England and Wales High Court (Commercial Court) Decisions

IN THE HIGH COURT OF JUSTICE
QUEEN'S BENCH DIVISION
COMMERCIAL COURT

		Claim No: 2007 Folio 553 Royal Courts of Justice Strand, London, WC2A 2LL
		31/03/2009

Mr Justice Blair:

1. On the evening of 4 November 2005, the starboard leg of the oil rig "Cendor MOPU" was lost at sea off the coast of South Africa just north of Durban. At the time, the rig was being towed on a barge from Texas to Malaysia, with its legs in place and elevated in the air above the deck. The following evening, the remaining two legs fell off in quick succession. The experts are agreed that the loss occurred because of fatigue cracking caused by the repeated bending of the legs under the influence of the motions of the barge as it was towed through the sea. The cracks propagated to a critical size, at which point sudden fracture finally happened. The question is whether the loss is covered by insurance or not. The answer to the question depends largely on issues of causation.
2. The claimants are the owners of the rig. They provide onshore and offshore production facilities out of Kuala Lumpur, Dubai and Singapore. The defendant is a Malaysian company providing Takaful, that is to say insurance on Islamic principles. It covered this voyage under a policy of cargo insurance. The parties agree that the weather experienced was within the range that could reasonably have been contemplated. The claimants say that the loss of the rig was accidental, and so within the terms of the "all risks" cover. They submit that the proximate cause was the failure to ensure that adequate repairs were carried out at Saldanha Bay near Cape Town, where the tow stopped en route. The defendant says that the cause was "inherent vice" in the legs themselves, alternatively that their loss was an inevitable consequence of the voyage embarked upon. Since the term "all risks" covers risks and not certainties, the defendant says that it is not liable for the loss.
3. Before analysing the facts, I should say something about the burden of proof, about which the parties are in agreement. The insurance policy incorporated the Institute Cargo Clauses (A) ("ICC(A)") of 1 January 1982, the insured perils being "all risks" subject to specified exclusions, including the inherent vice exclusion which is relied on by the defendant insurer. The cover being all risks, the claimants are only required to show that the cause of the loss was accidental, or to put it another way, that the loss was not inevitable. Once the claimants discharge that burden, the burden is then on the insurer to make out the exclusion relied on. Quantum is not in dispute.

