Environmental law - Air pollution - Company charged with causing pollution and emitting an unreasonable emission of chlorine gas - Negligence - Whether emission was foreseeable - Whether company exercised due diligence - Whether company failed to take reasonable precautions - Whether Magistrate erred in dismissing the complaint

Environmental Protection Act 1986 (WA), s 49(3), s 49(5), s 74(1), s 74(1a)

Canale v Bayens [2001] WASCA 383

Council of the Shire of Wyong v Shirt [1980] HCA 12; (1980) 146 CLR 40

Executive Director of Health v Lily Creek International Pty Ltd [2000] WASCA 258; (2000) 22 WAR 510

Maloney v Commissioner for Railways (1978) 18 ALR 147

Osland v The Queen [1998] HCA 75; (1998) 197 CLR 316

Romeo v Conservation Commission of the Northern Territory (1998) 192 CLR 431

State Rail Authority of New South Wales v Earthline Constructions Pty Ltd [1999] HCA 3; (1999) 160 ALR 588

Suvaal v Cessnock City Council [2003] HCA 41; (2003) 200 ALR 1

Thielemann v Commonwealth [1982] VicRp 71; [1982] VR 713

Universal Telecasters (Qld) Ltd v Guthrie (1978) 18 ALR 531

Warren v Coombes [1979] HCA 9; (1979) 142 CLR 531

Webb v South Australia (1982) 43 ALR 465

Abalos v Australian Postal Commission [1990] HCA 47; (1990) 171 CLR 167

Beneficial Finance Corporation v Karavis (1991) 23 NSWLR 256

Dempster v Mallina Holdings Ltd (1994) 13 WAR 124

Devries v Australian National Railways Commission [1993] HCA 78; (1993) 112 ALR 641

Edwards v Noble [1971] HCA 54; (1971) 125 CLR 296

Fox v Percy [2003] HCA 22; (2003) 197 ALR 201

Howells v Murray River North Pty Ltd [2004] WASCA 276

Jones v Hyde [1989] HCA 20; (1989) 85 ALR 23

1 This is an appeal by leave under Justices Act 1902 (WA), s 187, against the dismissal of two complaints against the respondent company. These complaints were for two offences, under the Environmental Protection Act 1986 (WA), relating to the discharge into the atmosphere, from a stack at the respondent's plant, of gas containing chlorine. The appeal raises a narrow issue, although one of some specialised complexity, of the application of the defences to these two offences. Those defences go in different terms to the absence of fault of the person charged with the offences. The narrow issue in their application goes to the correctness of the learned Magistrate's conclusion that the defences were made out because the discharge was not foreseeable. This conclusion was based on the evidence of witnesses called for the respondent, which the learned Magistrate preferred to that of witnesses called by the appellant. Before me no issue was taken with that preference. The narrow issue thus goes to whether drawing that conclusion from that evidence, as the learned Magistrate did, was wrong.

2 I will begin by reviewing the background to this appeal, before setting out the grounds for which leave to appeal from the decision of the learned Magistrate was granted. I will then consider the standard to be applied to review of decisions on matters of the sort to which the grounds relate, before dealing with those grounds.

3 At all material times the respondent had a plant in the Kwinana area, the Tiwest Joint Venture Pigment Plant. The Plant took synthetic rutile ore and in a process using coke and chlorine, beginning in a chlorinator, converted it first into titanium tetrachloride. Further treatment converted that compound into titanium dioxide and liberated chlorine, which was recycled under pressure into an early stage of the conversion process, in a chlorinator in the "200 Unit" of the Plant. The liberation occurred in an oxidiser. The titanium dioxide was initially a powder which ultimately became packaged for a variety of applications, including paper, paint and toothpaste. There was at least one other by-product of the conversion, hydrochloric acid. Plant operations were controlled by a Distributive Control System, or DCS.

4 The Plant was apparently the only one of its kind in this state, and was modelled initially on a plant in the United States. Subsequently, some modifications were made to the Kwinana Plant, as part of a process of audits and internal reviews. There were at least three such audits by outside contractors before the incident in question. The respondent's internal reviews used a methodology called HAZOPS, which refers to hazard and operability studies for the purposes of identifying process plant hazards.

5 One of the modifications that resulted from an internal review was in 1996 or 1997, to the operation of the valves which controlled certain gases containing chlorine, among other things, vented to the system of scrubbers which were to remove the chlorine. The operation was modified so that, in the event of a shut down of the Plant or a part of it, the valves would shut, or fail shut, as it was put. This would prevent the venting of gas to the scrubbers, where the system initially would have opened the valves, a fail open arrangement.

6 As I have indicated, a part of the Plant was a system of scrubbers, and particularly important for the purposes of this appeal were two towers called Snake Scrubbers, or Snakes, that used lime. These towers served to take gas containing chlorine which was vented either by Plant operators or automatically, in, but not only in, emergency events. The Snakes took in gas under pressure, which was passed up from the bottom of the Snake so as to come into contact with a stream of lime slurry being fed in from the top of the Snake. That contact produced a solution of calcium hypochlorite which was drawn off from the bottom of the Snake. Scrubbed gas came out of the top of the Snake and after some further treatment ultimately was vented under pressure out of a stack into the atmosphere.

