The question whether the standard of care required by s 36(2) was breached must be assessed prospectively, no differently from any other question of breach.

41. [655]

    Fourthly, the onus lay with Rodriguez to adduce evidence that no public authority in Seqwater’s position could properly have considered keeping the major downstream bridges open, in the context of the possible outcomes adverted to above, to be a reasonable exercise of its functions.

42. [656]

    Stated more fully, the question is whether the asserted acts and omissions of Messrs Malone and Tibaldi, for which Seqwater was held to be vicariously liable, were so unreasonable that no dam operator with the flood mitigation function of Seqwater could properly consider them to be a reasonable exercise of that function. First and foremost, there is no evidence of this. That reflects the fact that that was not how the case was run at trial. No witness gave evidence of what a dam operator would or would not consider so unreasonable that what occurred could not properly be regarded as a reasonable exercise of that function.

43. [657]

    It is far from clear that a court, with no real understanding of the operation of a dam with water storage and flood mitigation functions, can supply that gap in the evidence. There are two distinct aspects to this concern.

    1. (1)

       It is quite unclear how prescriptive or how discretionary the Manual was reasonably understood to be, save that it was common ground that (i) the engineers were obliged to follow it, but (ii) much was left to professional judgment.

    2. (2)

       It is also quite unclear what weight would be given by a dam operator in the position of Seqwater to the fact that a flood engineer on duty was ordinarily required to adhere to the strategy determined by the Senior Flood Engineer. This point is developed in more detail below by reference to the facts confronting Messrs Malone and Tibaldi over the critical four days. But essentially it is one thing to conclude that it was a breach of the duty to take reasonable care for a more junior flood engineer on a shift not to make more releases; it is another thing to ask whether a flood engineer who failed to depart from the overall strategy determined by the Senior Flood Engineer was acting so unreasonably that no dam operator could properly consider adherence to the Senior Flood Engineer’s strategy to be a reasonable approach to dealing with that flood event.

44. [658]

    Rodriguez attempted to address the gaps in evidence by submitting that any breaches by the flood engineers were “flagrant”. Use of that pejorative ordinarily conveys conscious breach of rights (such as a flagrant infringement of copyright) or, at least, a reckless indifference to the position. But Rodriguez has not established that the flood engineers were not attempting to take steps which they sincerely believed at the time to be properly directed to mitigate flooding. The contemporaneous documents do not suggest that the flood engineers were doing other than attempting to perform that function. If there were evidence, or cross-examination, to the effect that the flood engineers were deliberately taking steps to inundate urban areas, or failing to take steps which would avoid inundating urban areas, or were recklessly indifferent to that prospect, then this Court was not taken to it and could not in any event assess it given the limited scope of Rodriguez’ contentions on the appeal.

45. [659]

    Secondly, that the four flood engineers appear to have acted by way of consensus, but subject ultimately to the strategy determined by the Senior Flood Engineer, Mr Ayre, has two presently relevant consequences.

    1. (1)

       One is the point made by Seqwater in oral submissions: [188]

    2. (2)

       The second is that, while it is rational to identify the conduct over 6, 7, 8 and 9 January as a whole, Seqwater could only be vicariously liable for breaches of duty by Messrs Malone and Tibaldi. That was the finding of the primary judge, and was reflected in the way the appeal was run in this Court. In particular, no part of Rodriguez’ submissions contended that Seqwater was liable for breaches of duty by Messrs Ayre or Ruffini. To do so would have required a further notice of contention, as well as being contrary to the proposition that two persons cannot be vicariously liable for the conduct of another: Oceanic Crest Shipping Company v Pilbara Harbour Services Pty Ltd, [189] (as the primary judge observed at Ch 11 [171]). While Seqwater’s employees Mr Malone and Mr Tibaldi may be taken to have agreed with the strategies put in place by Mr Ayre, the Manual provided that Mr Ayre was in charge. In order for breaches by Seqwater employees to satisfy s 36(2), Rodriguez must establish that no authority in Seqwater’s position would properly consider Messrs Malone and Tibaldi complying with Mr Ayre’s strategies to be reasonable. We deal with each man’s conduct in turn.

46. [660]

    Mr Tibaldi returned from leave to undertake the night shift. Just over an hour previously, Mr Ayre had issued his situation report, which stated that “[t]he current gate opening strategy will maintain flows of up to 1,600m3/s in the mid-Brisbane River throughout the evening”. It went on to say:

47. [661]

    Thus from Mr Tibaldi’s position, the strategy set by the Senior Flood Engineer had the dam in W1E, with a caution that in the next few days it would be reviewed to consider whether to shift to W2. Mr Tibaldi applied that strategy over his shift, including by opening the gates by two further increments in the early hours of the morning: and in fact dam levels fell slightly during his shift (from 68.65m when he started, to 68.57m when his shift ended).

48. [662]

    Rodriguez’ case at trial was that Mr Tibaldi was obliged to adopt strategy W4. Assuming that “proper modelling” took account of releases which might reasonably have been made, that case was not made out. [190] There was no basis in the material to which this Court was taken for an inference that the standard in s 36(2) was breached.

49. [663]

    On Thursday, 6 January, Wivenhoe levels ranged between 67.29m and 67.45m, rising by slightly less than 1cm per hour throughout the entire day. The rate of increase changed at around midnight. Water levels rose by around 3-4cm per hour for the first 11 hours of Friday, 7 January. In the period from 11:00 until 14:00 on that Friday, water levels rose by 7cm, 6cm and 5cm each hour, and thereafter at a slower rate. The radial gates commenced to open at 16:00 on 7 January, at which time the water level was 68.06m.

50. [664]

    Mr Malone’s original view, communicated by email at 12:14 on 6 January was to open one gate by five increments between 18:00 and 22:00, so as to keep Burtons Bridge open. At 13:30 Mr Malone changed his mind, and said in his “Revised Gate Opening Strategy”:

51. [665]

    By the conclusion of Mr Malone’s shift on 6 January, the water level was merely 67.41m. His situation report at 17:33 stated:

52. [666]

    Even if this involved a failure to apply the Manual, it was not an unreasonable approach to take, and it has certainly not been shown that it was so unreasonable that no dam operator could properly consider it to be reasonable. Throughout the shift, the dam levels were only between 6cm and 16cm higher than the minimum level at which the radial gates could be opened. Even at the end of the shift, almost the entire flood storage of Wivenhoe (more than 95%) remained available. To be fair, this first shift of Mr Malone was not at the forefront of Rodriguez’ case based on s 36(2).

53. [667]

    The second shift, on Friday 7 January, was preceded by Mr Ayre’s situation report at 06:06. The primary judge accepted that in issuing that report, “Mr Ayre was setting an ‘overall strategy’ or ‘general strategy’ for the management of at least part of the flood event to the effect that releases would be delayed until Burtons Bridge was inundated, and then gradually increased over the subsequent 24 to 30 hours to approximately 1200m3/s.”

54. [668]

    Mr Ayre’s situation report stated that “it was proposed that Wivenhoe releases will commence late Friday/early Saturday” (ie late on 7 January or early on 8 January). Mr Malone appears to have brought that forward, by determining at midday (when he issued Directive 1) to commence gate openings at 15:00 on Friday afternoon.