The facts

The acquisition of the rig

4. Only two witnesses of fact gave evidence, both on behalf of the claimants. Most of the evidence has been in the form of documentary and expert evidence. The facts as I find them are as follows. The rig is technically known as a "self elevating mat supported jack-up rig". It was designed and built at the Bethlehem Shipyard in Singapore in 1978 for use in the offshore oil industry. It consists of a watertight working platform called the jackhouse, which can be jacked up and down the legs according to the sea depth at the in-service location. The legs are made of welded steel cylindrically shaped with an outside diameter of 12 feet and a length of 312 feet. Each of these steel tubes weighs 404 tons. There is a mat at the bottom of the legs that sits on the sea floor when she is in service.
5. The jacking system works by engaging steel pins into pinholes in the legs. Each leg has 45 sets of six pinholes at six foot intervals. To describe them as "pinholes" does not quite do credit to the holes in question, which are 16 inches wide and 10 inches high. Discontinuities such as these holes cause stresses in the steel structure to be increased locally, and since sharp corners tend to give rise to high stress concentrations, the corner of each pinhole was designed to incorporate a roughly circular hole of an inch and a half in diameter. The effect of such "rounding" is considerably to reduce stress levels at the corners. The state of the corners, and in particular the presence or otherwise of cracking at the corners, has been a key issue in this case.
6. Prior to its acquisition by the claimants, the rig was called the "Odin Liberty". It had something of a chequered history. In May 1996, whilst underway from Singapore to Nigeria on board a barge, the port leg broke off at approximately 20 feet above the jack-house. Coincidentally, this incident happened not far from where the present incident took place, and on that occasion the rig put into Durban for repairs. Following service in Nigeria it must have crossed the Atlantic at some stage, because from about July 1999 it was laid up at the Gulf Copper Shipyard in Galveston, Texas. It was purchased by the claimants on 20 May 2005 for conversion into a mobile offshore production unit (MOPU) for use in the Cendor oil field some 200 kilometres off the coast of East Malaysia on a contract with Petrofac Malaysia Ltd. It was not in class at this time.
7. At this point, I should introduce the two factual witnesses called by the claimants. The first was Mr Keith Walker who is (and was at the time) the Group General Manager, reporting to Mr Clint Elgar who is Group President. The second was Mr Mike Ooley who was the project manager, and indeed still has responsibility for the rig, which I may say has functioned satisfactorily since the legs were replaced. Both are seasoned professionals, and I found them to be reliable witnesses. Their evidence is that the claimants are not in the business of marine transportation, and engaged outside experts to organise the transit for them. The rig was to be dry towed aboard a barge to Lumut Port, Malaysia, where conversion work was be done on it prior to its deployment in the oil field. A company called Proceanic Engineering Services Pte Ltd was engaged as project management consultant. The arrangements were going to have to be approved by a marine surveyor. Proceanic recommended Noble Denton as having a good track record acting on jack-up vessels, and as such likely to be acceptable to most underwriters. And then there is something that has featured importantly in the evidence in the case. Consultants were required to perform calculations as to the structural integrity of the legs, and a company called Viking Systems Inc was engaged for that purpose. The scope of Viking's work encompassed both transport and in-place stress analyses (the latter being required after arrival in Malaysia in respect of the contract with Petrofac).
8. Though there was a pre-purchase inspection, the claimants appear to have been largely content to purchase the rig "as was", reflecting the brisk demand for rigs in the oil industry at that time. The rig had been laid up for nearly six years. One of the defendant's experts (Mr Jeremy Colman) says that the ABS survey carried out after purchase presents a picture of a decayed and dilapidated structure. The claimants point out however that the legs were subject to ultrasonic gauging for classification purposes conducted from inside the legs, and that no serious wastage to the legs was found. Mr Ooley told me that the overall impression was that the legs were in "good condition". In any case, the expert evidence shows that what really matters when it comes to metal fatigue is not dilapidation in a general sense, but cracking, here cracking particularly in the pinholes in the legs, and I will come back to the cracks that were found and what was done about them.
9. At the outset, consideration had to be given as to how the rig was to be transported, and it is plain that this was a tricky question. Mr Ooley accepted that there is a history of this type of leg failing in the course of transit. In non-technical terms, the problem is the degree of stress that the legs are subjected to as the barge pitches and rolls in the sea. This is easily envisaged when one appreciates that the legs were hollow steel columns sticking into the air over three hundred feet high. Mr Ooley told me that the predominant jack-up design now is a lattice-type structure, which has different characteristics from the cylindrical design in the present case.
10. There were alternatives to transporting the rig with its legs intact and in the air. The legs could have been removed and carried on the barge, or they could have been shortened (or "cropped" as the witnesses put it). On 2 June 2005, Mr Bob Harris of SPS emailed the claimants with his views about the transport arrangements. SPS is part of Swire which as Mr Walker explained was consulted from time to time during the early stages of the project on an ad hoc basis. Mr Harris said that given the method of transport and the time of year in which the tow would be undertaken, the "consensus of opinion" was that the legs would have to be shortened. He cited a number of factors in this respect. It was, he said, generally accepted that stress limits within the legs would be exceeded with a 20 degree roll with a period of 10 seconds. Given the tow route, and the known weather conditions at and around the Cape of Good Hope, it was highly likely (he thought) that these criteria would be met. He said that a motions analysis with a model of the columns in their current condition would almost certainly confirm this but, "Why go to the time and expense confirming without doubt what we already know?" The recommendation that he conveyed to the claimants was that "you bite the bullet and get Proceanic to shorten the columns … without going to the additional expense of analysing leg strength further at this time".
11. This recommendation did not it is fair to say meet with any great enthusiasm. Proceanic responded that removal of sections of the legs at Galveston and their reattachment in Malaysia would involve a considerable amount of work, the cost of which was assessed as being about US$2m. There was then some further comment, but the exchange of emails was brought to an end by the claimants' Group President on 7 June 2005. He said that they were not to get the yard in Lumut involved in the cutting of the legs, and that the "objective here is to do everything we can to avoid cutting the legs, ie brace, stiffen etc". (Neither of these latter possibilities appears in the event to have been satisfactory.)
12. Based on this and an earlier email of 23 May, the defendant submits that the reality is that by the end of May 2005 the claimants had made a policy decision that the rig would be transported with its legs in place, and that this decision remained operative at every stage thereafter. No doubt, it is said, they would have given way had Noble Denton insisted on removing the legs, but subject to that, there was no prospect at all of the claimants taking the legs off. I think that this is putting it too high. I have no doubt that the claimants' strong preference was for the legs to remain uncut, and this is understandable because, apart from the expense, shortening or removal would have entailed the reconstruction of the legs when the tow reached Malaysia. But both Mr Walker and Mr Ooley told me that the claimants had not closed their options out at this stage, and I accept their evidence in that respect.
13. What is, as it seems to me, is of more significance in this "to cut or not to cut" exchange of emails, is the contemporaneous view of people who are independent of the parties to this litigation, and who were aware of the issues involved in transporting this kind of rig. Before any calculations were done, in broad terms they correctly predicted what would happen, where it would happen, and why it would happen. With the benefit of hindsight, a good deal of time and trouble would have been avoided if their recommendation had been followed, and the claimants now accept that.
14. On 28 June 2006 the claimants entered into a charter with Seaspan (Cyprus) Ltd for the carriage of the rig to Malaysia on an unmanned barge. The "Boabarge 8" was later designated for these purposes. The freight was agreed at US$3,495,750, with provision for demurrage at a rate of US$11,900 per day. In due course the "Atlantic Hickory" was nominated as the tug for the tow. With those arrangements out of the way, preparations for departure proceeded apace.
15. On 9 July 2005, Seaspan sent Mr Bader Diab of Noble Denton what was described as "preliminary engineering" for the Boabarge 8, including a sea motions analysis, together with weather data for the intended route. Such data, as will be seen, was essential for the preparation of fatigue assessments regarding the legs en route. Other essential data in the form of "design constraint (worst case)" response amplitude operators for the transit was sent by Seaspan's naval architects (Grand Marine) on 19 July 2005. On 18 July 2005, Noble Denton confirmed that it had inspected the rig and found her ready for the wet tow to the loadout location, and that preparations for the "Boabarge 8" were also complete. The loading of the rig onto the barge was accomplished at Pier 10, Galveston, between 22 and 23 July 2005.

The policy of insurance

16. Meanwhile, the claimants sought insurance through brokers called Insfield Insurance Brokers Sdn Bhd. A cover note issued by the defendant insurers was sent to them on 20 July 2005, and a placement slip, schedule and certificate followed. It is common ground that it was a condition of the policy that Noble Denton approved the arrangements for the tow. The terms are not in dispute and so far as material are as follows (the references to RM are to Malaysian ringgit):

"Cover Note

Period of Cover 20^th July 2005 to 30^th November 2005

Description of Risk On Cendor MOPU …

For commencement of loading operations in Galveston Texas until completion of discharge in Lumut Port, Perak, Malaysia.

Total sum covered: RM38 million (Equivalent to USD10 million)

Placement Slip

…

ENDORSEMENTS 9. Institute Cargo Clauses (A) 1.1.82

…

DEDUCTIBLE USD1,000,000 or equivalent to RM3,800,000 …"

17. Further, by reason of the incorporation of the ICC(A) terms, the policy of insurance incorporated the following terms:

"RISKS COVERED

1. This insurance covers all risks of loss of or damage to the subject- matter insured except as provided in Clauses 4, 5, 6 and 7 below.

...

EXCLUSIONS

4. In no case shall this insurance cover:

…

4.4 loss damage or expense caused by inherent vice or nature of the subject matter covered."

The policy was therefore an "all risks" policy, excluding (among other things) "inherent vice".