7 One of a number of possible sources to the Snake for gasses containing chlorine was through a pressure control valve whose operation was of central importance to this appeal. The valve, called the 3PCV098, or the 098 valve, could be programmed to open at a pressure set by an operator between 140 and 400 kPag. Under normal operations, the pressure in the relevant parts of the Plant was at about 180 kPag, and the pressure relief setting, at which 098 would open, was set at 260 kPag.

8 Critical to the capacity of the Snake was the feeding of lime slurry into the top of the device, which was done by an electric pump with several speeds. Under certain conditions, which indicated to the DCS that chlorine was likely to find its way to a Snake, the DCS would ramp pump speed up to increase the rate of delivery of the lime. Ramping up or down would be noticeable to operators monitoring the relevant part of the Plant. If the electrical pump failed, the DCS would activate a back-up turbine system, which, however, had only one speed, and whose operation would thus not involve any ramping to signal special demand on the Snake.

9 The Snake scrubber system had an overall scrubbing capacity which was limited, by reference to the scrubbing capacity of the lime slurry supply available relative to the rate of introduction of the noxious gas to be scrubbed. The system was designed to more than accommodate the venting, initially at high pressure, of the chlorine gas inventory in the Plant at any one time, a matter to which I will return.

10 However, the system was not designed to accommodate the venting to the Snake over an extended period through an open 098 valve of the continuing production of such gas at the pressure involved in the incident in this case. In 1996, at the time of work on an expansion of the system, the respondent considered the causes of incidents that would lead to the venting of inventory would themselves cause the arrest of further chlorine generation, through the work of the DCS or the "fail safe" configurations of key systems, such as the fail shut valve arrangements. The "worst case scenario" then considered by the respondent was an introduction of gas at a pressure higher than the scrubbers could deal with, which it was considered could only occur from the part of the system for the delivery of recycled chlorine, because of the very high pressures in that system. The pressures in the lines from the oxidisers were much less, and in cases of over pressures a rupture disk would open which would cause the shut down of oxidiser operation, through what was called a DCS interlock, with the result no new titanium tetrachloride would be introduced into the overall system.

11 The Plant also had a number of sensors on the boundaries of the site to detect and give warnings of emissions into the atmosphere of gas containing chlorine.

12 On 2 February 2002 at least part of the Plant was shut down because of problems in the chlorinators. A shut down of this sort occurred only about once per year. The Plant was started up again, which involved bringing the pressure in the lines up, and flushing them with a non-noxious gas to remove any chlorine and any other waste products. This start-up phase involved setting the pressure relief point for the 098 at a relatively low value, so that gases would vent to a Snake. This was done. However, there was then another shut-down because of another equipment fault, and the start-up routine involving the 098 valve was repeated.

13 The start-up routine involving the 098 valve called at its conclusion for the pressure relief point to be reset to 260 kPag. However, owing to operator error in entering the relevant numbers, the pressure relief point was set to 200, which was above normal operating pressure, although as the learned Magistrate found "higher running pressures could be expected from time to time" (AB 273). On the night in question there was overpressure, however. The consequence of this was that the 098 valve opened twice.

14 There was a further consequence. There was an "advisory" alarm, both audible and visible, which was triggered when pressure reached 245 kPag. There was a further, "critical" alarm, to advise that the 260 kPag point had been reached. There was no alarm (then) to indicate the 098 valve had been opened (at any point) because, it seems, of the need to open the valve in normal operations such as start-up, although as I will indicate shortly there was provision for a warning indicator to show the valve was open on one of the screens monitoring the Plant for the benefit of its operators. The 200 kPag setting meant the 098 valve could open, permitting the venting of gas containing chlorine to a Snake without a pressure alarm sounding, as long as pressure remained below 245 kPag.

15 On the night of 2 February there were two further problems. One was that the electrical pump system in the relevant Snake had failed before the 098 valve began to vent gas containing chlorine to the Snake, and the DCS had activated the turbine back-up. This meant that if gas containing chlorine was vented to the Snake, there would be no ramping which would have indicated to an operator that there was such demand on the system.

16 In addition, there was a fault that night in the system, called the SCADA (the meaning of the acronym was not apparent from the material for this appeal to which my attention was directed), which monitored Plant operations, maintaining a record of them and providing various indications as to them on screens accessible to operators. The SCADA was recording operations properly, but it was not providing screen indications at least of a sort that would have alerted operators to a possible venting of gas containing chlorine to a Snake. While it is not altogether clear, from the material before me to which I was directed, whether or not the SCADA had anything to do with what I have referred to as the warning light that would have showed the 098 valve was open, nothing was made of the warning light in the appeal as distinct from the SCADA failure.

17 On the night, the 098 valve, as I have said, opened twice; once for about 35 minutes, and later for about one hour and 40 minutes. The gas containing chlorine that was vented on the first occasion was successfully scrubbed, and no gas containing chlorine was emitted to the atmosphere. However, on the second occasion gas containing about 1400 kg of chlorine was so emitted, for about 30 minutes. The venting stopped after a report had been received from a neighbouring BP plant that the smell of chlorine had been noticed there. The sensors at the respondent's plant would have detected the emission of gas containing chlorine. However, as a result of wind conditions carrying the gas as they did that night over the BP plant and out to sea those sensors had not detected the gas being emitted.