55. [669]

    Rodriguez did not establish that Mr Malone’s implementation of Mr Ayre’s strategy was so unreasonable that no dam operator could properly consider it to be a reasonable thing to do. There was no evidence of that, and it is to be borne in mind that dam levels on that shift were from 67.68m to 68.17m.

56. [670]

    True it is that the primary judge considered that Mr Malone was not excused by that general strategy from determining the applicable Wivenhoe strategy: Ch 6 [262]. So much may be accepted. But it is another thing to conclude that implementing the Senior Flood Engineer’s strategy was something which was so unreasonable that no dam operator could consider it a reasonable response.

57. [671]

    It is also true that the primary judge addressed, in some detail in Ch 6, the likelihood of rainfall over the next four days exceeding 140mm of runoff as follows:

58. [672]

    This analysis has been considered above at [536]-[540]. It depends on the no release assumption applied over a four-day period as well as the coming to pass of a number of other possibilities. It does not address the reality that during the four days dealt with by the PME forecasts, it was anticipated there would be substantial and sustained releases. Once that is appreciated, it does not follow that there was a “serious and significant risk” that implementing the strategy set by Mr Ayre would lead to the need to make large releases days later because levels exceeded 74m.

59. [673]

    Turning to Mr Malone’s third shift, Wivenhoe was releasing around 1330m3/s at the commencement of his shift at 06:30 on Sunday 9 January, in accordance with the strategy of keeping the Mt Crosby Weir and Fernvale Bridges open. The primary judge recorded that Mr Malone made inquiries of the Bureau about the persisting heavy rainfall forecasts, and suggested that there be a discussion with all flood engineers to discuss upcoming operations: Ch 7 [166]. That meeting took place at 15:30.

60. [674]

    Mr Malone sent his email of 11:01 which has earlier been mentioned. It identified the very large run off into Wivenhoe which could be generated by forecast rainfall over the next three days (from 500,000 – 1,000,000Ml), compared the flood event to those of October 2010 and February 1999, and noted that “the starting level of the dams is much higher than in these historical events”.

61. [675]

    The following extract from his cross-examination, noted in Ch 7, was directed to this point in time:

62. [676]

    The primary judge regarded Mr Malone as conceding, based on what he knew at 11:00 on 9 January, that releases should have been increased. We agree. However, neither that concession, nor his Honour’s statement that “the only relevant increase that could be countenanced was one that inundated the remaining downstream bridges” entailed breach of the standard in s 36(2). Again, there was no evidence of what another dam operator would regard as so unreasonable that it could not be considered a reasonable response. Furthermore, our earlier breach analysis (at [602] above) concluded that it was not until the early evening on 9 January that a reasonably competent flood engineer taking account of proposed releases should reasonably have expected that the level of Wivenhoe was likely to exceed 74m, thereby requiring releases to a level above that sufficient to submerge the bridges. Moreover, Mr Ayre was in charge. What was Mr Malone to do? It is unrealistic to expect Mr Malone unilaterally to issue directives which would increase releases so as to inundate the major bridges. Mr Malone appears to have been involved in assembling a meeting, one aspect of which was, evidently, to review the strategy. We are not persuaded that taking those steps, and not in the meantime departing from the strategy which had been in place over the previous days, was so unreasonable that it could not properly be considered reasonable by a dam operator in Seqwater’s position.

63. [677]

    The contemporaneous note of the meeting that afternoon, which the primary judge regarded as accurate and speaking for itself, [191] bears repeating:

64. [678]

    The timing and content suggests that immediately after the conclusion of that meeting, Mr Malone made the telephone call recorded in the following entry in the Event Log, at 16:15:

65. [679]

    That note confirmed that there was a strategy to limit releases to levels to preserve the two major bridges: Ch 7 [200]. It was repeated in a note recording a materially identical communication by Mr Malone to Ipswich City Council at 16:20. A note at 16:27 recorded a similar conversation between Mr Malone and Mr Morris at Brisbane City Council, advising that “[f]low in the Lower Brisbane potentially might reach 3,000 cumecs by next Wednesday or Thursday”. That possibility was reiterated in a file note recording another communication with Brisbane City Council at 17:25.

66. [680]

    Thus, in accordance with contemporaneous documents recording communications with third parties, the strategy adopted at the meeting on the afternoon of Mr Malone’s shift was to keep releases to levels which would preserve Mr Crosby Weir and Fernvale Bridges, that is, around 1400m3/s, but with there being a real possibility that by Tuesday afternoon (11 January) or Wednesday substantially greater releases might occur.

67. [681]

    Mr Malone completed his 9 January shift with his situation report issued at 21:00 on the Sunday evening, in which he advised all (including Mr Morris at Brisbane City Council) that following further inflows:

68. [682]

    Thus, by the evening of Sunday 9 January, Mr Malone had unquestionably moved into Strategy W3. The question is not whether that should have occurred earlier, but rather whether Mr Malone’s failure to take that step earlier was so unreasonable that no dam operator in Seqwater’s position could properly consider his conduct to be a reasonable exercise of flood mitigation functions.

69. [683]

    Breach of this standard has not been made out. Mr Malone adhered to the strategy set by Mr Ayre, abided by the decision reached at a meeting convened by Mr Ayre, and when dam levels continued to rise, altered the strategy to W3. Although this should have occurred before the meeting of flood engineers on the afternoon of 9 January, it was not so unreasonable for Mr Malone to await the outcome of the meeting that no dam operator could properly consider that a reasonable thing to do.

70. [684]

    Undoubtedly the strongest of Rodriguez’ claims related to the operation of the dams in the period immediately preceding the peak releases from Wivenhoe which resulted in major flooding in Brisbane and Ipswich. However, it is important to bear in mind two matters which are only known with the benefit of hindsight:

    1. (1)

       Even if there were zero releases from Wivenhoe, there would have been urban inundation; the peak flows at Moggill without Wivenhoe exceeded 5,000m3/s.

    2. (2)

       Every simulation advanced by Dr Christensen required, in response to the tremendous inflows which took place in the late morning and early afternoon of 11 January (6,817m3/s at 06:00, 9,165m3/s at 09:00, 10,376m3/s at 10:00, 9,606m3/s at 11:00, 10,120m3/s at noon and a peak rate of 11,561m3/s at 13:00) substantial releases from Wivenhoe which contributed to the urban inundation.

71. [685]

    These two matters are well illustrated by the following graph provided by Rodriguez.

72. [686]

    The first point is clear from the lowest line on the graph (dotted black), representing “Moggill Flow without Wivenhoe”. The second may be seen from all of the other lines save the top-most (being the dashed black line representing “actual flows”). Effectively there were two contributing causes that led to urban inundation, that is to say flooding independent of any release from Wivenhoe and inflows into the dams on 11 January which required substantial releases from Wivenhoe. But on 7 January, and for most if not all of 9 January, it was not clear that there would be any urban inundation at all. It was certainly not clear that the issue confronting Mr Malone on 7 and 9 January was how best to minimise the urban inundation which (it is now with the benefit of hindsight known) was to occur on 12 and 13 January.

73. [687]

    For these reasons, Rodriguez could not discharge its onus in making out a contravention of the standard set by s 36(2) in respect of any alleged breach. Seqwater’s appeal with respect to the findings of breach on 6-10 January must be upheld. The finding of liability on the part of Seqwater must be set aside. The remaining grounds do not arise, but the issues may be noted, and, as far as appropriate, determined.

