[Home] [Databases] [World Law] [Multidatabase Search] [Help] [Feedback]
	England and Wales Court of Appeal (Civil Division) Decisions
You are here: BAILII >> Databases >> England and Wales Court of Appeal (Civil Division) Decisions >> United Marine Aggregates Ltd v G M Welding & Engineering Ltd & Ors [2013] EWCA Civ 516 (14 May 2013) URL: http://www.bailii.org/ew/cases/EWCA/Civ/2013/516.html Cite as: [2013] EWCA Civ 516

[New search] [Printable RTF version] [Help]

IN THE COURT OF APPEAL (CIVIL DIVISION)
ON APPEAL FROM THE HIGH COURT OF JUSTICE
QUEEN'S BENCH DIVISION
TECHOLOGY AND CONSTRUCTION COURT
Mr Justice Edwards-Stuart
[2012] EWHC 779 (TCC)

		Royal Courts of Justice Strand, London, WC2A 2LL
		14/05/2013

Lord Justice Tomlinson :

1. The Appellant, United Marine Aggregates Limited, Claimant at trial, to which I shall refer hereafter as "UMA", owns a large aggregate processing plant on the Thames near Greenwich. On Sunday 3 February 2008 fire broke out in a part of the plant known as the screening house. The first sign of the fire was smoke observed at about 09.50 emanating from the top of the screening house. Earlier that morning two employees of the Respondent, GM Welding and Engineering Limited, Defendant at trial, to which I shall refer hereafter as "GM", had been carrying out hot works in the screening house. In fact the relevant hot work had started shortly before 08.00 and had been completed shortly before 08.15. Shortly before 09.30 the two employees of GM, Messrs Smith and Percival, left the screening house to go for breakfast. It was whilst they were at breakfast just off the site that the fire manifested itself. It caused extensive damage.
2. The Appellant alleged at trial that the fire was caused by breaches of contractual and tortious duty owed by GM. GM in turn sought an indemnity from its public liability insurers, Novae Syndicates Limited, to which I shall refer hereafter as "Novae".
3. After an eight day trial confined to the issue of liability the judge held in favour of GM that it had not been in breach of duty. It had done its work properly and carefully and taken the agreed precautions against the risk of fire. To the extent that it may have departed in one respect from the agreed precautions, which UMA had failed to prove, the departure was not causative of the fire. The fire broke out in a manner and in circumstances which were not reasonably foreseeable.
4. UMA's claim accordingly failed. GM had no need of an indemnity from its insurers, although that claim too would have failed had GM had need of it because GM was in breach of a warranty in its insurance cover as to the manner in which combustible materials in the immediate vicinity of hot work should be covered and protected whilst the hot work was being carried out. That failure did not also amount to a breach of the duty owed to UMA – the adjacent combustible materials were during the hot work covered in the contractually agreed manner, by water sprayed carefully at high pressure. However the fact that a fire occurred demonstrates that the lining of the underpan was not both covered and protected at all times and thus that the warranty was broken.
5. UMA appeals against the judge's decision. Novae appeals against the judge's decision to award it only 50% of its costs.
6. The judgment of the judge below, Edwards-Stuart J sitting in the Technology and Construction Court, is careful, detailed and comprehensive. It runs to 60 single spaced pages and 303 paragraphs. It is quite unnecessary in order to dispose of this appeal to attempt to replicate that detail. The curious will find the judge's full account at [2012] EWHC 779 (TCC).
7. The relevant work was being carried out in and to a container in the screening house known as the top box. The top box is a rectangular container. It contains two further large steel vibrating containers known as screens through which aggregate of a particular size passes into the underpan below. The screens are braced by transverse steel support tubes which are secured to the structure by bolts which pass through steel flange plates welded to the ends of the tubes. The tubes are prone to substantial stress in service and from time to time require routine maintenance. The most common form of failure takes the form of cracks which propagate outwards from the bolt holes in the flanges. These cracks must from to time be "vee-ed out" and then repaired by welding. In order to do this the bolts have first to be removed. Using a small angle grinder a groove is then cut along the line of the crack which is then filled by welding.