18 The gas containing chlorine vented that night was found to have affected the health of three persons on neighbouring premises, two employees of BP and one of Chubb Security.

19 Two changes were made to the Plant in relation to the 098 valve in response to the incident. One was the immediate and temporary installation of an alarm to indicate the valve had opened. The other, somewhat later, and more enduring, was to introduce an automated process whereby, if toxic gases such as gas containing chlorine were being circulated in the system, the pressure relief point for the 098 valve would be reset automatically to 260 kPag. This "logic change" as it was called had been worked on before the incident as part of the development of what was called an Instrumented Protective System, or IPS, which was to change the way the Plant dealt with pressure build up beyond normal levels. Instead of the system, controlled by the DCS, dealing with overpressure through relief valves, and through pressure valves that vented to Snakes, the build-up to an overpressure problem would cause the IPS to shut down certain areas of the Plant. Development of the IPS was an involved process, however, because of the need to ensure, among other things, it worked in harmony with the DCS. The IPS, as at the trial before the Magistrate, had still not been implemented. However, the logic change was one that could be implemented earlier, and as I have indicated it was implemented, in response at least in part to the emission incident.

20 The respondent was charged with two offences under the Act in two complaints by the appellant. One offence was that of causing pollution contrary to Act s 49(3) and the other was of emitting an unreasonable emission contrary to Act s 49(5). The learned Magistrate dismissed the complaints in a lengthy and detailed decision extending over more than 250 pages in the appeal book.

21 No issue was taken in this appeal with the learned Magistrate's analysis of those offences or with his determination that all of their elements had been proved.

22 Before the learned Magistrate the respondent sought to rely on the defence to s 49(3) in Act s 74(1a) and on the defence to s 49(5) in s 74(1). The focus in this appeal was on those defences.

23 The defence in s 74(1a) to the offence of causing pollution in s 49(3) was, so far as its terms are material to this appeal, as follows:

"(1a) Subject to subsection (2), it shall be a defence to proceedings for a Tier 1 offence if the person charged with that offence proves that -

(a) the person took reasonable precautions and exercised due diligence to prevent the commission of the offence; and

(b) ..."

24 The defence in s 74(1) to the offence of emitting an unreasonable emission in s 49(5) was, so far as its terms are material to this appeal, as follows:

"(1) ..., it shall be a defence to proceedings for an offence under this Part in respect of the discharge of waste or the emission of noise, odour or electromagnetic radiation if the person charged with that offence proves that -

(i) ...

(ii) as a result of an accident caused otherwise than by the negligence of that person,

and that the occupier of the premises, if any, from which that discharge or emission occurred took all reasonable precautions to prevent that discharge or emission."

25 The learned Magistrate analysed these defences, and found they had been made out. The basis for this determination was his finding that the "the incident was not foreseeable" (AB 273). Although he accepted that "operator error simpliciter is foreseeable", he added "the situation is more complicated than that" (AB 273). He referred to the need for the system to run at "over pressure" for the 098 valve, set at 200 kPag, to open (AB 273). He accepted that "the longer the situation persists the more opportunity there is for the operator to discover his error and intervene", referring to access to screens and the requirement to constantly monitor the process and "other indicators such as the electric motor ramping and the vacuum flows that may provide an alert" (AB 273 - 274). He also referred to the weather conditions "helping as they did to blow the emission over the Tiwest boundary without being detected by the chlorine boundary" as "also part of the equation" (AB 274).

26 In this appeal, no issue was taken with the way in which the learned Magistrate analysed the terms of these defences, which it will be observed are not in identical terms. In particular, the learned Magistrate noted the requirement for "due diligence" in relation to the s 74(1a) defence but not in the s 74(1) defence. As will become apparent the appeal is concerned rather with the learned Magistrate's application of the defences, and in particular with the standard of care he used for the purpose. The standard, for what were reasonable precautions and due diligence, in its turn depended upon his assessment of the foreseeability of the problem in question, which was the focus of the appeal.

27 There were four grounds of appeal in respect of each of the s 49(3) and s 49(5) offences. They were in virtually identical language, with variations only for the difference in the terms of the offences, and their defences. No distinction between the two pairs of grounds was taken in the argument before me.

28 Accordingly, I reproduce the grounds for appeal against the dismissal of the complaint of the s 49(3) offence, with an indication of the variations for the appeal against the dismissal of the complaint of the s 49(5) offence, below:

"(1) the learned Magistrate erred in dismissing complaint no. PE 42416 [PE 42417] of 2002 in that:

(i) approximately 1400 kg of chlorine gas, harmful to the health of human beings, was released from the Respondent's premises over approximately 30 minutes; and

(ii) the emission chlorine gas was caused without intervening agency;

and having heard uncontested evidence from witnesses called by the Respondent that the Respondent had the technology to prevent the release of chlorine gas [the unreasonable emission], the learned Magistrate erred in law in finding the Respondent had exercise due diligence and taken reasonable precautions to prevent the commission of the offence [in finding that the unreasonable emission occurred as a result of an accident caused otherwise than by the negligence of the Respondent, which the Respondent took all reasonable precautions to prevent].