22 Causation (grounds 26, 27)

1. [688]

   As appears from the diagram at [685] above, natural flows at Moggill emanating from catchments below Wivenhoe, that is, without any allowance for water released from Wivenhoe, were above 4,000m3/s from about 22:00 on 11 January to about 22:00 on 13 January. Those downstream flows reached a peak of some 5,800m3/s at about 00:00 on 13 January.

2. [689]

   On the non-negligent hypothesis adopted by the primary judge (simulation C), the flows at Moggill would have substantially exceeded 4,000m3/s over a period from early afternoon on 11 January until about 23:00 on 13 January. In the actual event, there was a greater level of inundation with flows above 4,000m3/s from about 09:00 on 11 January until about 03:00 on 15 January, peaking at 10,700m3/s at in the early hours of 12 January. (The peak flow reached the city some hours later.) The factual question of causation turned upon the difference between the non-negligent and actual flows.

3. [690]

   The plaintiff at trial relied upon modelling prepared by Dr Altinakar to demonstrate the level of flooding caused by the negligence of the flood engineers. Ground 27 in the notice of appeal asserted that the judge had erred in relying on that modelling to support findings that the property of Rodriguez and certain group members would have been inundated had the flood engineers conducted operations substantially in accordance with simulation C. Dr Altinakar also prepared models for the other simulations, including F and H. In the event, ground 27 was not pressed.

4. [691]

   The only ground relied upon under the heading “Causation” (ground 26) alleged that the judge had erred in finding that the breaches of each of the flood engineers were sufficient to account for the occurrence of the particular harm suffered by the first respondent and other selected group members, again based on the difference in outflows between the actual events and simulation C. Putting to one side the relevant outflows to be compared, the underlying proposition was that a breach on a particular day by a particular flood engineer could not account for a specific level of harm at a particular property. There was some overlap between this ground and ground 28 under the heading “Apportionment”. However, apportionment turns on a number of assumptions which are theoretical, given the findings above as to the engagement of s 36(2). There is little immediate purpose in considering the extent of the responsibilities of the individual flood engineers, in circumstances where SunWater and the State (the employers of Mr Ayre and Mr Ruffini respectively) were not involved in the appeal before the Court. Nevertheless, because the ground must be rejected those problems may be put to one side.

5. [692]

   As Seqwater correctly submitted, the starting point in assessing questions of causation is s 11 of the Civil Liability Act, which provides:

6. [693]

   In considering the question of breach, it is necessary to have regard, prospectively, to a particular “risk of harm”; questions of causation are viewed retrospectively by reference to the relationship between any breach of duty which has been established and the “particular harm” which eventuated. The particular harm in the present case may be identified as the damage caused by inundation to the property of Rodriguez. The inundation occurred as a result of the flows in the Brisbane River exceeding a particular rate, for a period of time.

7. [694]

   Further, assuming that the appropriate non-negligent program of releases was that identified in simulations F and H, the peak flow at Moggill reached approximately 8,300m3/s. Accordingly, the owner of property at a sufficiently low level to be inundated by those flows could not allege that any breach of duty caused the damage suffered. As the actual release rate peaked at about 10,700m3/s, [192] on the same hypothesis the owners of properties which suffered inundation as a result of the flows exceeding 8,300m3/s were, on Rodriguez’ case, entitled to recover their losses.

8. [695]

   The primary judge approached the question of causation on the basis that there had been several breaches of duty over the period from 2 January until 10 January 2011. These breaches involved failures on the part of the flood engineers to release water from Wivenhoe at an appropriate rate prior to 10 January. The judge accepted that the losses suffered had been caused by the cumulative effect of the several breaches. In doing so, he applied a passage in the joint reasons of the High Court in Strong v Woolworths Ltd, [193] dealing with the equivalent provision, s 5D, in the New South Wales Civil Liability Act:

9. [696]

   Seqwater contended that there were difficulties in applying this reasoning to the present case which, on the primary judge’s findings, involved a series of sequential breaches by different parties. Its point may, perhaps, be illustrated by reference to two examples, based on an assumption that there were seven breaches each contributing an increased flow of 300m3/s. First, the owner of a property which was only inundated when the peak flow exceeded 10,400m3/s could legitimately argue that all the breaches of duty were necessary, in their cumulative effect, to cause damage to that property. In that sense, each breach was an element of a set of conditions, all of which were necessary and sufficient for the occurrence of the particular harm. (What was meant by the reference in Strong to “more than one set of conditions” may be disregarded; it is the concept of a “set” involving several elements which is critical.) A second example might be a property which was inundated when the peak flow reached 8,600m3/s. If the non-negligent flow reached 8,300m3/s and each breach contributed an increased flow of 300m3/s, only one breach was necessary to cause the harm to that particular property. Accordingly, as we understand the appellant’s submission, it cannot be said that all of the breaches were “necessary” to cause that harm and, accordingly, it cannot be said that Seqwater was responsible for the damage to that property. If there were merit in viewing the breaches sequentially, the loss might be attributable to the act of the flood engineer who was on duty at the time of the first breach.

10. [697]

    To present the problem in this way demonstrates a difficulty in so identifying breaches. It is true that Rodriguez pleaded breaches of duty by reference to separate steps taken on a daily basis and, understandably, that approach was followed in the course of the trial and in the judgment. However, the approach is artificial. It involved the dividing of a singular course of conduct into discrete temporal segments. Further, it assumed that each flood engineer could and should from time to time exercise independent judgment. In fact, the decision-making of the flood engineers was more constrained than this approach recognised. As has been explained, for the most part the flood engineers were acting in a collaborative manner in setting a course for dealing with the flood event. Secondly, while it may have been appropriate and indeed necessary for the planned operation to be varied as conditions changed, any change required evaluation from a new starting point and could not achieve large effects immediately. For example, to increase significantly the volume of water being released from Wivenhoe it was necessary to open gates according to a protocol established by the Manual. If there were a legitimate criticism of the engineers for not releasing more water more quickly when the flood event was declared in the early morning on 6 January 2011, to the extent that the flood engineers operated collaboratively, all (and vicariously their employers) were liable for each breach. The fact that a particular engineer was on duty at a particular time was not a critical factor. What was important was the time at which (or period over which) the engineers breached the appropriate standard of care, and the time at which the window of opportunity for pre-releases to create flood storage volume closed. Section 11 did not require the cutting and dicing of a particular course of conduct to determine separate specific breaches which could be attributed to the damage caused at particular properties.

11. [698]

    It follows that Seqwater’s complaint as to the way the primary judge dealt with the issue of causation must be rejected.

12. [699]

    That conclusion does not provide an answer to the question of apportionment. Apportionment between tortfeasors was required by Pt 2 of the Civil Liability Act and was to be assessed on a “just and equitable” basis: s 31(1)(a). It may be possible to identify different levels of culpability as between the flood engineers and those vicariously liable for their conduct. The approach to causation does not address that issue.