8. It is the removal of the nuts from the bolts securing the flanges which was here the relevant hot work. The nuts were on the outside of the container. They were cut off by Mr Smith using an oxy-propane cutting torch. Oxy-propane cutting produces "spatter". Spatter consists of globules of molten steel with temperatures of the order of 1500^oC. They can travel up to 10 metres. If the technique being used is "washing" rather than "lancing", the globules will be fewer in number but they will be larger and they will not travel so far. The technique being used here was washing, but there was opportunity for the spatter to pass through a gap in the side of the container below the nuts into the underpan beneath the screen. The underpan is lined with a combustible rubber lining material. The rubber lining is in turn attached to the steelwork by a combustible mastic sealant. At the top of the underpan, below the level of the nuts and at the bottom of the gap through which the spatter could pass, there is an exposed narrow band of mastic. This is just below a steel strip welded to the long side of the container at an angle of about 45 degrees to the vertical side.
9. The judge found that the fire was caused by a globule of molten steel, spatter, from the oxy-propane cutting which penetrated the band of mastic to which I have just referred. The hot globule went through the mastic band and penetrated some way into the interior of the lining. There it started a smouldering fire.
10. The whole working area around the screen was hosed down before any hot work started. Whilst Mr Smith was carrying out his cutting work water from a high pressure hose was continuously directed into the side of the underpan on which he was working. The judge was satisfied that Mr Smith, who he found to be a conscientious and highly competent craftsman of integrity, would have deployed the hose in the most effective manner for the work that he was doing at the time. It was work that he had done many times before. After the work was completed the area was again thoroughly hosed down. The judge expressly found that "the hosing down of the lining of the underpan by Mr Smith was carried out with reasonable care, as he had done it on many previous occasions."
11. Mr Smith and Mr Percival checked the area carefully both visually and using their sense of smell. They remained in the area of the underpan for at least one hour after the hot cutting had finished and one of them probably for about one and a quarter hours, in order to check for any signs of a fire. By the time that Mr Smith and Mr Percival left the area of the underpan, probably shortly before 09.30, there was no visible fire in the underpan and there was no reasonably detectable smell of burning.
12. The judge found that it was reasonably foreseeable that a molten globule of steel might avoid the water and hit and penetrate the horizontal band of mastic that protected the top of the lining just below the steel strip that was fixed to the top of the side of the underpan. However he went on to find that it was not reasonably foreseeable that a globule of molten metal might penetrate sufficiently far into the lining so as to start a smouldering fire that would not be extinguished when the underpan was hosed down on completion of the work and would not be detected by its smell when the area was being checked for signs of fire. The judge also found that it was not reasonably foreseeable that a fire might in this way be caused which could smoulder unseen and undetectable within the lining for about one and a half hours or more and then make the transition into a visible flaming fire.
13. The judge had evidence in the shape of a recommendation by the Fire Protection Agency that in relation to work such as was here being undertaken a fire watch should continue for at least thirty minutes after the hot work is completed, with further checks at regular intervals up to sixty minutes after completion. As it happens, this recommendation was effectively incorporated into the insurer's warranty in this case. The judge concluded that in order to meet the rare and unforeseeable situation which had here arisen a fire watch would have been required of considerably longer duration than that recommended.
14. The contractual provisions to which GM were obliged to adhere whilst doing this work, the UMA Hot Work Procedure, included the following:-

2.2 All sources of fuel within a 10M radius shall be removed where possible. Any which cannot be removed shall be adequately protected from heat and sparks.

2.3 Where it is not possible to remove sources of fuel, e.g. conveyor belt, wooden walkways, rubber decks or chute linings, etc, these should be protected by spreading non-flammable dust, fire blankets, covering with steel plates, etc, where possible the area should be damped down using water.

2.4 Where it is not possible to protect such items, extreme care and attention are required whilst carrying out hot work to prevent a fire occurring.

. . .

5.3 A constant check of the area to be made for any signs of fire.

5.4 The working area shall be monitored after the hot work has been completed for signs of heating or fire. The period of time will vary according to circumstances and the degree of risk involved (30 minutes to four hours or longer).