(b) he erred in fact in finding that the logic changes required to prevent the 098 valve opening while chlorine was circulating in the 200 unit could not be implemented prior to the incident;

(c) he erred in law and in fact in finding that the failure to implement, before the incident, the logic changes that were made by the Respondent after the incident did not constitute a failure to take reasonable precautions when the logic changes were presumed to have been implemented by the Respondent in its assessment of a 'worst case scenario release' in 1996.

(d) he erred in law and fact in finding that the release of chlorine into the environment via the 098 valve as a result of operator error was not foreseeable and that therefore the Respondent had exercised due diligence and taken reasonable precautions to prevent the commission of the offence [and that therefore the unreasonable emission occurred as a result of an accident caused otherwise that by the negligence of the Respondent, which the Respondent took all reasonable precautions to prevent] when there was uncontested evidence from witnesses called by the Respondent that:

(i) a competent HAZOP process would have identified the possibility of a release via the 098 valve;

(ii) the logic changes that would have prevented the release were presumed to have been implemented by the Respondent in its assessment of a 'worst case scenario release' in 1996."

29 After considering the standard for review of decisions in cases of this kind, I will consider the grounds of appeal separately, although as will become apparent there is substantial overlap between them.

30 While the Act is silent as to the nature of the appeal by leave under s 187 (as is its successor, the Criminal Appeals Act 2004 (WA)), the appeal is accepted to be one by way of rehearing, and not either an appeal "in the strict sense" or a hearing de novo: Canale v Bayens [2001] WASCA 383, Pullin J; see also Brown, "Criminal Law Western Australia", [11.225.1]. However, in an appeal under the Act "the appellant must still demonstrate some legal, factual or discretionary error in the decision of the Court under appeal": Canale, per Pullin J, at [32].

31 This is not an appeal in which the matter at stake is the determination of the finder of fact below as to the credibility of witnesses. As I have indicated, no issue has been taken with the learned Magistrate's preference for certain witnesses for the respondent over certain witnesses for the appellant. Rather, the matter at stake is the correctness of certain inferences drawn from primary evidence found by the learned Magistrate. Those inferences go primarily to the foreseeability of the emission of gas containing chlorine on which to found a determination of a failure to take reasonable precautions and a lack of due diligence. In relation to matters of that sort, the correct approach appears to be that described in relation to civil appeals on determinations of matters of negligence, in Warren v Coombes [1979] HCA 9; (1979) 142 CLR 531, at 551, per Gibbs, Jacobs and Murphy JJ:

"There is in our respectful opinion no authority that entitles us to depart from the doctrine expounded in this court in cases before and including Paterson v. Paterson [(1953) [1953] HCA 74; 89 CLR 212] and in the House of Lords in Benmax v. Austin Motor Co Ltd [[1955] AC 370]. The balance of opinion in cases since Edwards v. Noble [(1971) [1971] HCA 54; 125 CLR 296] inclines in favour of adherence to that doctrine. Shortly expressed, the established principles are, we think, that in general an appellate court is in as good a position as the trial judge to decide on the proper inference to be drawn from facts which are undisputed or which, having been disputed, are established by the findings of the trial judge. In deciding what is the proper inference to be drawn, the appellate court will give respect and weight to the conclusion of the trial judge, but, once having reached its own conclusion, will not shrink from giving effect to it. These principles, we venture to think, are not only sound in law, but beneficial in their operation.

With the very greatest respect for the opinion of Windeyer J., we can see no reason to favour the suggestion, which he himself recognizes as heretical, that in a case of negligence, where the primary facts are not in question, the decision of the trial judge should be treated as the equivalent of the verdict of a jury. That suggestion has not found favour with any other member of this Court and we need say no more about it than that the traditional and practical reasons for the reluctance of an appellate court to interfere with the verdict of a jury do not exist where the judgment is that of a judge sitting alone; for one thing, the judge gives reasons, whereas the verdict of the jury is, as Lord Denning M.R. has said, 'as inscrutable as the sphinx' (Ward v. James [[1966] 1 QB 273 at 301; [1965] 1 All ER 563 at 575]). Again with the greatest respect, we can see no justification for holding that an appellate court, which, after having carefully considered the judgment of the trial judge, has decided that he was wrong in drawing inferences from established facts, should nevertheless uphold his erroneous decision. To perpetuate error which has been demonstrated would seem to us a complete denial of the purpose of the appellate process. The duty of the appellate court is to decide the case - the facts as well as the law - for itself. In so doing it must recognize the advantages enjoyed by the judge who conducted the trial. But if the judges of appeal consider that in the circumstances the trial judge was in no better position to decide the particular question than they are themselves, or if, after giving full weight to his decision, they consider that it was wrong, they must discharge their duty and give effect to their own judgment. Further there is, in our opinion, no reason in logic or policy to regard the question whether the facts found do or do not give rise to the inference that a party was negligent as one which should be treated as peculiarly within the province of the trial judge. On the contrary we should have thought that the trial judge can enjoy no significant advantage in deciding such a question. The only arguments that can be advanced in favour of the view that an appellate court should defer to the decision of the trial judge on such a question are that opinions on these matters very frequently differ, and that it is in the public interest that there should be finality in litigation. The fact that judges differ often and markedly as to what would in particular circumstances be expected of a reasonable man seems to us in itself to be a reason why no narrow view should be taken of the appellate function. The resolution of these questions by courts of appeal should lead ultimately not to uncertainty but to consistency and predictability, besides being more likely to result in the attainment of justice in individual cases. The interest of the community in the speedy termination of litigation might, no doubt, be an argument in favour of the complete abolition of appeals, although that would be far too high a price to pay merely for finality. However, if the law confers a right of appeal, the appeal should be a reality, not an illusion; if the judges of an appellate court hold the decision of the trial judge to be wrong, they should correct it."