Pre-shipment preparations

18. The claimants submit that because everybody knew that the legs were at risk of fatigue, a great deal of care was taken to assess their fatigue life before Noble Denton (whose mandate was to act in the interests of the defendant) was willing to approve the tow to depart at all. I would certainly accept the premise of this statement, but the proposition itself has to be examined with some care. The formal process by which the fatigue life of this rig was assessed began when Viking sent a methodology for their calculations to ABS and Noble Denton for review and approval in early August 2005. At about the same time, National Marine Services performed a liquid penetrant inspection and visual check of pinholes at the 24 foot level on each leg. This was within what has been called the "danger zone" so far as stress during transit was concerned. No surface cracks or imperfections were found. Seaspan was also providing weather data to be input into the transit calculations, which was evidently the subject of some debate.
19. There were in fact a number of fatigue analyses provided by Viking, initially in the form of "simplistic" fatigue assessments. (As I explain below, the term "simplistic" has to do with the data upon which the analysis is based, which does not include sea states sector by sector.) To anticipate the expert evidence, fatigue life is assumed to be expended when a damage ratio of unity (1.0) is achieved. The claimants' expert Dr Aston explained that the maximum allowable value for critical structures (giving a safety factor to allow for uncertainties) published by DNV (Det Norske Veritas) is 0.5 where inspection and repairs during the operation are planned, otherwise 0.33. On 20 August 2005, Viking submitted fatigue calculations of the pinhole subject to the highest stress showing a total damage ratio of 0.26, which was well within these parameters.
20. Despite this, when voyage approval was requested on 23 August, Noble Denton's draft recommendations stated that the rig did not have adequate fatigue life for the transit. As the defendant points out, a couple of hours later "do not" had become "may not", and the draft recommendations were toned down. What happened in between is not known. The claimants are not in a position to say exactly what material was available to Noble Denton, since day to day liaison was left to Proceanic, and they point out that these recommendations were produced in draft, and that discussion and change was not surprising.
21. Gulf Copper conducted an inspection of the legs, and I agree with the claimants that though not explicitly stated in the report, this was clearly an MPI (magnetic particle) inspection. According to Gulf Copper's report of 23 August, after thorough inspection of 18 pinholes on three levels up from the jackhouse, no imperfections were found on the bow or starboard leg, and on the port leg there were "five minor imperfections on the second row [which was probably the 12 ft level] in three pinholes. The minor infractions were repaired, ground and welded, and re-tested and found to be satisfactory." The defendant describes the repairs at Galveston as "minimal in the extreme". The claimants take issue with this description, pointing out that the inspections and repairs were as mandated by Noble Denton, and that workmen inspected the legs using rope access. I think it is fair to say that although not "minimal in the extreme", the repairs were relatively minor. Mr Ooley confirmed that there was no attempt to "re-set" the fatigue life of the legs.
22. The Noble Denton Certificate of Approval was issued at 18.30 on 23 August 2005. It recommended among other things that the barge roll motions should be kept under 5 degrees. With regard to the legs it says as follows:

"1 The ODIN LIBERTY legs have been the subject of a simplified fatigue analysis. Taking into account the fact that some of the fatigue life has already been used in the rig's previous history, and the uncertainty associated with the simplified fatigue analysis for the wet tow from Galveston to Lumut, it is possible that the legs in way of the pinholes may not have sufficient fatigue life to undertake the full tow to Lumut.

2 Since the fatigue analysis shows possible damage, it is required that the legs be re-inspected at Capetown for crack initiation in way of the six levels of 'pinholes' above the mat. Capetown is, broadly speaking, the half way point and remedial work could be undertaken should it be found necessary. Inspection should be using eddy current or equivalent NDT [non-destructive testing] technique."

With approval from Noble Denton thus in hand, the tow sailed away from Galveston on the same day.

The passage from Galveston to Saldanha Bay

23. As it passed through the Gulf of Mexico, the tug managed to avoid Hurricane Katrina. In fact, no problems of an exceptional nature appear to have been encountered en route. During the course of the tow, Viking (as the claimants put it) "progressed work" on the transport strength and fatigue analyses, which went through a number of revisions. It is clear that these were the subject of discussions with Proceanic and Noble Denton.
24. The last simplistic fatigue assessment was revision D, which was issued by Viking on September 20, 2005, that is some four weeks after the tow had sailed. According to the summary, the "fatigue analysis results show that the highest fatigue damage to the leg occurs at a pinhole corner detail. The highest ratio is approximately 0.68 and as such the legs are expected to withstand the fatigue damage during the transportation from the Gulf of Mexico to Malaysia via the Cape of Good Hope". It seems that the 0.68 figure came from an earlier draft, because the maximum damage ratio was estimated at 0.89, which is in fact the final number which these analyses produced. (It may be compared with the first total damage ratio calculated by Viking of 0.26 on 20 August 2005.)
25. The assessment states that, "Due to the inherent uncertainties with simplistic fatigue analysis it is recommended that a spectral fatigue analysis be carried out. The spectral fatigue analysis should be based on actual wave headings, wave periods, and wave height experienced along the route". In fact, a spectral fatigue analysis had been under discussion since the end of August, the principal reason according to the claimants being to "benefit the in-place analysis that Viking was already working on, since GPS needed to demonstrate a minimum 10 year in-place life expectancy for the Petrofac contract".
26. Whatever the reason, the necessary calculations must have been done by this stage, because Viking's spectral fatigue analysis was dated September 21, 2005, the day after revision D. The summary begins identically to the summary contained in revision D, saying that, "The fatigue analysis result shows that the highest fatigue damage to the legs occurs at a pinhole corner detail". But the highest ratio is now stated as approximately 2.13. The words about the legs being expected to withstand the fatigue damage during the transportation from the Gulf of Mexico to Malaysia via the Cape of Good Hope are omitted. Clearly on this number, such a statement would have been untenable. Instead the report recommends that "the legs be inspected for presence of cracks prior to entering the Cape Good Hope portion of the voyage". A considerable number of pages of calculations follow showing how that result was reached.
27. Noble Denton had been provided with revision D of the simplistic fatigue analysis dated 20 September, but it appears that they were not provided with the spectral fatigue analysis, which is something of a mystery. Mr Ooley mentioned his "sort of assumption that Noble Denton was privy to this". He thought it "just was not transmitted formally from GPS to Noble Denton". However to jump forward in time, when Noble Denton issued the Certificate of Approval for the tow to proceed from Saldanha Bay on the final sector of the voyage on 23 October 2005, the recommendations say that the rig's legs had "been the subject of a simplified fatigue analysis" showing "possible damage". That implies that Noble Denton did not see the spectral fatigue analysis. A reference to "possible damage" could clearly not have been squared with an analysis which predicted the consumption of the fatigue life more than twice over.
28. Ms Claire Blanchard, counsel for the claimants, submits that the Viking spectral fatigue analysis is a "red herring", and irrelevant to the issues which the Court has to decide. In formal terms that may be correct, but it is an unsatisfactory feature of this case that the crucial revision of the fatigue life did not find its way to the marine surveyors. And at the least, as with the "to cut or not to cut" emails, one finds again that the failure of the legs was predicted before it happened.