15. There was considerable controversy at trial as to whether Mr Percival, who was the only candidate for the task, was carrying out a "dedicated" fire watch whilst the hot work was being carried out by Mr Smith. However the judge found that whether he was or was not was of no relevance, as the fire broke out in a manner which would not have been seen or detected by a dedicated fire watcher, however diligent. All that a fire watcher would have seen was that some spatter went into the underpan to be apparently extinguished. There was no fire of any sort to be seen in the underpan before Mr Smith and Mr Percival left the area, approximately one and a quarter hours after completion of the work, at which stage no fire was yet visible. Paragraph 5.4 of the Hot Work Procedure is unspecific as to the period of time over which the working area should be monitored, after completion of the hot work, for signs of heating or fire. UMA made no submission at trial as to what was a suitable period. The judge found that in the light of the Fire Protection Agency recommendation UMA could not plausibly assert that GM was obliged to leave one man in the area to act as a fire watch for a period in excess of one hour, and that it was only a fire watch of the order of, say, two hours that would have been able to observe the transition from a hidden smouldering fire into a flaming fire.
16. It followed therefore that GM had acted properly and carefully and was in breach of no contractual requirement. The fire had started in a manner which was not reasonably foreseeable and no asserted breach of duty would in any event, if proved, have been causative of the fire.
17. The appeal, as it was developed by Mr Ronald Walker QC for the Appellant UMA, was more ambitious than that foreshadowed by the Grounds of Appeal upon the basis of which permission to appeal to this court was granted. No appeal against the judge's conclusion can succeed unless his essential findings of fact are undermined. Skilfully though Mr Walker developed his arguments, they were essentially at the periphery of the judge's findings, and moreover directed at areas which were not ultimately determinative. I deal first with the Grounds of Appeal.
18. Ground 1

"The learned judge erred in law in holding that the "Safe System of Work" document of February 2005 was not contractually binding, notwithstanding that (a) as a matter of construction it was, and (b) Mr Marshall accepted that it was.

Had the judge so found he would have been bound to find that the Defendant was in breach of the contractual obligations (a) to have a "dedicated person operating hose and for fire watching" and (b) "a dedicated person will remain on the hose for the duration of any hot work"."

The Hot Work Procedure to which I have already referred was incorporated into the June 2007 version of UMA's Health and Safety Manual. Another relevant document was the Permit to Work, issued on 2 February 2008, to which I refer below. The allegation that a further document, "Safe System of Work", introduced additional obligations, over and above those contained in the admittedly applicable Hot Work Procedure and Permit to Work, was not clearly made before trial and was born out of confusion as to the provenance of the Safe System of Work document itself. It was always an implausible allegation and it is also irrelevant unless the judge's findings on causation can successfully be attacked. The only additional obligation which the Safe System of Work document potentially introduces is expressed in the document as "dedicated person operating hose and for fire-watching" and "a dedicated person will remain on the hose for the duration of any hot work." The judge found that any failure in this regard was not causative of the fire – the hose could have been deployed no more effectively by a dedicated operative and nor would a dedicated fire watcher have observed the onset of fire.

19. In any event the judge's reasons for regarding this further document as of no relevance are compelling. The document produced was undated, albeit it had been countersigned by four GM fitters carrying out work on another occasion on 4 February 2005, three years earlier than the fire in February 2008 with which this claim is concerned, and by a Mr Langton for UMA. It was plainly a document which would be produced on an ad hoc basis for agreement and signature in relation to specific work. Moreover it is clear both from the heading "Safe System of Work for Rebuilding Support Frames etc" and from at least paragraphs 9-13 of the text that it is directed to significant rebuilding work involving cutting, removal and installation of heavy steel work rather than routine maintenance. UMA also led no evidence relevant to the incorporation of contractual terms. The natural inference from the material before the court was that the undated document, even if once relevant, was superseded by the Hot Work Procedure incorporated into the May 2007 version of the UMA Health and Safety Policies and Procedures Manual. The answer extracted from Mr Gary Marshall (effectively the owner of GM) in cross-examination to the effect that the work carried out in February 2008 "to a degree" fell within the description of the work outlined in the Safe System of Work document is of no relevance to the true construction of the document, still less to the question whether it had contractual effect in relation to the relevant work in February 2008.
20. Ground 2

"The learned judge erred in failing to find that there was a breach of UMA's Hot Work Procedure (which he did find to be contractually binding) because there was not a "constant check of the area . . . for any signs of fire."