32 I note that the appellant in this case appeared to accept that he had to demonstrate a clear error in the determination below. This appears to me to be the same approach as that I have quoted from Warren.

33 I would distinguish the approach I have just described from that of review of determinations of credibility, where the standard for review appears to be significantly higher. The respondent cited a number of authorities to me in this regard, including Suvaal v Cessnock City Council [2003] HCA 41; (2003) 200 ALR 1, per McHugh and Kirby JJ, at [76] and State Rail Authority of New South Wales v Earthline Constructions Pty Ltd [1999] HCA 3; (1999) 160 ALR 588, per Kirby J, at [90], which used the language (from Suvaal) of decisions which need to be "compellingly" or "palpably" erroneous before reversal would follow. See also "Cross on Evidence", 7th Australian ed (2004), at [11150].

34 The approach I will follow will be that I quoted from Warren.

Ground (a): error in law in finding no lack of reasonable precautions or due diligence

35 This ground rests on the quantity and duration of the emission, the lack of an intervening agency to cause that emission, and the respondent's possession before the incident of the technology to prevent it. However, this combination of circumstances is not enough to preclude any finding that the relevant defence has been made out, as the appellant appears to have conceded. If the emission was shown not to be foreseeable, or, if foreseeable, was a risk "so remote that 'a reasonable man, careful of the safety of his neighbour, would think it right to neglect it'", then the defence would be made out: Romeo v Conservation Commission of the Northern Territory (1998) 192 CLR 431, per Kirby J, at [128] (on civil negligence); see also Council of the Shire of Wyong v Shirt [1980] HCA 12; (1980) 146 CLR 40, per Mason J, at 47 - 48, where his Honour's frequently cited dicta about the need to undertake a "balancing" process in this area appear. Foreseeability is raised by ground (d) below, where I consider the learned Magistrate's finding in that respect. That is also the convenient context in which to consider whether, if the risk was foreseeable, a reasonable person would have thought it "right to neglect it", although I will also have something to say on the matter under ground (c).

Ground (b): error in fact in finding logic changes could not be implemented before the incident

36 The logic changes the subject of this ground were those that were a result of the development work on the IPS that had been done before the incident. In my view this ground must fail, as the learned Magistrate did not in fact determine that the logic changes could not be so implemented.

37 The origins of the logic changes emerge most clearly from evidence of a witness for the respondent, Karen Lee Boyce, who was the process engineering group leader for the respondent. There was the following exchange with her concerning the logic changes in her examination-in-chief (AB 408):

"Now, there is a spin-off to this in relation to the 098 valve, is there not? --- The IPS is a stand-alone computer-run system and because it would carry out some -- it will drive certain parts of the plant in certain ways, we needed to make some changes to the distributed control system, or DCS system, that normally controls the plant so that the two are compatible. We didn't want these two systems trying to tell the plant to do different things. In order to make a change - in order to make 098 compatible with the IPS system, there was a need to make some logic changes to that. Those logic changes had been planned before this incident actually occurred. So once the incident had actually occurred, we reviewed those changes and said that -- and determined that they were still appropriate in light of this incident and we made those logic changes and that was actually done mid-2002, that we actually made the logic changes to the logic for 098 and the result of having made those logic changes is now that the situation that we actually have had occur in this incident cannot occur again."

38 Later, in cross-examination there was the following exchange (AB 423):

"There is nothing to stop you implementing -- there's nothing technologically to stop you implementing the logic change that you implemented in June 2002. There was nothing to stop you doing that prior to February 2002? -- No, but there was no trigger to do it prior to 2002."

39 There is similar evidence from another witness for the respondent, Geradus Hendricus Theunissen, who was manager of process technology for the respondent. There was the following exchange in his cross-examination, in which the learned Magistrate also participated (AB 643):

"Yes, and we've heard some evidence that there has been a change since the event in relation to the IPS development which puts into place the system that you cannot have the 098 valve open if the system is running. That is, if the oxidisers are putting chlorine into the system. Is that correct? --- There's a conflict in what you're saying there, in that yes, there have been changes made to the valve, but I see no connection between that and the IPS. I see that as nothing more than re-configuration of the DCS to prevent that valve from opening. It has nothing to do with the IPS, as I see.

HIS WORSHIP: Wasn't it described as a change in logic rather than the IPS? Because the IPS is still developmental, isn't it? --- Correct."

40 In his decision the learned Magistrate dealt with the matter as follows (AB 272 - 273):

"I find that the logic changes made by Tiwest after the incident did not constitute a failure to take reasonable precautions prior to the incident.