23. Apportionment (ground 28)

1. [700]

   The primary judge undertook an apportionment of liability as between the defendants on the basis that s 28 of the Civil Liability Act was engaged. That section stated:

2. [701]

   Instead of the usual principle that two or more wrongdoers are jointly and severally liable for the whole of the loss suffered by the plaintiff, where s 28(1) is engaged, each will be a “concurrent wrongdoer”, responsible only for the proportion of the loss determined by the court as just and equitable, pursuant to s 31(1)(a). There is no right of contribution between concurrent wrongdoers: s 32A. [196]

3. [702]

   Consistently with s 28(4), s 30 provides:

4. [703]

   There appears to have been no contest at trial that the claims were apportionable claims within Pt 2 of the Civil Liability Act, which is headed “Proportionate liability”. In Rodriguez (23) the primary judge stated that it had been found in Rodriguez (22) that the plaintiff’s claim in negligence was an “apportionable claim” within the meaning of s 28(1). [197] The judge determined that it was “highly likely” that the relevant proportionate liability provisions were those contained in the Queensland Civil Liability Act and not the New South Wales statute: Ch 14 [85]. The judge then identified the definition of “apportionable claim” in s 28(1), observing that, “[t]here are certain exclusions in s 28(3) and 28(4) but they are not relevant.” There was no reference to s 30.

5. [704]

   Perhaps because Rodriguez did not dispute that the apportionment pleaded by the defendants should be undertaken, there was no consideration of the effect of s 28(4) and s 30(1). Nor was it noted that, in this respect, there was a significant difference between Pt 4 of the Civil Liability Act (NSW) and Pt 2 of the Queensland Act. There is no equivalent in the New South Wales Act to the exclusion contained in s 28(4) of claims where, by statute, liability is “joint and several.” Further, whereas the definition of “concurrent wrongdoer” in s 30(1) involves two or more persons whose actions caused “independently of each other” the loss or damage the subject of the claim, the New South Wales Act covers a person who is one of two or more persons whose acts or omissions caused “independently of each other or jointly” the damage or loss the subject of the claim. [198] That Seqwater was the sole licensee entitled to release water from Wivenhoe Dam, and that SunWater and the State were only involved because their employees operated as flood engineers in combination with Seqwater’s employees, casts real doubt on whether apportionment was appropriate or even available. [199] The difference between the legislative schemes in this respect appears to be fundamental.

6. [705]

   The judge apportioned the loss and damage suffered by Rodriguez by reference to each of the flood engineers, namely Mr Ayre (30%), Mr Malone (37.5%), Mr Ruffini (20%) and Mr Tibaldi (12.5%): Rodriguez (23) at [112]. The result was that Seqwater, being liable for the acts of Messrs Malone and Tibaldi, bore 50% of the loss and damage.

7. [706]

   The primary judge was confronted with several differing positions as to the proper method for assessing culpability, as between the defendants. These included an attempt to assess which flood engineer was responsible for releasing particular volumes of water in excess of those deemed reasonable under simulation C. The judge opted, however, for a simpler formula, which turned primarily on the number of shifts served by each engineer during the relevant period.

8. [707]

   Subject to one qualification, Seqwater did not seek to depart from this approach. Its primary submission turned on the number of relevant shifts. That in turn depended upon the degree of success with respect to the challenges to the judge’s findings as to breaches of duty. Thus, it was submitted that if the breaches on 2, 3, 4, 5, 6 and 7 January were set aside, there would have been a significant reduction in Mr Malone’s contribution and hence the apportionment to Seqwater. [200] The qualification was that the judge had given insufficient weight, in Seqwater’s submission, to the greater responsibility of Mr Ayre as senior flood engineer. [201] Seqwater recognised that if all its challenges were upheld and it was not in breach, the Court would not need to consider apportionment. [202]

9. [708]

   An issue was raised in the course of submissions as to whether any finding on apportionment would bind the other defendants, namely SunWater and the State, both of which had settled with the plaintiff and did not participate in the hearing of Seqwater’s appeal. However, that concern may be put to one side; the Civil Liability Act permits an apportionment of liability where not all concurrent wrongdoers are parties to the proceedings: s 31(3) and (4). Further, the Act makes specific provision with respect to separate claims against concurrent wrongdoers: s 32B, s 32C(2).

10. [709]

    Because the Court has concluded that Seqwater is not liable to the plaintiff or group members, it is not appropriate to address the question of apportionment on a contingent basis, where the contingency is undefined. In other words, apportionment will depend upon the basis and the extent of any liability which is to be assumed, contrary to the findings set out above. The inappropriateness of pursuing the issue is reinforced by the doubt that the defendants were in fact “concurrent wrongdoers” within s 30(1) of the Civil Liability Act, and the absence of submissions by either party on that provision.

11. [710]

    It is, however, convenient to note one consideration which might have rendered it inappropriate to reduce Seqwater’s liability below 50%. Although the parties did not address the issue either in this Court or before the primary judge, as has been noted above, Seqwater was the sole party responsible as licensee for controlling the release of water from the dams into the Brisbane River. Whatever arrangement it had with third parties, including SunWater and the State, to engage flood engineers employed by them, it is difficult to envisage that its liability would be less than 50%.

24. Damages for cost of cleaning by volunteers (ground 31)

1. [711]

   Ground 31 concerned the calculation of the reasonable commercial cost of cleaning and reinstating damaged property following the flood. It was formulated as follows:

2. [712]

   On the basis that Seqwater was not liable to the plaintiff or the group members, this ground does not arise. Whether or not the Court should address the issue will turn on principles of judicial economy articulated by the High Court in Boensch v Pascoe. [203] The effect of the statements in Boensch was to dilute the convention established in Kuru v New South Wales, [204] by which intermediate courts of appeal considered whether to deal with grounds which were not dispositive, given conclusions reached on another ground or grounds.

3. [713]

   In favour of disposing of the issues raised and argued, it may be said that there is an issue of principle involved which falls within a narrow compass and is therefore readily resolvable.

4. [714]

   On the other hand, there are three factors which diminish the appropriateness of addressing this ground. First, it is entirely discrete and could, if necessary, readily be dealt with in the event that the High Court set aside this Court’s findings on liability. Secondly, it is not apparent that there is any practical utility in addressing the issue in this case. The proceedings involving the other two defendants have been settled; while there may be outstanding issues as to the distribution of any payments resulting from the settlement, it is unlikely that the resolution of the present issue will affect those steps. Thirdly, for reasons which warrant some further explanation, the issue is not easily defined in terms which will readily resolve any contingent factual disputes.

5. [715]

   The first factual complexity lies in the concept of “cleaning costs”. The relevant principles have more commonly been stated in cases involving repairs for damaged chattels or fixtures on land. As explained by McPherson J in Davidson v J S Gilbert Fabrications Pty Ltd: [205]

6. [716]

   There are also cases where a repaired chattel may not attain its full pre-damage value, in which case the cost of repairs, though reasonable in themselves, may not cover the full loss suffered by the owner. In some cases, cleaning costs of a chattel may exceed the value of replacement by a comparable item; a similar possibility may arise in relation to fixtures damaged by water. The value of cleaning costs will undoubtedly be a significant element in the calculation of loss, but it may not equate to the loss suffered by the property owner.

7. [717]

   Any dispute that the owners were entitled to recover compensation for the cleaning services which they or volunteers had undertaken gratuitously was resolved in Rodriguez (22) and was not raised by the appeal. The issue addressed by ground 31 was how those services were to be valued. As explained by the primary judge in Rodriguez (27), [207] the different approaches were identified as follows:

8. [718]

   It may be seen that, as a matter of principle, both approaches used a market price, or commercial quotation, for undertaking the work. That there were different methodologies underlying the calculations does not appear to involve any question of law. The exercise was no doubt difficult and there were practical considerations affecting how it should be approached. Seqwater’s concern was apparently about not paying for work which had not been done. However, this had two mutually inconsistent elements. On the one hand, Seqwater contended that it should not have to pay for professional cleaning services which were not in fact provided; on the other hand, it should not have to pay at an hourly rate for work which was done inefficiently or unnecessarily.