The stopover at Saldanha Bay

29. On 10 October 2005, that is about seven weeks after leaving Galveston, the tug and barge arrived at Saldanha Bay, just north of Cape Town. Although as I have said, there do not appear to have been any exceptional events on the voyage from Galveston, it is clear that there was a considerable degree of cracking around the pinholes by the time of arrival. Because of the way the claimants put their case, the steps that were taken at Saldanha Bay to repair the rig are central, and I shall deal with them after analysing the expert evidence.

The loss of the legs

30. On 28 October 2005, the tow departed Saldanha Bay with two tugs in attendance, the "Smit Amandla" having joined the "Atlantic Hickory", in accordance with Noble Denton's stipulation, for the passage through the rough waters round the Cape. As planned, on 3 November 2005 the "Smit Amandla" let go, and the "Atlantic Hickory" continued on its own. North of Durban on 4 November 2005 at 20:00 hours, the starboard leg of the rig broke off at the 30 foot level. The following day at 20.50 hours, the forward leg broke off at the 30 foot level, and about half an hour after that, the port leg broke off at the 18 foot level. The legs all fell into the sea, fortunately without damaging the barge too substantially. The tow limped into Richards Bay on 11 November, departing three days later. In Mr Ooley's words, it then continued without further incident, arriving off Lumut on 9 December 2005.

The parties' pleaded cases on causation

31. The case originally pleaded on behalf of the claimants in paragraph 14.1.1 of the Reply was that the loss of the legs (or more precisely the starboard leg which went first) was caused by a combination of the weather including the waves and wind, the handling of the tugs following departure from Saldanha Bay, the heading and speed of the tugs and the consequent effect on the roll and pitch of the barge. There is no longer an issue about the behaviour of the tugs. Following the exchange of experts' reports, the Reply was amended at the beginning of the trial to plead a primary case that the proximate cause of the loss was inadequate repairs at Saldanha Bay. The amendments further identify as the cause of the loss as follows:

14.1.1.4 The repairs at Saldanha Bay being inadequately performed and/or inadequately engineered as particularised in the expert reports of John Aston and Robert Andrews. Specifically:

14.1.1.4.1 The object of the stop over at Saldanha Bay was to ensure that they would survive the second leg of the passage. Self evidently, that object was not achieved and the repairs were in fact inadequate to achieve that purpose; alternatively

14.1.1.4.2 The repairs were inadequately performed in that they did not succeed in removing all cracks and/or were not painted or dressed or coated with a temporary protective compound or grease and/or did not achieve DNV Class C and/or had the features particularized in sub-sub-sub-paragraphs 27.2(b)(ii) to (iv) of the Amended Defence and/or did not achieve the shape designed by Viking and/or did not "re-set" the fatigue life of the legs and/or did not return the legs to at least as good a condition as they were on departure Galveston; alternatively

14.1.1.4.3 The repairs were inadequately engineered in that, in so far as in known, they were not engineered to achieve the level of a "machined" curve.

The claimants' case is further that had adequate repairs been carried out, the loss of the legs was not inevitable.

32. The defendant's case was also subject to a late amendment. In its final form, it is said that the damage was caused by inherent vice in that the legs were not capable of withstanding the normal incidents of the tow, as demonstrated by the fact that they failed in weather conditions within what could reasonably have been expected. "Adequate" repairs (in the sense of repairs resetting the fatigue life of the legs) were not intended or attempted, and were not practically possible. Even if such repairs had been successfully performed, there would have been insufficient fatigue life in the legs to complete the voyage to Lumut. In the alternative, the defendant says that the fractures to the legs were inevitable.

The expert evidence

33. Four experts gave oral evidence at the trial. For the claimants, these were Dr John Aston (naval architecture), and Dr R.M. Andrews (metallurgy/fatigue). For the defendant, these were Mr Jeremy Colman (naval architecture), and Dr Jonathan Sykes (metallurgy/fatigue). It may be noted that Dr Sykes inspected what was left of the legs shortly after the rig arrived in Lumut, and was able to obtain samples. Each of the experts is experienced and well qualified, and they were able to reach a measure of agreement at their pre-trial meeting. There were also weather experts on each side, but because the claimants accepted that the weather experienced was within the range that could reasonably have been contemplated (albeit their expert puts it at the upper end), it was possible to proceed on the basis that their respective views could be treated as establishing an upper and lower range, and oral evidence from them was unnecessary.
34. It is common ground that the cause of the loss of the legs was fatigue cracking, which was caused by the repeated bending of the legs under the motions of the barge in the sea. The cracks propagated until they reached a critical size at which time, with the application of sufficient stress, the legs failed. The starboard leg failed first followed by the forward leg, and finally the port leg. It is also common ground that once the first leg failed, the effects operating on the remaining two legs would have been exacerbated. Nor is it in dispute that the second sector of the voyage in which the loss occurred included the seas which would be expected to be the roughest of the voyage, on account of the southern ocean depressions which circle the globe between Antarctica and the Cape of Good Hope.
35. The process of fatigue failure is well understood in the scientific community, and was helpfully described in the expert evidence. Taking Dr Sykes' summary, fatigue is a progressive cracking mechanism caused by repeated or fluctuating (cyclic) stresses at a level lower than that required to cause fracture upon a single application to an un-cracked component. In the present case, such cyclic stresses would have been generated in the legs of the rig by the height and direction of the waves as they hit the barge. In general, higher motions mean more stress. It is common ground that the fatigue damage caused varies with the cube of the stress range applied, whereas the stress range is proportional to the wave height. Thus, small increases in wave height can lead to large increases in damage. And as Ms Blanchard points out, the converse is also true.
36. The evidence is that there are generally considered to be three stages to the fatigue failure of any given component—fatigue crack initiation, fatigue crack propagation and final fracture. Fatigue crack initiation is the term given to the early nucleation of a fatigue crack. Fatigue cracks invariably initiate at regions of stress raising features, where locally increased stresses are concentrated, such as a corner or notch. In this case, the corners of the pinholes would have acted as stress raising features. In general, the smaller the radius of a corner the greater the magnitude by which the stress is concentrated. The corner of each pinhole of the rig, as I mentioned earlier, incorporated a roughly circular hole of an inch and a half in diameter in an effort to reduce such stress. In terms of the greatest concentration of stress while the rig was under tow with its legs in the air, it is common ground between the experts that this was at the 30 foot level.
37. Fatigue is often categorised as being either "high cycle" or "low cycle". High cycle fatigue generally involves a very large number of stress cycles at levels well below the yield strength of the material in question (here, steel). Put differently, the deformations generated under high cycle fatigue conditions are "elastic", that is to say they are recoverable. Low cycle fatigue is usually associated with less than 10,000 cycles to failure and involves "plastic", that is to say permanent deformations. Consequently, they occur at stress levels in excess of the material yield strength. Or to put it in the more colloquial language of Mr Luke Parsons QC, counsel for the defendant, once you apply a yield stress to something, you only need to apply that 10,000 times for it to crack. Dr Sykes' evidence is that the very high localised stresses that were predicted at the pinhole corners would have caused cracks to propagate by a mechanism of low cycle fatigue. Under those conditions, large fatigue cracks could have been generated under the influence of a relatively small number of stress cycles. The view of Dr Andrews (whose explanations I also found most helpful) was not greatly different, in that assuming a stress at the pinholes of about 500 or 600 Mpa, he would have expected to get a failure after 20,000 to 30,000 cycles. (For present purposes a megapascal or Mpa is the unit for stress.)