I accept Boyce's evidence that Tiwest were developing the IPS prior the incident and it was not able to be implemented. I accept that logic changes were necessary to make the 098 valve compatible with the IPS. These were reviewed after the incident, found to be appropriate and implemented. It is the situation now that the introduction of chlorine or titanium tetrachloride into the system forces the 098 valve closed and resets the pressure point to 260 kPag."

41 Although the matter could be clearer, I am of the view it is reasonably clear from this passage, when read with the evidence to which I have referred, that the learned Magistrate was not proceeding on the (erroneous) basis that the logic changes could not have been made prior to the incident.

Ground (c): error in law and fact in finding the failure to implement the logic changes before the incident was not a failure to take reasonable precautions

42 I have just referred to the learned Magistrate's determination in this regard. The determination was challenged on the basis that the respondent had indeed, when considering in 1996 its then new Snake, considered the possibility of chlorine gas being vented to the Snake while such gas was being produced by the oxidisers, and should have recognised that there was a risk that warranted address by a precaution such as the one in question.

43 The consideration appears in an internal memorandum from Mr Theunissen "Snake Lime Scrubber - 80,000 TPA Expansion" dated 4 September 1996 and one of the parts of a facsimile transmission from the respondent to the Department of Environmental Protection for the purposes of a meeting with representatives of the Department on 6 September 1996. These materials formed exhibit P 27 at the trial. The 4 September 1996 memorandum says this (AB 778):

"Although it is not intended that the Oxidisers be vented to the Snake with flows still active, and in fact it can't instantly happen while the inventory is venting, the calculations show that even after an inventory dump, with an initial lime strength of 10% W/V Ca(OH)2, there is still sufficient lime strength to cope with a hypothetical event."

44 In cross-examination of the author of the memorandum, Mr Theunissen, the following exchange occurred (AB 685 - 686):

"MS THATCHER: Getting back to this, you know that in order to limit -- you know at the time you're doing these calculations that you need -- in order to say that your worst case scenario is the venting situations that you have put in you need to have something that stops venting beyond the inventory of chlorine, don't you? --- It's desirable to have something that stops even inventory from venting.

Perhaps you could just answer the question. Your state of knowledge at that stage is that for your worst case scenario to be effectively the inventory of chlorine there has to be some way of ensuring that you're not at any stage venting any more than that to the stack - to the snake system? --- The exercise in this report was to demonstrate the scrubbing capacity of the snake, not to prove the integrity of the plant that was feeding to it, and I took what I considered to be pretty severe scenarios and used them as what I considered to be realistic worst case events, and, clearly, an event has occurred which was outside the boundaries of those parameters which I set myself when we did this and --

Can I -- I'm sorry, please proceed? --- And, yes, clearly things could have been done to stop that happening. At the time this was being done my charter was not to do a complete --- was not to do a complete HAZOP review of the plant and look for any possible flaws in our management strategy there. This document was purely to try and demonstrate what the capacity of the snake was.

MS THATCHER: Can I take you back to page 228 which I think you should still have open --?--- Yes.

-- to you. The second of the two paragraphs that you read commencing, "Note" I'll read them again:

'Note that in all incidents considered, venting of the chlorine inventory only is considered as the causes of such incidents will cause the arrest of further chlorine generation in those systems.'

WITNESS: Correct. That's what's written.

MS THATCHER: As you write it there, a person reading it would assume, then, that once you have commenced venting there is no more chlorine production no matter which of the scenarios you take? --- And that is a design philosophy we which -- which we would like to ensure is within the plant design.

Mm? --- And it clearly wasn't.

It clearly wasn't in place? --- Correct."

45 An earlier letter from the respondent, dated 12 March 1996, to the Department of Environmental Protection headed the "Snake Scrubbing System" attached a document "Overview of Chlorine Within the Process at the Kwinana Pigment Plant" which reviewed the capacity of the Snakes to absorb chlorine inventory. This letter and its attachment formed exhibit P31 at the trial. The attachment said this (AB 790):

"Reflection on these numbers clearly emphasises the need to arrest chlorine generation in any venting situations. Thus, in all situations where there is chlorine flow to the Snake Scrubber, the prime focus must be on stopping the production of more chlorine in order to limit the total amount of chlorine ultimately discharged from the system. This is addressed in two ways.

Firstly, configuration of the DCS prevents continued production of chlorine while the system is venting. Interlocks will act automatically and the Operator initiated CSD [sic?] system will stop the process and vent the system in a carefully programmed sequence. This function is routinely tested to ensure its reliability. It is subject to T.H.C.P and associated audit.

Secondly, in the event that there are systems failures (power, compressed air etc.) the fail-safe positions of critical control valves will halt the process and vent pressure from the System to the Snake Scrubber. This instance probably represents a 'worst-case' and has occurred without major problems. In the early history of the Plant, some problems were experienced with reliability of the emergency generator and steam turbine governors, however, these have been rectified. Regular tests of the systems ensures they are always serviceable and will perform on demand."

46 The reference to "T.H.C.P" above is to the respondent's "Total Hazard Control Plan", to which, with its "associated audit", I will return below.

47 This attachment confirmed, the appellant said, that the respondent had recognised the need to have a system in place to deal with the risk of venting of gas containing chlorine to the Snakes while production continued. The logic change represented such a system known to the respondent before the incident in question.