9. [719]

   After explaining and applying the reasoning of the Victorian Court of Appeal in Powercor Australia Ltd v Thomas, [208] and noting arguments which had been considered and rejected in that case, the judge stated:

10. [720]

    Whether the point of distinction sought to be articulated in this passage was valid would depend upon particular circumstances. That may be illustrated by reference to the circumstances and the reasoning in Powercor upon which the primary judge relied.

11. [721]

    In Powercor, propositions which were not in dispute in the present case were at the heart of the appeal. Thus, one issue was whether the farmer (Mr Thomas) whose farm had suffered significant damage as a result of a fire which commenced through the negligence of the appellant, could recover the cost of repairing fences himself. Osborn JA stated:

12. [722]

    Powercor dealt with work done by volunteers, in the following terms:

13. [723]

    The judgment in Powercor also referred to the principles applied by the High Court in Zheng v Cai, [212] a case in which payments made by a religious organisation for whom a passenger injured in a car accident did voluntary work were, as the High Court held, not to be taken into account in assessing damages for economic loss. [213] That led to two propositions. First, consistently with the approach in Zheng v Cai, “the gifts of voluntary labour received by Thomas should be disregarded from the assessment of damages.” Secondly, the contention that the case fell to be determined by reference to the principles stated in Griffiths v Kerkemeyer was rejected. [214]

14. [724]

    Both propositions relate to personal injury cases; one is applied, the other not. As to the first proposition, in the present case the relevance of amounts paid gratuitously to victims of the floods by the State was separately considered and rejected. The question is whether services gratuitously provided to meet a specific compensable need should be dealt with on the same basis. The payments of money in relief of suffering, as in National Insurance Co of New Zealand Ltd v Espagne [215] and Redding v Lee, [216] are “subventions” to adopt the terminology of Fullagar J in Blundell v Musgrave. [217] The reasoning of the majority and the minority (Dixon CJ and Fullagar J) in Blundell v Musgrave was not in conflict and, as noted below, has recently been approved by the High Court. The plaintiff, Musgrave, was a member of the Naval Forces, who was injured when hit by a motor vehicle driven by Blundell. The appeal concerned the liability of the defendant for the cost of medical treatment provided to the plaintiff at naval facilities, for which the Navy sought to charge him. The liability of the defendant for those costs turned on the likelihood of the Navy recovering them from the plaintiff. Fullagar J referred to two cases, one of which involved a domestic servant who, prevented by injury from remaining in her employment, received free board and lodging following the accident from her father. The second case involved a plaintiff who had lost wages during a period of disability but had received by way of pension and sick pay amounts equivalent to the wages. In each case the plaintiff was allowed to recover the amount in dispute. Fullagar J observed: [218]

15. [725]

    In relation to the second proposition, in Van Gervan v Fenton, [219] a case involving gratuitous domestic care, the High Court rejected the claim that a care provider’s loss was an appropriate measure of the injured person’s loss. The joint reasons of Mason CJ, Toohey and McHugh JJ stated: [220]

16. [726]

    However, the basis for recovery in Griffiths v Kerkemeyer was noted in CSR Ltd v Eddy, [221] Gleeson CJ, Gummow and Heydon JJ stating that the case was “anomalous in departing from the usual rule that damages other than damages payable for loss not measurable in money are not recoverable for an injury unless the injury produces actual financial loss.” The three available heads of recovery were identified as general damages (payable for non-pecuniary losses, including pain and suffering), loss of earning capacity and actual financial expense (including medical, hospital and related expenses). The latter two categories have been characterised as economic loss, and payments as “special damages”. It was further noted in CSR that the traditional approach to compensable loss was that stated by Dixon CJ in Blundell v Musgrave, with whom Fullagar J agreed, the majority accepting the same principles, but differing only in their application. [222] As McHugh J observed, Griffiths v Kerkemeyer turned on acceptance that a plaintiff’s “injury-caused needs”, resulting from the tortfeasor’s negligence, should not lead to unrecoverable loss. [223] He continued:

17. [727]

    It is therefore somewhat unclear to what extent, beyond the specific (anomalous) awards made pursuant to Griffiths v Kerkemeyer, a plaintiff can recover the cost of services for which no payment has been made, no obligation to pay incurred and for which no payment will be required in the future. On one view, the proper analysis in the present circumstances is dependent upon whether the principles governing charitable gifts, as applied in Zheng v Cai, [224] apply to the gratuitous provision of services to assist those in need. If they do, is it necessary to ask whether the benefit was conferred independently of any right or redress the plaintiff might have against others, so that the recipient might enjoy the benefit even if damages were obtained? [225]

18. [728]

    This point was noted by Handley JA in Screenco Pty Ltd v R L Dew Pty Ltd: [226]

19. [729]

    There was no challenge in this Court to the proposition that the plaintiff and group members were entitled to recover cleaning costs provided by members of the public without charge. However, the legal basis upon which such services are recoverable may affect the appropriate method of valuation. The uncertainty noted by Fullagar J as to the true basis for such claims has not been resolved in the last 65 years. They do not fall within the tripartite classification accepted in CSR v Eddy. Uncertainty in this regard provides a further reason for this Court not to address the question raised in circumstances where it does not need to.

20. [730]

    The issue sought to be raised by Seqwater, however packaged, is little more than a dispute as to the best means to quantify that element of compensation. There would be no benefit, and significant difficulties, in seeking to resolve that issue further in the present circumstances.

25. Pre-judgment interest on damages

1. [731]

   As with ground 31, there is a live issue as to whether it is appropriate to address the two related issues which are covered by this topic. However, both the issues and the context were different from those discussed in relation to ground 31, as will be explained below.

2. [732]

   Two separate but interrelated questions arose with respect to claims for pre-judgment interest on damages. The first was whether the primary judge erred in not awarding interest on damages for cleaning of fixtures and fittings, undertaken by volunteers, but calculated by reference to the commercial cost of the labour. This was the subject of Rodriguez’ application for leave to appeal. The second was whether the judge was correct in awarding interest on that proportion of the damages equivalent to the amount received in Round 3 funding from the Premier’s Disaster Relief Fund. This was the subject of ground 32 in Seqwater’s appeal.

3. [733]

   The plaintiff’s claim for interest was pleaded as “interest in accordance with s 100 of the Civil Procedure Act 2005 (NSW)”. The pleading raised two issues. First, it assumed that interest was available, if at all, pursuant to New South Wales law: it is necessary to consider whether that was correct, or whether the claim was properly to be assessed under Queensland law. Secondly, what was sought was interest “on damages”, pursuant to statute; there was no claim for interest “as damages” with respect to tortious injury to property. A claim in the latter terms would have raised an issue as to the application of Sempra Metals Ltd v Inland Revenue Commissioners [227] in Australia. As noted in McGregor on Damages, [228] both Lord Nicholls and Lord Scott held that a claim for interest as common law damages for loss could arise in tort: at [100] and [132] respectively.