Fatigue life

38. The expert evidence has been concerned in particular with the rig's fatigue life at various points in time. The fatigue life of a component is usually defined as the number of stress cycles that it can withstand before failure occurs. The experts are agreed that fatigue performance is measured using "S-N" curves in which the stress range, S, is plotted against the number of cycles to failure, N. S-N curves are fatigue life curves that are derived by the fatigue testing of small specimens in test laboratories. In these fatigue tests, samples are subjected to repeated (cyclic) loading at various stress magnitudes, and the number of cycles required to cause fatigue failure is determined and plotted onto a graph. As Dr Andrews explained, the basic fatigue design S-N curves are usually generated using constant stress cycles. To quantify the effects of stress under actual conditions, Miner's rule assumes that the loading has been broken down into a spectrum of stress ranges and that the number of occurrences of each stress range is known. Expressed mathematically, if the summation or damage ratio D is less than unity—that is to say less than 1—failure is not predicted to occur. For values greater than unity, failure is predicted to occur. As Dr Andrews makes clear, the damage summation is not a sharp failure criterion—it is not the case that D = 0.99 represents completely satisfactory performance or that D = 1.01 represents failure. This is another way of saying what is common ground between the parties, namely that the fatigue life expenditure produced by these calculations is of an indicative nature only. It is a predictor of success or failure, and it is important to keep that limitation in mind when approaching this evidence.
39. The legs of this rig were unusual in that the fatigue life was controlled by the pinhole corners where there is a high stress concentration. DNV C Class was used by Viking to assess the pinhole corners, and the naval architects in their reports set out the results of a spectral fatigue analysis similar to that done by Viking in September 2005. The term "spectral" here refers to the spectrum of sea states by geographic area. As Dr Aston explained, this method is more involved than the "simplistic" fatigue analysis carried out by Viking prior to the rig's departure from Galveston on 23 August 2005. It calculates the likely distribution of stress ranges based on actual wave headings, wave periods, and wave heights to be experienced along the route derived from long term wave statistics of the different sea areas passed through on the route.
40. The purpose of these analyses is to assess how much of the fatigue life of the legs at the pinhole corners would have been expected to have been consumed during the voyage from Galveston to Lumut. At trial, much of the debate between the experts turned on these analyses. The differences in their results depended partly (but not entirely) on the data inputted, and in particular on the appropriate "motions analysis". As I have indicated, the motions analysis is to do with the state of the sea in the various sectors of the planned voyage. This in turn is largely based on hindcasts produced by the weather experts (a weather hindcast compared with the more familiar weather forecast is based on historic data). The significance of this input is that the action of the waves, and in particular the height of the waves, was responsible for the cycle of stresses imposed on the rig during its voyage. That determines how much stress there was, and how much of the fatigue life it would eat up.
41. As I have explained above, fatigue life is assumed to be expended when a damage ratio of unity (1.0) is achieved. If the summation or damage ratio is less than unity, i.e. less than 1, failure is not predicted to occur. For values greater than 1, failure is predicted to occur. In summary, where the evidence got to in terms of the prediction of the likely degree of fatigue life consumption in respect of this voyage was as follows. As already stated, Viking's September 21, 2005 spectral analysis produced a figure for the consumption of fatigue life of 2.13. It is agreed between the experts that this was an underestimate, though there is a dispute as to what degree it was underestimated. To quote the defendant's closing submissions, the claimants' expert Dr. Aston's estimate for the consumption of fatigue life is 2.90. Ms Blanchard emphasised that this was a "maximum" figure based on built in assumptions, and that Dr Aston was not stating his view that such assumptions were likely to eventuate. Accepting that submission, I take Dr Aston's position from his supplemental report, in which he explains that after various calculations: "The net effect is that I am now getting a fatigue life damage ratio (based on DNV C Class performance) of the order of 2.9 with the higher motions as compared to about 2.4 before [in his original report] – i.e. an increase of about 20%". It is further to be noted that his oral evidence was that increasing the Viking calculation by 30% in accordance with his preferred motions input gave a consumption of 4.70. The Defendants' expert Mr. Colman's estimate was in excess of 5.00. However the defendant submits, in my view correctly, that the crucial point is that each spectral analysis predicts failure. These are all indicative figures, and they are all subject to assumptions, but they all predict failure.
42. For reasons I shall explain in due course, in my view the precise differences between the calculations of Dr Aston and Mr Colman are not decisive. I must say however that I think there is force in the defendant's submission that the process by which Dr Aston reaches his figure of 2.90 is not entirely evident. (In that regard, I have not overlooked some further calculations he produced shortly before giving evidence.) Specifically as regards motions analysis, he told me that he did not trust motions analysis in roll. He took the view that Mr Colman's motions are far too high for a barge. However he found it difficult adequately to explain why his reports use the Viking motions with an uplift of 30%, or to explain his results compared to those of Mr Colman when he used his own in-house programme for the calculations. Despite Dr Aston's undoubted experience as regard barges, I think it is fair to say that Mr Colman was the more practical of the two witnesses, and I generally prefer his evidence save where I indicate to the contrary. That includes the dispute between them as to the effect of "roll damping" and "wave spreading" in this context.
43. Dr. Aston's fatigue consumption estimate of 2.9 is affected by his use of a "reduction factor" to allow for the fact that a detail is under compression. The effect is to reduce his fatigue estimates by one half. The defendant submits that this factor was wrongly applied, with the result that his fatigue damage consumption comes up to 5.8. It is submitted that the reduction factor should not be used where significant residual stresses exist in the material, and that such stresses would have existed in way of the pin-holes due to the welding and grinding repairs performed at Galveston and Saldanha. It is also submitted that a reduction factor is of no relevance where low cycle (i.e. yield stress) fatigue is operating, and that such fatigue would have been inevitable in the waters off the Cape. However on the reduction factor point, Dr Aston has the support of Dr Andrews who said that this approach has been recognised and accepted in the industry for over 30 years. I also prefer Dr Aston's evidence in this regard.
44. Each of the estimates of fatigue consumption assumes that no fatigue life had been expended at the time the rig left Galveston. It is however as I have mentioned the case that she had suffered a previous leg failure due to fatigue cracking in 1996, also off the eastern coast of South Africa. Following repairs in Durban, the rig worked in Nigeria and was eventually transported to Galveston. The defendant submits that it is likely that a proportion of the fatigue life of the legs had already been expended when the claimants purchased the rig. Indeed, this is described as a "fact" in the Noble Denton Certificate of Approval. Dr. Aston accepted that there would have been some consumption of fatigue life. Dr. Andrews considered that an allowance of about 0.2 to 0.3 of a fatigue life would be appropriate to take account of it. Ms Blanchard objects in her supplementary closing submissions that his evidence was to the effect that to allow for any previously expended fatigue life, he would want the calculated damage sum to be down to 0.2 or 0.3. Whilst this may be the correct way of looking at his evidence, it simply goes to underline how far the calculations which the court has been provided with differ from what would be an acceptable number. In any case, I should say that the reason why one would ascribe what might seem a relatively small number to the expended fatigue life of the rig (given its age and general decrepitude) is, as Dr Andrews put it, that "mileage equals cracking at the end of the day", which summarises the expert evidence as I have understood it.
45. Also of significance in my view is the fact that Dr. Aston accepted that his model could not explain the 400mm crack shown in a photo taken by Dr Sykes in December 2005 of the stump of the port leg. Dr Sykes maintained that the crack was not caused by the trauma of the leg being broken, because it shows no deformation. If so, it is visual evidence of the actual extent of cracking in the legs at the time of failure. To put this crack into perspective, it is agreed that the remote stress necessary to break a leg containing a 120mm fracture is 126Mpa.