This is an attack on one of the judge's principal findings of fact. A constant check may be made by the person carrying out the hot work, i.e. here Mr Smith, as the judge observed at paragraph 198 of his judgment. In fact the relevant Permit to Work required a second person to be present, as Mr Percival was. Moreover it was the evidence of Mr Smith that Mr Percival, who did not himself give evidence, was looking out for signs of fire whilst he, Smith, was carrying out the cutting. The judge was entitled to accept the evidence of Mr Smith to the effect that he, Smith, exercised extreme care and vigilance to ensure that none of the sparks from the cutting equipment ignited anything. In other words Mr Smith made a constant check of the area for any signs of fire, both during the work and for as long as was reasonably required thereafter. Mr Smith's preoccupation would have been with his own work. The evidence painted a confused picture so far as concerns the precise nature of the work, other than keeping a look out for signs of fire, being undertaken by Mr Percival whilst Mr Smith was cutting off the nuts. Accordingly, the judge was not prepared to go so far as to find that Mr Percival was also keeping a constant check for fire whilst the cutting work was in progress. He may have been carrying out another task, albeit in the same area, the screen in which Mr Smith was working. However it was for UMA to prove, if relevant, that Mr Percival was not in fact keeping a fire watch whilst the cutting was in progress. As it happens, given the judge's findings as to the nature of the fire, it was not relevant, but UMA did not adduce material which justified the positive inference that Mr Percival was not maintaining a proper fire watch. The material upon which Mr Walker relied on the appeal was relevant rather to the question whether Mr Percival was maintaining a dedicated fire watch. In my judgment it falls far short of justifying a conclusion that Mr Percival was not in fact maintaining a proper fire watch, albeit there is uncertainty as to what other tasks he might have been carrying out at the same time.

21. Ground 3

"The learned judge erred in finding that there had been an (unpleaded) waiver of the obligations imposed by UMA's Hot Work Procedure."

There were in play at trial four potentially relevant sources of contractual precautions: the Risk Assessment, the Method Statement, the Permit to Work and the Hot Work Procedure. The Risk Assessment envisaged the use of fire blankets. The Method Statement referred to the need for a constant fire watch and live water at hand at all times. The Permit to Work required two men to be present. I have already set out the salient parts of the Hot Work Procedure. The judge's findings are contained in the following paragraphs:-

"195. When one considers these documents, together with UMA's Hot Work Procedure and what actually happened on site prior to the weekend of 2/3 February 2008, the evidence as to how the work was to be carried out is, in my judgment, all one way. It was clear from his oral evidence that Mr Farla, [the UMA Wharf Manager], like his predecessor no doubt, knew exactly how the Defendants were intending to carry out the work to the A side primary screen. In particular, he knew that the Defendants would not be using welding blankets to protect the lining of the underpan, because that would be done by protecting it with a constant supply of water. So, apart from the skill of the operator of the cutting torch, the precautions to prevent fire that were known to or were expected by him consisted of:

- thoroughly damping down the area of work (principally the lining of the underpan) with water before the work, and

- having a bed of sand/ballast on the bottom of the underpan, and

- having one man on fire watch during the hot work, and

- having a high pressure hose in the relevant side of the underpan trained so as to provide a constant spray of water in the vicinity of the work whilst it was being carried out, and

- thoroughly hosing down the area of work with water after the work, and

- carrying out regular checks for fire for at least 30 minutes after the work in accordance with UMA's Hot Work Procedure and, if later, until the area had cooled and there were no signs of any remaining heat or smell of combustion.

196. To the extent that these precautions were more limited than those required by UMA's Hot Work Procedure, which in the ordinary course of events would be contractually binding on the Defendants, I consider that strict compliance with the Hot Work Procedure was waived by UMA. I do not reach that conclusion from the issue of the Permit to Work alone, because I do not regard Mr Curtis (the person who issued it) as a person who can be regarded for these purposes as a directing mind of UMA. However, the position is different in the case of Mr Farla, who was the Wharf Manager. His position in the organisation was such that I consider that if he gave his informed consent to a particular set of precautions that did not comply strictly with UMA's Hot Work Procedure then his doing so constituted a waiver by UMA of strict compliance with that procedure.