4. [734]

   It is necessary first to identify the applicable law. To the extent that an award of pre-judgment interest depends on statute, both Queensland and New South Wales have statutory provisions in broadly similar terms. Section 58 of the Civil Proceedings Act 2011 (Qld) relevantly provides:

5. [735]

   Each of these provisions is in the form of a conferral of power on a court giving judgment for a payment of money, a power which has been described as “discretionary”. [229] However, the discretion is not open-ended: the purpose for which interest is awarded is compensatory and where matters have been proved to the satisfaction of the court warranting a grant of interest, a grant will be obligatory, in the proper exercise of the discretion, in order to fully compensate the victim of a tort or other wrong. Thus, although in form each provision confers a power on a court (which should be understood as a court of Queensland and New South Wales respectively), in substance it is a conferral of an entitlement to a payment in the nature of compensation. Each provision has the dual function of creating a contingent right in the victim of a wrong, and conferring on a court the power to provide the remedy.

6. [736]

   An award of interest being compensatory, it follows that the right to an award is a matter of substance governed by the lex loci delicti, consistently with the statement in John Pfeiffer Pty Ltd v Rogerson, [230] that “all questions about the kinds of damage, or amount of damages that may be recovered, would … be treated as substantive issues”. The relevant provision is therefore s 58 of the Queensland Act. (As discussed earlier in these reasons, this Court’s jurisdiction does not depend on s 4(1) of the Jurisdiction of Courts (Cross-vesting) Act 1987 (Qld), and s 11 of that Act is not engaged, though if it were, it would produce the same result.)

7. [737]

   Given the compensatory purpose of an award of interest, in each case where interest is sought it will be necessary to have regard to two particular considerations. The first is that, as more expansively stated in s 100(2) of the New South Wales Act, there may be good reason not to require payment of interest where a pre-judgment payment has been made of “the whole or part of the debt or damages” the subject of the proceedings. This raises the question identified by Seqwater, namely whether it was correct in principle to disregard the payments made to victims of the flooding from the State relief fund in awarding interest.

8. [738]

   Secondly, it will be necessary to consider whether payment of interest would involve double recovery. That might occur, for example, where a loss which accrued in 2011 is compensated by a payment of damages assessed in terms of current values at the date of judgment where interest in part allows for losses due to inflation. [231] It might also occur where a revenue-producing asset has required repair and been out of service, the owner being compensated by recovery of the cost of repairs (with interest from the date of payment) and an award of damages for lost revenue during the period before the repairs were completed. Accordingly, the issues of principle sought to be raised cannot be answered in terms which disregard the context in which they come to be applied.

9. [739]

   The limitations on the value of general rulings may be demonstrated by the lack of final resolution of the rulings with respect to interest in Rodriguez (23), even as applied to the circumstances of the individual cases under consideration. Thus, in Rodriguez (27), delivered on 26 February 2021, the primary judge returned to answer questions about interest on damages where payments had been made by the Queensland Rural Adjustment Authority and the Natural Disaster Relief and Recovery Arrangements, in relation to two particular group members. The issue arose partly in relation to whether those payments were intended to be enjoyed independently of, and cumulatively upon, any right to recover interest in respect of the loss occasioned, and “how a claim for statutory interest would interact with a subsidised loan”: at [84] and [87].

10. [740]

    There are, therefore, considerations which weigh against dealing with the question of pre-judgment interest. On the other hand, the issues raised by the parties are formulated in terms which identify two specific criteria as determinative for or against the award of interest. It is possible as a matter of principle to address those questions, although not in terms which would necessarily provide a common outcome for all group members, had liability been established.

11. [741]

    Rodriguez sought leave to appeal, whilst submitting that leave was not required. The issue turned on the amount of interest claimed on the award for cleaning services being less than $100,000, thus engaging the requirement for leave in s 101(2)(r) of the Supreme Court Act 1970 (NSW). Its answer was that because the issue would affect many class members, including the representative sample of three class members whose damages were assessed by the primary judge, the amount in issue would in practical terms certainly exceed $100,000.

12. [742]

    That submission may be accepted, but it demonstrates that the requirement for leave lies elsewhere, namely in the fact that the judgment obtained by Rodriguez was interlocutory, for the reasons identified in relation to Seqwater’s appeal. The proceedings have not been finally disposed of by the answering of separate questions in a manner giving rise to liability, but not resolving all issues of damages. On one view, the fact that the issue is now moot in the light of the findings as to liability militates against a grant to Rodriguez of leave to appeal. On the other hand, if the Court were otherwise minded to address the issue, the need for leave to appeal should not stand in the way of that outcome.

13. [743]

    The issue sought to be raised was addressed in Rodriguez (23). The parties had agreed damages of $200,968, comprising $101,517 for trading and inventory loss, $10,377 for loss of fixtures and $89,074 for “the cost of cleaning and repairing the plaintiff’s shop, fixtures and stock after the flooding”, which was in fact “undertaken for no charge by a director of the plaintiff, Mr Rodriguez, members of his family and community minded volunteers”: at [8]. The defendants contended that interest should not be awarded on the third component of the damages, characterised as “gratuitous services”, even if that characterisation did not affect the plaintiff’s entitlement to damages.

14. [744]

    The judge observed that an award of interest was intended to compensate a plaintiff for having been kept out of money due between the time of the wrongful act and the time of judgment: at [20]. After considering whether the effect would be to confer a windfall benefit on the plaintiff, he concluded that to award interest “would not be consistent with the compensatory basis on which interest is awarded”: at [33].

15. [745]

    Scattered through the decided cases, and the submissions of the parties in this case, are rhetorical statements reflecting these alternative positions. Thus, defendants generally allege that to award damages calculated as the commercial value of services provided gratuitously may be seen as a “windfall” which does not compensate for any financial loss; to award interest on top of those damages would be to exacerbate the nature of the windfall. Plaintiffs generally argue that such services were not intended to relieve defendants of their liability to pay full compensation at commercial rates, so that to decline to award interest on the payment would equally be to relieve a defendant of part of its liability. Both submissions are consistent with the compensatory principle, but neither assists in resolving the dilemma of requiring a defendant to pay interest on a financial outlay that was not in fact incurred.

16. [746]

    The difficulty in providing a coherent and principled answer lies in the disparate conclusions reached by the courts in the past. The variable factors may be grouped as follows:

17. [747]

    Some common areas of concern have been removed by statutory provisions. In Queensland, a court cannot award interest on general damages or damages for gratuitous services provided to an injured person, in respect of personal injury. [232] In New South Wales, that prohibition extends to damages awarded for the loss of a claimant’s capacity to provide gratuitous domestic services to his or her dependants. [233] Thus, the decision in Grincelis v House [234] requiring an award of interest on damages for past gratuitous care services no longer operates in Queensland (or in this State). The principle may, of course, have broader operation. However, it is by no means clear that that is so. The Court in Grincelis referred to the adoption in Gogic of the compensatory principle. [235] That in turn was identified by reference to the reasoning of Gibbs CJ in Batchelor v Burke. [236] However, the question in Batchelor was whether interest was payable on so much of an award of damages as related to loss of earning capacity where the earnings lost before trial had been replaced by payments of worker’s compensation. Gibbs CJ referred in the course of his reasons to the principle established in Fire & All Risks Insurance Co Ltd v Callinan [237] that, in the case of loss of earning capacity, interest should be allowed “only on that part of the damages awarded under that head which represents compensation for those detriments the practical impact of which, in terms of economic loss actually incurred, has already, at the date of judgment, been experienced by the plaintiff.” [238] There was nothing in Grincelis to cast doubt on those statements, rather the contrary.