The timing point

46. Both counsel encouraged me (based on the case law) to take a common sense approach to the causation issue. One of the claimants' "common sense" points is the fact that the legs survived the first stage of the tow to the half way mark at Saldanha Bay. This it is submitted is an important piece of "real life evidence" to be brought into account in considering whether the legs could have survived the second stage of the tow. In Dr Aston's view, one can answer the question whether the legs could have survived the second stage simply by asking (1) whether the legs could have been put in a condition at Saldanha Bay which was at least as good as the condition they were in at Galveston and (2) whether the weather on the second stage could have been no more damaging than the weather on the first stage. In Dr Aston's view, the answer to both questions is clearly yes and on that basis, the legs could have survived the second stage.
47. It is not in dispute that the repair matrix at Saldanha Bay shows widespread and severe cracks at all levels between -6 feet and 30 feet, save for the 18 feet level—there is an issue as to whether this level was inspected. It is common ground that this was the part of the legs most prone to fatigue damage. The matrix shows over 200 hundred cracks in total. Why then, say the claimants, had the legs not broken already as predicted in the various spectral fatigue analyses, particularly those of Mr Colman at the top end of the range?
48. The evidence is that a developed crack would not, on its own, have been sufficient to cause a leg to come off. That requires, in addition, a "leg breaking" or "final straw" stress, that finally fractures the weakened steel. I found the evidence of Mr Colman convincing in this regard. He told me that:

"When we have these fatigue cracks, they're 100 or 200 millimetres long – that's 4 or 8 inches – but remember that we have a leg which is 12 feet in diameter, a circumference of about 40 feet. So even quite a lot of these little cracks still leave a very large amount of good steel an inch and a half thick. This isn't light plate; this is very heavy steel, and that's an enormously strong structure. So you've got to catch it just right, if you want to make it actually fail all the way round. I'm not surprised that it takes special event to make it go, but I think it was very lucky to have survived the first leg, as indeed the speed with which it failed during the second leg showed".

49. I asked him to explain further what he meant by "catching it just right" and he said:

"Just right in terms of roll angle, the direction of the motion and the amplitude of the motion and the presence of a crack of the right orientation in the right place, I suppose. Perhaps it's worth bearing in mind that the legs aren't just rolling from side to side or pitching backwards and forwards. They're getting a combination of those motions. So they're going around in great circles, or ellipses in different directions, and the stressing that is caused by that is complex. So somewhere all of the ingredients have to come together, but I think once you have used up a lot of fatigue life and you have cracks everywhere, then all you need is probably the 2, 3, 4-metre sea states that the Cape waters can provide."

The agreed range of wave heights demonstrates that waves in excess of 3 metres were in fact regularly experienced during the second stage of the voyage as the rig navigated the waters of the Cape.

Conclusions as to fatigue life

50. It is no disrespect to the experts, each of whom in my view was of considerable assistance in this case, to say that the numbers can be run almost indefinitely to produce figures for a consumption of fatigue life all of which would be within the bounds of plausibility. I am satisfied that these are inherently indeterminate calculations, the main purpose of which in this case would (or should) have been to predict whether the voyage could safely be undertaken with the legs attached to the rig. As it happens, the only contemporary calculation was that of Viking, which showed that it could not. And it is common ground that the Viking figure of 2.13 is in fact too low. The maximum figure of the claimants' own expert, Dr Aston, is 2.90. So far as it is necessary to reach any conclusions, I find that even this is probably too low. Equally, I have concluded that Mr Colman's original figures of upwards of 5.00 are too high. I refer to my findings above in relation to the details. A reasonable figure is probably somewhere between Mr Colman's estimate and that of Dr Aston. It follows that I do not accept the suggestion in Mr Colman's second report that the figure should be of the order of not less than 10, and agree with Ms Blanchard's observation that this further number only goes to illustrate the sensitivity of the analysis to the input factors.
51. One reason for adjusting Mr Colman's figures downwards relates to the routing. The tug's daily position reports contain a map showing a snapshot of the tow's position. The report for 14 September 2005 (which is the first report with a map showing the Cape), together with all subsequent reports, show the route as being via the Mozambique Channel, which would be more sheltered than the originally projected direct route across the Indian ocean. Moreover, the telex daily position reports sent at about the time the tow departed Saldanha Bay show that it was intending to take a route which avoided the worst weather. Whilst Mr Ooley (in characteristically honest fashion) said that when the tow left Saldanha Bay it was still the intention to proceed to Malaysia by the original direct route, there was no reason for him to be told about changes in the route. It would have made sense for the tug to proceed by the Mozambique Channel given all the concerns about the legs, and on balance I find that this was the intention. However the impact on the fatigue life consumption figure of positing a change of route is substantial but not dramatic. Using simple directional wave headings, Dr Aston's figure of 2.9 comes down 2.2 for the whole voyage. Importantly, the waters of the Cape would still have had to have been navigated. The Mozambique Channel route would have made no difference in that regard. So far as it is necessary to allocate fatigue life expenditure between the two parts of the voyage, I prefer the evidence of Mr Colman that most of the fatigue life would be consumed in the sector after Saldanha Bay (he said in a proportion of 1 to 2) to that of Dr Aston, who considered that the damage split is about even.
52. The fact is that at these levels, the differences in the various fatigue life expenditure calculations become fairly academic. Taking issue with the defendant's closing submissions, the claimants say that it "is not common ground that the legs did not have sufficient 'inherent fatigue life' to complete the tow on departure Galveston". Common ground or not, I am satisfied that this was the position. The consequence can be seen by looking at the claimants' own evidence as to what this conclusion implies. Dr Aston was asked:

Q. But if we compare whatever the right figures are, Viking had 2.13; I think you are now at 2.9 for the original voyage; Mr Colman is much higher than that?

A. Correct.

Q. But each of those would not comply with the DNV rule, would they?

A. No.

Q. Indeed, they're so far away from it that you would not, as a warranty surveyor, have approved this tow would you?

…

A. I would like to think I wouldn't have approved it.

MR JUSTICE BLAIR: Would you or wouldn't you?

A. I wouldn't.

MR PARSONS: The reason for that is it is so far away from the DNV rule that one could predict damage in this case, couldn't one.

A. You could predict damage but it doesn't necessarily mean that damage -- sorry, you could predict damage and you might expect damage, but it doesn't necessarily mean that catastrophic loss would necessarily occur.

Q. You would be expecting cracks of the order that catastrophic loss is likely to occur?

A. You would be expecting cracks and they could lead to catastrophic loss, but I don't think they would inevitably lead to catastrophic loss.

Q. We'll come back to inevitability later, but you certainly would agree you would be predicting damage of such a sort that --

A. You would be predicting damage.

Q. And you could not sign a certificate saying that this project could withstand the ordinary incidents of the expected voyage, could you?

A. If I knew those figures, no.

53. The last part of this exchange shows one difference between the parties' experts, namely as to the inevitability of loss. I shall place questions of inevitability on one side for the moment. The same line of questioning was put to the claimants' other expert, Dr Andrews. He was asked to hypothesise that he was at Galveston and had these figures to hand:

A. … are you saying I'm standing at Galveston with a spectral analysis giving a damage of 2?

Q. Yes

A. I'm afraid -- I'd have to say, don't do it.

Q. Don't do it?

A. Yes.

Q. Why not do it?

A. Because a damage sum of 1 on that design curve is intended, actually, for structures that can be inspected and are probably not redundant, and, as Dr Aston discussed in his report, looking at some numbers -- one or two of the other codes, given that this structure is not redundant and you're not going to inspect during the voyage, I would actually have wanted a lower damage sum even, than 1. So in this hypothetical situation, I would have to say professionally if I'd seen a damage sum of 2 at Galveston, I would have had to say, don't cast off, stop.

As I have said, the DNV recommendation is about 0.3 if there cannot be an inspection during the voyage, and 0.5 if there can. So it is not at all surprising that Dr Andrews would have said, "Don't cast off, stop".

The probabilities

54. The claimants accept that as the number of fatigue lives expended increases, the odds on survival decrease. They also submit (correctly in my view) that the predicted expenditure has to be measured against the rest of the evidence. As Ms Blanchard puts it, not all pinholes are equal, because the stress experienced on each individual pinhole varies with its orientation to the motions of the barge. Nevertheless, the probabilities are unambiguous in this case, and the defendant relies on them. The claimants' expert was asked about the probabilities. Dr Andrews accepted that at a fatigue life consumption of about 1, that is to say unity, there is a 1 in 40 chance of one pin-hole corner failing. Applied to the thirty foot level by reference to the two pinholes most at risk, this translates to a chance of survival of about 80%, or conversely a risk of failure of about 20%. But as the defendant submits, if one makes the assumption (which I am satisfied on the evidence is a conservative one) that 2 pinholes per level at 3 levels on each leg are at risk, then there are not 8 corners to take into account but 72 corners, and the chances of success become 16.16%, or conversely an 83% risk of failure.
55. But if one then assumes that the rig was to expend, not one, but between 2.5 and 2.8 fatigue lives during the tow, which were numbers put by the defendant in cross examination, Dr. Andrews' evidence was that, taking the eight corners of the two pinholes at the 30 foot level most at risk, "your chance of success has unfortunately dropped to about 1:250. So your chance of failure is high although not – 249 in 250". This as the defendant says is a chance of failure of 99.609%. However, since there were three legs and failure could have been within at least three levels of pinholes within the "danger zone" in each of the three legs, the defendant points out the chances of failure would be higher still.

Resetting the fatigue life—the repairs at Saldanha Bay

56. The position thus far reached is that the rig's legs failed as predicted by the various spectral fatigue analyses, including the contemporary one, and the one conducted by the claimants' expert. They failed about seven days into the second part of the voyage. This was the sector that included the rougher waters of the Cape in which the legs were most likely to be subject to severe stress. The weather experienced was within the range that could reasonably have been contemplated.
57. It was in the light of these facts that the claimants' case came to be amended. As put in its closing submissions, it is that the "common sense conclusion" is that the loss was caused by either (1) the inadequate repairs performed at Cape Town or (2) the perils of the sea, the first being the central plank of their case as the extract set out above from the amended pleading makes clear. In summary, their case is that the repairs at Saldanha Bay could have reset the fatigue life to zero, and that with the legs in an "as new" condition, or in a "machined" better than new condition, the tow could have safely have proceeded to Malaysia.
58. The defendant's case is that the failure to perform an "adequate repair" was not the proximate cause of the loss, or indeed causative at all, because no adequate repair was possible, no adequate repair was contemplated, intended, or attempted, and no repair that has been suggested would have made any difference.