197. This conclusion is consistent with the concession made by Mr Walker, in his written closing submissions, that the agreement between Mr Marshall and Mr Farla that it was not necessary to de-mat the B side screen amounted to a variation of the contract. Whether it is properly characterised as a variation of the contract or a waiver of the requirement to comply with a particular contractual obligation probably does not matter in the context of this case.

198. As it happens, with one possible exception, I cannot detect any material difference between the precautions that the Defendants were proposing to adopt and the precautions that were required by UMA's Hot Work Procedure as applied to this situation. The possible exception is that paragraph 5.3, which requires a constant check of the area to be made for any signs of fire, does not say that this has to be done by a second dedicated fire watcher. It may be that it could be complied with if the person carrying out the work paused every minute or so and carried out a visual check of the area. However, that question does not arise in this case because, as I have already mentioned, the Permit to Work required a second person to be present."

22. It is also relevant at this point to note the important concessions recorded by the judge at paragraphs 183 and 184 of his judgment:-

"183. Having disposed of this point, I now turn back to the Hot Work Procedure document. In his oral closing submissions, Mr Walker accepted, correctly in my view, that adequate protection of the lining from heat and sparks could be achieved with water. He therefore accepted that "etc" in paragraph 2.3 could include the protection of rubber linings by continuous spraying of the relevant area with running water during the hot work, as an alternative to the other precautions listed in paragraph (namely, "spreading non-flammable dust, fire blankets, covering with steel plates")."

184. Accordingly, it followed also, as Mr Walker accepted in his oral closing submissions, that paragraph 2.4, with its reference to "extreme care and attention" did not apply. As he put it, this was because "an adequate amount of water would do" to satisfy the requirement imposed by paragraph 2.3. I consider that, in the light of the evidence, Mr Walker was right to make these concessions."

23. These paragraphs reflect Mr Walker's acceptance at trial that it was not practical to cover the underpan with fire blankets or metal sheets during the hot work. Spraying with a high pressure hose was accepted to be adequate and moreover comprised within the catalogue of permitted precautions set out in paragraph 2.3 of the Hot Work Procedure. The judge's conclusion about waiver was therefore irrelevant and paragraph 198 of his judgment is to some extent not easy to follow. If there was a departure from the Hot Work Procedure, it consisted only in the presence of an additional man, on the assumption that such was not required by the reference to a "constant check of the area". The only real relevance of the judge's conclusions in this regard is, as Mr Stuart Hornett for GM pointed out, that UMA approved at management level the precautions against fire which were to be taken. This rightly informed the judge's ultimate conclusion as to the ambit of GM's duty of care in the particular circumstances of this case.
24. Ground 4

"The learned judge erred in failing to find that the Defendant had not appointed a dedicated fire watch and that no-one had performed that function. In particular he erred in failing to draw an adverse inference from the failure of the Defendant to call as a witness Mr Percival - who was allegedly the dedicated fire watcher – although there was, and could be, no inference which could sensibly be drawn, other than that he would have conceded (a) that he was not the dedicated fire watcher and (b) did not in fact act as fire watcher while Mr Smith was carrying out the hot works which (as the judge found) caused the fire."

I have already dealt with the issues raised by this ground.

25. Ground 5

"In the premises the learned judge also erred in finding that the Defendant had not been negligent."

This point simply does not arise but it is in any event a non-sequitur.

26. Ground 6

"Having accepted that there was not a dedicated person operating the hose, and that Mr Smith did not direct the hose at the side of the underpan where he was working from time to time, the learned judge erred in finding that complying with the Defendant's contractual obligations would not have prevented the fire."

This ground mischaracterises the judge's findings as to Mr Smith's operation of the hose. His findings are at paragraphs 255 and 256 of his judgment as follows:-

"255. As I have indicated, I am satisfied that whilst Mr Smith was carrying out the hot cutting water was being directed continuously into the underpan towards the area where he was working. However, I am unable to decide whether the hose was directed at the central divider in the underpan so that a spray of water was redirected back to the area where he was working, or whether the hose was pointed directly to the side of the underpan where he was working. Immediately after the fire he appears to have been saying the former, whereas in evidence he said the latter. I suspect that on this point Mr Smith's memory has failed him. It may well be that it was his practice to employ one or other technique according to the precise area where the work was being carried out. I see no reason why directing a high-pressure hose at the central divider so as to create a spray of water, as opposed to a straight jet, might not have been a more effective technique in some situations. Since no-one has carried out any tests on this, it remains an area of uncertainty.