18. [748]

    Whether cases involving personal injury resulting in loss of earning capacity, or the ability to care for oneself, raise different principles from those applicable to loss of use of chattels, and whether in the case of chattels, different principles apply with respect to those used for commercial and non-commercial purposes, are questions not readily answered in the abstract. In Screenco Pty Ltd v R L Dew Pty Ltd, [239] to which reference has already been made, this Court reviewed a judgment of McClellan J refusing to award interest on the loss of a large outdoor screen used in sporting events, the screen having been destroyed by the negligent construction of the scaffolding supports. While there was no doubt that Screenco, as the owner, was entitled to the value of the screen, a dispute arose as to whether it was entitled to interest on that amount from the date of the loss of the screen to the date of judgment. The case was unusual because Screenco had not paid for the screen, which it had purchased from its parent company. There was no evidence that it was required to make a payment to its parent at any particular date, nor that interest was payable on the outstanding amount. In addition to the cases in the High Court referred to above, Handley JA referred to a number of Admiralty cases which “are important because they recognise that interest and loss of earnings are alternative bases for assessing compensation and both should not be awarded for the same period”: at [30]. He also referred to cases in which motorists who had suffered loss of the use of a vehicle were entitled to recover damages reflecting the loss of opportunity to use the vehicle, but where substitute vehicles had been made available free of charge, no interest being payable on that head of damages. Handley JA quoted a passage from the reasons of Lord Mustill in Giles v Thompson [240] to the following effect: [241]

19. [749]

    The result in Screenco was explained as follows:

20. [750]

    The primary judge, dealing with a loss of chattels or damage to property, considered he should apply the reasoning in Screenco. There was no error in taking that approach; indeed, it would have been wrong not to. It follows that Rodriguez cannot succeed in the present case unless it establishes that Screenco was wrongly decided in a matter of principle. Grincelis, upon which Rodriguez relied, was decided before Screenco and was not understood to be inconsistent with the approach adopted in the earlier High Court cases which were themselves applied in Screenco.

21. [751]

    The submissions made in this Court were quite limited in their scope. It is clear that the claims for interest will vary in a number of respects between members of the group: some will have claims for cleaning costs involving commercial property, others with respect to the cleaning of residential property. It is inappropriate for this Court to address the issues of principle in circumstances where, on the finding as to liability, those issues do not arise. The appropriate course in such circumstances is to refuse Rodriguez leave to appeal.

22. [752]

    The second aspect of the claim for interest arose from the award by the primary judge of interest on damages to the extent that claimants had obtained funding from the Premier’s Disaster Relief Fund (PDRF). Ground 32 of the Seqwater’s further amended notice of appeal was in the following terms:

23. [753]

    The primary judge decided this point chiefly on the papers, shortly before the hearing of the appeal: Rodriguez & Sons Pty Ltd v Queensland Bulk Water Supply Authority t/as Seqwater (No 29). [242]

24. [754]

    The PDRF was a charitable trust that solicited and received donations from members of the public in order to provide grants to relieve people in distress following the flooding in late 2010 and 2011. There were three rounds of PDRF funding, but only the third was relevant, pursuant to which grants were made to persons who had suffered structural damage to their homes. The primary judge concluded (at [14]) that “the circumstances of those payments indicated they were made to alleviate hardship and they were intended to be received by flood victims ‘in addition to whatever rights [they] may have to recover’ elsewhere”, citing National Insurance Co of New Zealand Ltd v Espagne. [243] Accordingly, the benefits obtained by group members from Round 3 of the PDRF were not to reduce the amount of damages to which group members were entitled.

25. [755]

    Seqwater did not challenge this finding. Rather, it submitted that the benefits obtained from the PDRF should nevertheless operate to reduce the interest awarded to group members, relying on the decision in Rodriguez (23), discussed above. However, the judge observed that Rodriguez (23) rested on the correspondence between the services provided and the loss suffered. He continued:

26. [756]

    The essence of Seqwater’s submission was as follows: [244]

27. [757]

    Seqwater thus called upon the principle espoused in Screenco. It is true that Screenco turned on its very particular facts: as was explained by Allsop P in CHEP Australia Ltd v Bunnings Group Ltd [245] and Bunnings Group Ltd v CHEP Australia Ltd, [246] the ultimate question was whether or not the award of interest would over-compensate the plaintiff.

28. [758]

    The same “governing principle” applicable to awards of personal injury damages which Windeyer J in Espagne identified in the cases involving gratuitous provision of third party assistance to an injured plaintiff should ordinarily apply also by analogy to the discretionary power to award interest. Accordingly, absent special circumstances, “benefits that a plaintiff has received or is to receive from any source other than the defendant are not to be regarded as mitigating his loss, if … they were given or promised to him by way of bounty, to the intent that he should enjoy them in addition to and not in diminution of any claim for damages”. [247]

29. [759]

    Subject to the issues of principle in relation to when and on what basis subventions operate to reduce damages, the approach of the primary judge accorded with current practice. There is no benefit in addressing this ground further.

26. Costs of trial – apportionment (ground 30)

1. [760]

   Ground 30 of Seqwater’s appeal separately challenged the apportionment, as between the three defendants, of their joint and several liability to pay the plaintiff’s costs. Notwithstanding a suggestion by Rodriguez that leave was required, Seqwater is entitled to advance this ground as of right: Housman v Camuglia. [248] In any event, it involved a substantial amount of money, and raised a question of principle, namely, the relationship in a case of apportionable claims between the effect of the apportionment legislation upon those claims and the costs discretion.

2. [761]

   In Rodriguez & Sons Pty Ltd v Queensland Bulk Water Supply Authority t/as Seqwater (No 24) [249] it was ordered that the three defendants each pay the plaintiff’s costs of the proceedings (with some presently irrelevant exceptions). Additionally, the primary judge made the following declaration:

3. [762]

   The primary judge thus apportioned liability for costs amongst the defendants in the proportions for which they had been found liable in damages. His Honour proceeded on the basis that the source of power was either s 7 of the Law Reform Act 1995 (Qld) or s 98 of the Civil Procedure Act 2005 (NSW) and that “[t]hese sources of power, or at least s 98, are supplemented by the Supreme Court Act 1970, and that which follows from the Court’s status as a superior court of record, to enable the Court to grant the appropriate relief to reflect a determination of the amount of each defendant’s contribution towards the plaintiff’s costs”. [250] Because none of the plaintiff’s costs had at that time been paid, “the appropriate form of relief is a declaration as to the defendants’ respective obligations to each other”.

4. [763]

   The primary judge regarded the starting point as being that “a tort was committed for which the three defendants are responsible in proportions of 50%, 30% and 20% respectively”. This informed the “extent” of a “person’s responsibility for the damage” under s 7 of the Law Reform Act and also the discretion under s 98(1)(b) of the Civil Procedure Act. His Honour then rejected the submission that the relative proportions should be adjusted by reference to the conduct of the litigation. The dispositive reasoning was:

5. [764]

   Because Seqwater has succeeded in setting aside orders as to its liability, orders as to the costs of the trial must be set aside and this ground does not arise. What follows proceeds on the basis that the primary judge’s apportionment of liability between Seqwater, SunWater and the State survives. That was the basis upon which the ground was argued.