48 However, this evidence also demonstrates that the respondent had in place systems to deal with what it understood was the risk it faced of such venting. This goes to whether the risk of an incident of the sort in this case was foreseeable. I return to this evidence in that context below. However, there is another significance to the evidence which goes to whether, if the risk was foreseeable, any further precautions were warranted.

49 It is established, both for the law of negligence generally, and defences to criminal liability formulated in terms of exercising due diligence and taking reasonable precautions, that "perfection or the use of increased knowledge or experience embraced in hindsight after the event should form no part of what is reasonable in the circumstances": Maloney v Commissioner for Railways (1978) 18 ALR 147, per Barwick CJ, at 148; see also Universal Telecasters (Qld) Ltd v Guthrie (1978) 18 ALR 531, per Bowen CJ, at 534. Further, while the fact that the "absence of past mishaps" does not entitle a person to ignore safeguards against dangers, "it is equally true that years of experience without accidents may tend to confirm" an assessment that "the risks of harm are negligible": Romeo (supra), per Kirby J, at [128]. In a case of this sort, "the magnitude of the risk and its degree of probability remain to be considered with other relevant factors": Shire of Wyong (supra), per Mason J, at 48.

50 It was put to me that the material and other evidence which I have set out under this ground of appeal show that the respondent had set the standard for a reasonable response to the foreseeable risk in question. It was that standard I should follow.

51 However, to the extent that the conduct of the respondent before the incident might be probative of common practice (a difficult matter in itself, given, as I have indicated, the relative uniqueness of the respondent's Plant), it seems to me it points away from the conclusion sought, apart from considerations of the magnitude of the risk and the probability of the incident. I will return below, under the final ground of appeal, to those considerations. It was put to me that the fact the respondent had arrangements in place through the DCS interlocks to shut down the Plant, or at least certain operations in it in other events, was an indicator of the appropriate response to the risk in this case. However, that analysis does not take account of the need to do the balancing called for by the balancing approach to negligence from Shire of Wyong (supra).

52 To the extent that the conduct of the respondent after the incident (in installing the logic changes) is also relied upon, the response is that such conduct may be relied upon to show that a particular precaution was practicable, but not to show that the failure to use it earlier was negligence: Thielemann v Commonwealth [1982] VicRp 71; [1982] VR 713; and Fleming, J G, "The Law of Torts", 9th ed (1998), at page 134.

53 It follows, in my view, that the fate of this ground of appeal turns on that of the next ground.

Ground (d): error in law and in fact in finding release of chlorine via the 098 valve as a result of operator error was not foreseeable

54 The learned Magistrate had this to say in relation to the matter of the foreseeability of operator error and of consequential release of chlorine (AB 273):

"I find that this incident was not foreseeable.

In this regard I accept the evidence of Allen.

I accept that operator error simpliciter is foreseeable. However the situation is more complicated than that."

55 The reference to "Allen" was to the evidence of Christopher John Allen, a consultant in the areas of risk management whom the learned Magistrate accepted was a specialist with expertise in chemical engineering and substantial experience in risk management and accident investigation, and in particular in relation to production plants handling dangerous goods and hazardous materials. He was retained by the respondent to evaluate a report of an expert produced for the Crown Solicitors Office on the incident, and to comment on the adequacy and efficiency of the systems and controls the respondent had in place at the time of the incident. Mr Allen's report was exhibit D 36. That report's principal findings for the purposes of this ground of appeal were (exhibit D 36, AB at 1097 - 1098) as follows, the first and second of which refer to the complications invoked by the learned Magistrate:

• the incident was the "result of a complex set of circumstances that are unlikely to be repeated";

• the probability of the incident occurring was very low, at approximately once every 4000 years;

• the "root cause" of the incident was "the inability of extensive risk analysis to identify and quantify the risk";

• the consequences of the release were "not life threatening" although the consequences "may have been different had the wind been blowing from another direction"; and

• "even if the risk had been foreseen, Tiwest would not have been required to take any action as it was sufficiently low".

56 The report's reference to the incident was to the undetected release of chlorine from the stack servicing the Snake, alternative causes for which it was "assumed" would have "other effects" that would have been "noticed by the operator" (exhibit D 36, at AB 1119).

57 The report's assignment of a frequency to the risk was on the basis of the assignment of probabilities to the causes leading to the incident, making up a "fault tree". The learned Magistrate determined that this approach was "an appropriate methodology" (AB 261).

58 The report's reference to "extensive risk analysis" was to the HAZOPS performed by the respondent, as well as three different risk assessments done by three different consultants before the incident, and the THCP of the respondent, which had been audited. The report further indicated that (exhibit D 36, at AB 1139):

"A root cause of the incident was the inability to identify the potential loss of containment scenario caused by an inappropriate set point on 3PCV908.

It is not possible to identify whether successive HAZOPs considered this scenario. As an experienced HAZOP facilitator, the author considers it highly unlikely that the possibility would not have been considered. The best assessment therefore is that the HAZOP team considered the risk to be acceptable and/or that the operator would receive adequate warnings in other ways."