266. However, having watched and heard Mr Smith being cross-examined for several hours I formed the clear impression that he is a conscientious and highly competent craftsman. I have no doubt that he would have deployed the hose in the most effective manner for the work that he was doing at the time. There was absolutely no point in doing otherwise when the protection afforded by the water was of such importance. I did not get the impression that Mr Smith was a man who was willing to cut corners, or that Mr Marshall would have retained his services for so long if that was the case. Whilst I have expressed reservations about some of the evidence given by Mr Marshall, I do not have any reservations about reaching the conclusion that he is a highly competent and safety conscious operator who runs, as he said, a tight ship. There was no challenge to his assertion that the Defendants had an unblemished safety record for the 20 years prior to the fire. That sort of record is not likely to have been achieved by accident, and of course Mr Smith has been working for the Defendants throughout that period."

27. Ground 7

"The learned judge erred in law in failing to apply the principle exemplified by cases such as Drake v Harbour and Vaile v London Borough of Havering, namely that where a claimant proves a breach of duty on the part of the defendant and damage occurs which is of a kind that performance of the duty was designed to prevent, the court should (and ordinarily does) draw the inference that the breach of duty caused the damage."

The judge found no breach of duty.

28. Ground 8

"Instead the learned judge erroneously embarked upon an exercise of considering various alternative mechanisms by which the fire might have started and then selecting, as being most likely one that would not have been observed by a person keeping a fire watch during the cutting operations."

I do not understand this criticism. It was common ground between the experts, as recorded at paragraph 264 of the judgement, that the fire could have started as a smouldering fire in the mastic sealant which would produce very little outward sign of the smoulder in relation to smoke or odour. It was also common ground that this smoulder could have continued to develop underneath the surface of the combustible material such that it could not have been reached by the quenching water. It was further common ground that the smouldering could have progressed for longer than the fire watch and thereafter undergone transition to a flaming fire. The judge's conclusion as to the probable cause of the fire was thus consistent both with the agreed expert evidence and with his primary findings of fact, in particular his conclusion that Mr Smith was a careful and conscientious worker.

Conclusions

29. The difficulty faced by UMA on this appeal is that its case resolves to the simple but obviously flawed submission that the very occurrence of the fire in itself shows either that the quenching water cannot have been properly applied or that a proper fire watch cannot have been kept, or both. The latter point is doubly hopeless given that, as recorded by the judge at paragraph 186 of his judgment, Mr Walker did not at trial criticise the fire checks that were carried out after the hot work was completed. Mr Walker suggested that a dedicated fire watcher may have seen spatter being showered onto an area of the underpan which was unprotected by water and may have seen spatter getting into an area where it should not have been allowed to penetrate. A dedicated fire watcher might, he submitted, have seen things happening that might give rise to risk. The difficulty about these submissions, quite apart from the judge's finding that there was no requirement for a dedicated fire watcher, is that they ignore the judge's findings as to the unforeseeable nature of the process of initiation and development of the fire. Mr Walker pointed to the circumstance that the same work had been done for years without mishap and suggested that the fact that on this occasion a fire occurred calls for an explanation. If that is so, it by no means follows that the only explanation is one consistent with the precautions against fire having been applied with insufficient care. More generally, this approach is simply inappropriate where the judge has found that the fire was both initiated and developed in an unforeseeable manner. The judge decided that the agreed procedures simply did not cater for a very, very, rare occurrence. No-one had or should have appreciated the risk of a fire starting in this manner.
30. Mr Walker was, as I have already indicated, highly critical of the judge's conclusion as to the manner in which the fire started. Mr Walker submitted that it was "not permissible" for the judge to find that spatter entered at a point which could neither be observed nor prevented. Again this mischaracterises a more nuanced finding. The judge's finding was essentially that the diligent fire watcher would simply observe spatter being apparently quenched and falling harmlessly. This of course is because of the judge's finding, criticised by Mr Walker, as to the manner in which the fire started. In my judgement however that finding was open to the judge on the evidence and I do not think that he is to be criticised for embarking on the exercise which led to it. Indeed, given GM's case as to the care which was taken and the common ground in the expert evidence to which I have referred above, it was as it seems to me incumbent upon the judge to explore the question whether there were ways in which the fire could have started consistent both with GM's evidence and the experts' agreed conclusions.
31. I agree with Mr Hornett that unless the judge's conclusion as to the manner in which the fire started can be undermined, this appeal cannot succeed. This court will not lightly overturn a conclusion of that sort reached by a trial judge but for the reasons which I have already given I can see no basis upon which it can in fact be impugned. The same is true of the judge's conclusion that this was not reasonably foreseeable. The appeal must therefore fail. The judge's further conclusions that there was no want of the care required to be taken and that such want, if proved, would not have been causative, are not of course entirely independent of his conclusions as to the manner in which the fire started. It is right however to record that UMA has not in my judgment come close to persuading me that those further conclusions are capable of serious challenge.