6. [765]

   The first question is to identify the power being exercised. Section 6 of the Law Reform Act 1995 (Qld) confers a right of contribution between tortfeasors, and s 7 identifies that the contribution is to be “just and equitable” as determined by a court. The source of the defendants’ shared liability to the plaintiff was the Court’s order made on 20 October 2020. That order conferred a partial indemnity for legal costs incurred by the plaintiff in proceedings commenced and prosecuted in the Supreme Court of New South Wales. Until and unless an order for costs was made, there was no liability for any defendant to pay any part of the plaintiff’s assessed costs.

7. [766]

   In these circumstances, the Law Reform Act 1995 (Qld) was not applicable. That is not because the Queensland Act was incapable of applying to the liability to pay a plaintiff’s costs. It is established that the right to recover statutory contribution pursuant to statutes in the form of ss 6 and 7 of the Law Reform Act (which enact the substance of s 5 of the Law Reform (Miscellaneous Provisions) Act 1946 (NSW)) extends to the costs payable to the plaintiff in addition to the damages: see James Hardie & Coy Pty Ltd v Wyong Shire Council; [251] South West Helicopters Pty Ltd v Stephenson (No 2). [252] (Seqwater made a formal submission to the primary judge that James Hardie was wrongly decided, but that submission was not renewed in this Court.)

8. [767]

   Rather, the Queensland statute is inapplicable because the question is not determined by the law of Queensland. The services which generated the plaintiff’s costs were, at least in large measure, performed in New South Wales in order to conduct litigation in the Supreme Court of New South Wales. The closest connection between the costs order and any Australian jurisdiction was with New South Wales. In the event of dispute about the quantification of those costs, the order would fall to be assessed under the New South Wales assessment regime. The power to order costs is quintessentially procedural, rather than substantive, as Miles CJ observed in Labuda v Langford. [253] Professor Garnett has stated that the position is reasonably settled in Commonwealth countries. [254]

9. [768]

   Separately from the above, Seqwater submitted that s 32A of the Civil Liability Act prevented the Law Reform Act applying as between concurrent wrongdoers. Section 32A provides:

10. [769]

    Seqwater submitted that a declaration that it was liable to contribute to costs payable by a co-defendant contravened this provision. There is a large question as to whether, par (a) being limited to contribution to “damages”, the indemnification referred to in par (b) is also so limited, or whether it extends to costs. However, it is not necessary in order to resolve this ground to express a concluded view on the effect of s 32A, especially in its operation to persons found to be concurrent wrongdoers in litigation in another State. Rodriguez submitted, without contradiction, that the primary judge had not been referred to the section.

11. [770]

    The source of power is the broadly worded s 98 of the Civil Procedure Act 2005 (NSW), which gives the court “full power to determine by whom, to whom and to what extent costs are to be paid”.

12. [771]

    Seqwater had two objections against the order made. One was purely formal. It was that a declaration of right could not be made in the form it was. The power was confined to granting a declaration of existing legal rights or legal rights which will come into existence upon the happening of future events. However, Seqwater accepted that s 98(1)(b) conferred a power to make orders that the defendants pay certain proportions of Rodriguez’ costs. Examples may be seen in the form of the orders made in Voli v Inglewood Shire Council [255] and Soblusky v Egan. [256]

13. [772]

    Seqwater’s substantive point was that the appropriate starting point was not the apportionment of liability for damages between the three defendants, which was not relevant to an assessment of the defendants’ respective responsibilities for the plaintiff’s incurring of legal costs. The point was illustrated by a simple example: suppose two defendants were found liable for 75% and 25% of a plaintiff’s damages, but the former had only disputed quantum while the latter had disputed both liability and quantum. Seqwater submitted that it would be neither just nor reasonable to apportion responsibility for the plaintiff’s costs in accordance with the apportionment for damages. In the proceedings at first instance, it was submitted that the three defendants had in effect made common cause and the starting point for the exercise of the costs discretion was not the apportionment reached at the end of the trial, but the separate liability for the costs of the proceedings which should be divided equally. Seqwater also maintained that that position was supported by contribution in equity in respect of the co-ordinate liability created by the joint and several costs order.

14. [773]

    It is to be borne steadily in mind that the primary judge merely took the apportionment of liability for damages as a starting point for the exercise of the costs discretion. His Honour was not persuaded that Seqwater had established a sufficiently strong basis to depart from it. That is an approach reflected in many cases. For example, in Chapman v Hearse, [257] the order which was ultimately confirmed by the High Court was that Mr Hearse was entitled to contribution from Mr Chapman in the amount of one-fourth of the damages and costs he was obliged to pay following the death of Dr Cherry.

15. [774]

    Indeed, Seqwater’s example proves too much. If a defendant who had conceded liability but disputed quantum was ordered to pay 75% of the plaintiff’s costs, while the other defendant who had contested liability and quantum was only ordered to pay 25% of the plaintiff’s costs, on the basis that the first defendant was ultimately liable to pay 75% of the damages and nothing more, then there would be House v The King [258] error. A material consideration to the exercise of the costs discretion would be the fact that the entirety of the plaintiff’s costs incurred in relation to liability were attributable to the stance taken by the second defendant. But that proves merely that the damages ordered against multiple defendants can only be an appropriate starting point for the exercise of the discretion as to costs. The primary judge considered Seqwater’s submission, but determined not to depart from that starting point. That approach does not disclose appealable error. The fact that his Honour relied on an additional source of power which was not available did not vitiate the exercise of discretion.

16. [775]

    For those reasons, had this ground arisen, we would have recast the orders in the light of Seqwater’s objection to their form, but not interfered with their substance.

27. Costs – generally

1. [776]

   Because Seqwater has succeeded in setting aside the judgment against it, the order as to costs of the trial, which followed the event, must also be set aside. As to the appropriate substitute order, this Court does not have before it sufficient information to reassess the costs of the trial, which involved numerous issues and a multiplicity of separate judgments over several years. There would be merit in returning this issue to the mediator who has been involved with the case in the past. However, this is a matter for the parties to consider. It may be that agreement can be reached in any event. The Court will not make orders as to the costs of the trial at this stage, but will grant leave to the parties to seek further orders, such leave to be exercised within 28 days, subject to further order of a judge of the Court.

2. [777]

   It should be stated expressly that the Court does not intend itself to resolve a dispute as to the proper order for the costs of the trial, in the light of the substantive orders made on the appeal. If the matter cannot be resolved extra-curially, it will be remitted to the primary judge: an order to that effect will be made.

3. [778]

   Seqwater, having been successful in setting aside the orders made at trial with respect to its liability, is entitled to its costs of the appeal. Its success turned on the primary ground on the appeal concerning the engagement and application of s 36(2) of the Civil Liability Act. It also succeeded on various grounds concerning the proper construction of the Manual, with the result that all findings of breach of duty have been reversed.

4. [779]

   Although it has been unsuccessful with respect to specific grounds, these do not warrant a reduction in the award of costs.

28. Orders

1. [780]

   The Court makes the following orders:

   1. (1)

      In relation to orders relating to group members not the subject of final orders, grant Seqwater leave to appeal.

   2. (2)

      Allow the appeal and

   3. (3)

      Subject to the remittal provided in order (2)(c), dismiss the proceedings in the Common Law Division in matter 2014/200854 as against Seqwater.

   4. (4)

      Order that the first respondent (Rodriguez & Sons Pty Ltd) pay the appellant's costs of the appeal.