59 The report's evaluation of the consequences of the incident emphasised the low concentrations of the chlorine at ground level in the neighbouring BP plant across which, to the ocean, the wind carrying the gas containing chlorine was blowing. The consequences were put at "moderate" for "moderate injuries" (exhibit D 36, at AB 1127), although consideration was also given to the risk of more serious injury, in the "severe" category, which was also put at the "very unlikely" level (exhibit D 36, at AB 1129).

60 In the face of this material, it seems difficult for me to come to a different view from, or to find fault with, the learned Magistrate's view of Mr Allen's evidence from his report, which it was not suggested to me he had qualified in his testimony, that it indicates the undetected release of chlorine in this incident was an "unforeseeable" event. As counsel for the respondent put to me, reliance on evidence of an expert for the purposes of such a conclusion is appropriate in a case such as this one, involving an area outside the ordinary experience or knowledge of the trier of fact that is the subject of a reliable body of knowledge or experience: see Executive Director of Health v Lily Creek International Pty Ltd [2000] WASCA 258; (2000) 22 WAR 510, per Ipp J, Owen and Miller JJ agreeing, at 519, quoting from Osland v The Queen [1998] HCA 75; (1998) 197 CLR 316, per Gaudron and Gummow JJ at 336.

61 However, the appellant put to me that the matter of foreseeability should not have been tested as the learned Magistrate did, drawing as he did on Mr Allen's evidence, and referring to the combination of circumstances that led to the incident. It was sufficient, it was said, there was evidence, from Mr Allen himself, that the possibility of a release of gas containing chlorine through an incorrect setting point for the 098 valve was something that HAZOPS analysis would have identified, even if it would have been concluded that "the operator would quickly become aware of the situation and remedy it before any significant release occurred" (exhibit D 36, at AB 1111).

62 This submission, in my view, does not take proper account of the basis, emerging from his report, for the assumption Mr Allen made as to the other causes that might lead to a release to the atmosphere than the particular ones in this case. That basis, upon which the learned Magistrate had drawn (AB 273 - 274), was Mr Allen's conclusion that it had required a very unlikely combination of circumstances to produce the incident, of an undetected emission of chlorine from the stack servicing the Snake. In particular, there was, in addition to the operator error, the combination of the failure of the operator to notice the error over a period of time contributed to by the failure of the monitoring screens, the role in relation to which of the failure in the SCADA system I have earlier referred to; the ramping of the lime slurry electric pump which could not occur because the turbine back-up pump had been engaged; and the weather conditions which meant the respondent's boundary sensors had not been engaged. This was not a case like that involving the construction of a false curb leaving an open gap leading to a pedestrian sustaining an injury, although in an unexpected way, in Webb v South Australia (1982) 43 ALR 465. In this case the very risk itself was only the product of the combination of circumstances the subject of Mr Allen's fault tree analysis. This case is more like the first and the third of those distinguished in Webb, per Mason J, Brennan and Deane JJ, at 466, as follows:

"The case is to be distinguished from those in which the risk of injury is not foreseeable because it is the outcome of a complex of unpredictable events. The present facts are to be distinguished from [Shire of Wyong (supra)] and Overseas Tankship (UK) Ltd v Miller Steamship Co Pty Ltd (the 'Wagon Mound' (No 2)) [1966] UKPC 1; [1967] ALR 97; [1967] 1 AC 617, where the chain of events creating the risk of injury was more exotic - yet in each instance the risk of injury was held to be foreseeable. Nor is it a risk which, though foreseeable, can, like that in Bolton v Stone [1951] UKHL 2; [1951] AC 850; [1951] 1 All ER 1078, be dismissed as being 'fanciful' or 'far-fetched' because the prospects of its occurrence are so very unlikely."

63 There is no doubt there was expert evidence that the combination of events creating the risk of injury included some that were predictable, such as the weather conditions, and others, most notably the failure of the operator to notice the error, that were not. The assessment by the learned Magistrate of the expert evidence as to the overall effect of this combination does not disclose any error, in my view, nor is it one I would not have made. I would note that in addition to the evidence of Mr Allen, there was the evidence of Ms Boyce, to which I have already referred, that there was no "trigger" to introduce the logic change prior to the incident (AB 796), and the evidence of Mr Theunissen, to which I have also referred, that "venting of the chlorine inventory only is considered as the causes of such incidents will cause the arrest of further chlorine generation in those systems" (AB 686). While neither appears to have conducted an assessment of the sort Mr Allen undertook, their evidence is consistent with his.

64 The learned Magistrate did not go to consider whether or not, if the risk of the undetected emission of chlorine had been foreseeable, it could have been dismissed as being "fanciful" or "far-fetched", and thus, in the language of Kirby J in Romeo (supra) "so remote that 'a reasonable man, careful of the safety of his neighbour, would think it right to neglect it'". However, it follows from my view of the evidence on which he did rely that, even if the undetected emission of gas containing chlorine to the atmosphere had been foreseen, it would not have been an error for the learned Magistrate to have concluded that it was a risk of such a character, nor would it have been a conclusion I would not have reached.

65 Thus, I would not uphold this ground of appeal. It follows from the analysis I used under grounds (a) and (c) that I would not uphold either of them for the same reasons.

66 I thus have not upheld any of the grounds of appeal in this matter, and would dismiss the appeal.