Novae's appeal on costs

32. The judge was invited to deal with costs on the basis of written submissions. In those submissions neither UMA nor GM suggested that Novae should be deprived of any part of its costs. The only issue which they raised relevant to Novae's costs was as to who should pay them as between themselves, or in what proportion they should bear them on the unspoken assumption that Novae would effect a full recovery.
33. In his written judgment on costs handed down on 30 April 2012, without prior circulation to the parties in draft, the judge said this:-

"The costs of the Part 20 claim

22. I have already expressed my concern at the amount of Novae's costs. In addition, it abandoned/failed on two if its allegations of breach of warranty. I consider the cross-examination of the Defendant's witnesses on behalf of Novae went much further than was necessary. This is not intended as a criticism of Novae's legal advisers – for all I know, that may be what they were instructed to do. Nevertheless, I am satisfied that it prolonged the trial.

23. Whilst it is true that the claim against Novae was potentially very substantial – probably in excess of £4 million – the only point raised by Novae shortly after the fire was that the Defendant had not complied with the requirement to ensure that all inflammable materials in the immediate vicinity of the work were "covered and protected". If the two fire experts instructed by UMA and Novae were right about how the fire started, the scope of the issue of whether or not there was a breach of this provision was very narrow. It raised no points of law and the facts were not complex.

24. The point in relation to the lack of a fire watch was not raised until October 2008, and then only tentatively. As far as I can tell, the point about the absence of fire extinguishers first appeared in the pleadings – at any rate, very much later. The former did not succeed and the latter was abandoned. I regard both of those points as opportunistic.

25. In these circumstances I consider that Novae should be limited to recovering only 50% of its costs (as assessed), assuming that it is in principle entitled to recover costs from one or other of the parties."

The judge then went on to conclude that 75% of those costs should be borne by the Claimant and 25% by the Defendant. There is no appeal against that part of the judge's order.

34. Novae immediately asked the judge for permission to appeal against his costs order, protesting that it had been made on a basis not argued or advanced by any party and on which therefore Novae had had no opportunity either to be heard or to address argument. Novae also complained, inter alia, that it was wrong in principle for the judge to deprive it of 50% of its costs on the basis that he thought that the costs claimed were excessive, expenditure on an unreasonable or disproportionate scale being a matter for assessment.
35. The judge gave his reasons for refusing permission to appeal as follows:-

"1. Novae's primary defence was whether or not the material that was ignited by the molten cutting particles was "covered and protected" within the meaning of the warranty in the policy. It asserted in its opening that the Defendant's own account of its method of working effectively amounted to an admission of a breach of warranty.

2. However, Novae pleaded and ran two other alleged breaches of the warranty, neither of which was supported by its loss adjuster or the experts who investigated the fire. They raised issues of fact that were disputed. Novae's fairly lengthy cross-examination of some of the Defendant's witnesses contributed to the 2 day overrun of the evidence. One of these defences was abandoned at the end of the trial, and I rejected the other.

3. In these circumstances it seems to me that an "issue-based" costs order against Novae would not have been open to challenge. I did not make such an order because it would probably have been difficult and costly to identify and assess the relevant costs. So instead I considered that the appropriate solution was to reflect the outcome by disallowing a proportion of Novae's costs. I see no real prospect of showing that this was wrong in principle and I therefore refuse permission.

4. The precise proportion was a matter of overall impression and discretion (taking into account the fact that Novae was not being ordered to pay any part of the Defendant's costs).

5. The parties agreed that costs should be dealt with on paper on the basis of written submissions. The very substantial amount of Novae's costs (disclosed with those submissions) underlined the unfairness that would result to the other parties if an order in relation to Novae's costs was not made to reflect the matters I have mentioned."

36. Mr Hornett, who alone addressed us in opposition to Novae's appeal, accepted both that the judge was wrong not to have circulated his costs judgment in draft before handing down and that it was irregular to have denied Novae an opportunity to be heard if the judge was minded to deprive it of a proportion of costs in a hitherto unsuggested manner.
37. I also consider that it is plain that the judge erred in principle in reflecting his concern as to the level of costs incurred by Novae by making a reduction in the proportion of those costs which would be recoverable. That was a matter for assessment. The judge could of course have made clear his concerns for the benefit of the costs judge.
38. Mr Hornett nonetheless strove valiantly to uphold the judge's order on the basis, essentially, that the judge had rightly formed the impression that no stone had been left unturned by Novae. Novae had, he said, cross-examined on the fire watch point and the cross examination of Mr Smith in particular occupied 44 pages of the transcript. The only point upon which Novae had succeeded had been the question whether the combustible materials on the underpan had been both covered and protected at all times, so the judge was right to be critical of the manner in which it had conducted its case.
39. In a trial in which the principal combatants are competently represented the intervention of a third party on issues already canvassed can be an irritant, and I have little doubt that that is what happened here. I sympathise with the judge, who no doubt felt that on certain issues the ground had been more than adequately covered before Mr Philip Shepherd QC for Novae had his opportunity to contribute. But Novae faced a substantial claim and it was entitled to defend its refusal to indemnify. It allied itself with the Claimant in seeking, successfully, to refute its insured's case to the effect that the fire had started not in the screening house but in the scalping house through activities in which both UMA and GM were involved. This was an important point for Novae, as if the fire started in the screening house then it could demonstrate that the breach of warranty, if proved, was directly causative of the fire, whereas if the fire started in the scalping house that was not so. Mr Shepherd analysed for our benefit the time spent in cross-examination in a manner which he had not had the opportunity to do for the judge. Whilst obviously Novae's cross-examination of the witnesses must have extended the trial, it did not do so to any very substantial or disproportionate extent. The judge was evidently unimpressed by cross-examination directed to demonstrating that GM's employees were unaware of the terms of the burning and welding warranty in the insurance cover. Again, I can understand this, but Novae was entitled to make the point, for what it was worth, that if GM's employees were unaware of the term, they are unlikely to have planned to comply with it. Mr Shepherd also made the point that no-one at the time suggested that his cross-examination was inappropriate and that on occasion during it the judge asked his own supplementary questions.
40. It is plain both from the judge's costs judgment and from paragraph 5 of his reasons for refusing permission to appeal that his order was substantially influenced by his view that Novae had incurred disproportionately high costs. I have already indicated my view that in this respect the judge fell into clear error. Looking at the matter in the round I am satisfied that the judge's order is not sustainable. The only matter which gives me pause for thought is whether it should be remitted to the judge for reconsideration. However on balance I have concluded that the appropriate course is to substitute our own order. We have heard full argument on the question what is the appropriate order. In my judgment the matters identified by the judge do not justify depriving Novae of any part of its costs. Just as GM unsuccessfully raised at trial the major issue concerning the origin of the fire but was nonetheless awarded all of its costs, so too Novae should not in my judgment be penalised for having pursued some issues on which it lost or which proved peripheral.
41. I would therefore dismiss UMA's appeal but allow Novae's appeal. Subject to the further submissions of counsel, effect can be given to the success of Novae's appeal simply by setting aside the second paragraph numbered 4 in the judge's order of 17 May 2012.

Lord Justice McFarlane:

42. I agree.

Lord Justice Rix:

43. I also agree.