What happened at Saldanha Bay

59. As a matter of fact, it is not in dispute that legs of the rig suffered extensive cracking on the first part of the voyage. The repair matrix reveals over 200 cracks in a range of sizes, and although most are under 50mm in length, there was one longer crack of 116mm, which is close to the length of the 120mm crack or cracks that Dr. Andrews believes led to final failure. He appeared to accept that about 1 fatigue life had been consumed by then. Whatever the number, I am satisfied that by this time, the legs were close to failure.
60. The stop at Saldanha Bay was of course a scheduled stop, the Noble Denton Certificate of Approval of 23 August 2005 having stipulated that, "Since the fatigue analysis shows possible damage, it is required that the legs be re-inspected at Capetown for crack initiation in way of the six levels of 'pin holes' above the mat. Capetown is, broadly speaking, the half way point and remedial work could be undertaken should it be found necessary". It is common ground that the reference to the fatigue analysis is to the simplistic analysis carried out by Viking.
61. On 10 October 2005, that is about seven weeks after leaving Galveston, the tug and barge arrived at Saldanha Bay. Mr Ooley who had earlier flown to Cape Town described the place as follows:

"Saldanha Bay, it's primarily -- I don't know what is the correct word for it, but it is a bulk carrier loading. There's coal loading facilities there. So it's sheltered. It has quite significant depth, so it can take quite large vessels. There's very little there in way of sort of public ship repair, if you like. So when we were there, we were actually doing it while we were moored in the centre of the bay. So there were very limited facilities in Saldanha."

62. What happened, or failed to happen, at Saldanha Bay is now the lynchpin of the claimants' case. There was a limited amount of evidence from Mr Ooley dealing with the events there and as he explained, the claimants left the technical side to the people they had retained to do particular jobs. Otherwise the story must be pieced together from emails and the like. As the claimants' closing submissions put it, the records of the repairs do not descend into enormous detail. So those caveats about the quality of the evidence need to be made at the outset.
63. A company called Atlantic Verification Cape (Pty) Ltd ("AVCAPE") was engaged to conduct inspections on the rig. Mr Ooley says that a preliminary inspection was conducted from the outside using rope access by the dye penetrant method. He believes that the preliminary inspection covered the first three levels of pinholes above the jack house, namely the 6 ft, 12 ft and 18 ft levels above the jackhouse. There is a potentially significant point here however, in that the defendant's case is that the likelihood is that no inspection at all was carried out at the 18 ft level at Saldanha. This is one of the "high risk" levels so far as the tow was concerned, and indeed was the level at which the port leg failed. Mr Ooley could not give first hand evidence, because he travelled out to the rig on a supply vessel with the inspection team, left them there, and came back.
64. As soon as it became apparent that repairs were required, Mr Neville Freeman of ABS (who was based in Durban) attended. The claimants say that he effectively deputised for Noble Denton (which was unable to get a surveyor on site in time) as far as physical inspections were concerned. The results of the inspection were sent to Noble Denton, which said that "the cracks found should be repaired in the same manner as was undertaken in Galveston prior to the rig leaving the USA" and recommended that the inspection be extended below the upper guide into the jack house.
65. There was a debate as to the proposed repairs between the various parties involved, Mr Wallace of Proceanic emailing Mr Cox of Noble Denton (neither of whom was physically present in South Africa) on 14 October 2005 saying:

"... Also, if the holes are due to fatigue, and fatigue comes from imperfections and surface defects at the intergranular level and high stresses (which are also at the extreme fibres), then grinding out the surface to eliminate all undetectable cracks and cracks which are less than the ground out length, should essentially reset the fatigue clock close to zero."

66. Mr Ooley's notebook is to the effect that repairs were complete on 16 October 2005. But Noble Denton subsequently required additional inspections to be carried out and his evidence is that this necessitated the scaffolding to be remobilised between about 18 and 23 October. On 18 October Mr Cox sent another email to Mr Wallace and Mr Ooley saying:

"Our engineers are increasingly worried that the repairs being undertaken are insufficient for the Cape passage. Bader [Diab of Noble Denton] is back in the office now and we would very much like to speak to you, or someone on site in Capetown, to make sure we are correctly understanding the extent of damages/repairs before committing ourselves to permit the transport to proceed."

67. Mr Ooley emailed back Mr Cox saying that the consensus was that the "drilled hole will arrest the crack for transit purposes until it can be dealt with in control conditions … Please note that these repairs have been completed so we require a resolution today/tomorrow latest …". Still on 18 October 2005, Mr Diab of Noble Denton refused to approve the repairs, emailing:

"... As you appreciate, no engineering has been presented to us that document the acceptability, in terms of fatigue performance, of the drilled hole repair method. As such we have no basis for any approval.

Based on the above we cannot approve the sailaway of the rig from the Cape. We would recommend the following:

- Proceanic need to propose an alternative repair method to drilled holes at the end of cracks. ...

- Proceanic need to inspect the holes at the 24' level for evidence of cracks and repair as necessary. Inspections at the next elevation (30') need to be carried out and cracks repaired if necessary."

68. Mr Ooley emailed back that flame cutting of the drilled corners would be carried out and that the 24 ft level had already been inspected. His notes record that inspections were to be performed at the 30 ft and 36 ft levels. On 19 October 2005, Proceanic sent Mr Diab sketches of proposed repairs on five locations which needed longer cracks to be removed, saying:

"... It is our understanding that the geometry of the finished repair will have a better profile from a fatigue perspective than the original design and that the reduction in the section of the leg at the repair locations is negligible. Additionally, the fatigue clock will be reset at zero in these locations."

69. Proceanic asked whether any further engineering was required by Noble Denton. Mr Diab requested that the stress concentration factor be derived for one of the five locations, saying that provided the SCF [stress concentration factor] was below 5.0, "we should be in good shape". Viking appears to have been engaged to perform a stress concentration calculation, and on 20 October 2005 Mr Cox of Noble Denton emailed saying:

"We have reviewed the modified pin hole stress and conclude that the new, reduced SCF's are acceptable.

It is now incumbent upon Proceanic through their various sub-contractors to ensure that the pin holes are modified as per the new design and to ensure that all cracks are fully removed."

70. On 23 October 2005, Mr Ooley sent Mr Diab the matrix showing the defects found on inspection, and saying that the repairs had been completed that day by way of enlarged radius on the pinhole corners or simple grinding repair in the cases of minor surface cracking. He asked Mr Diab to "forward the required approval to proceed to be issued as soon as possible. That day, Noble Denton duly issued a Certificate of Approval for the tow to proceed. The recommendations were revised to read:

"The ODIN LIBERTY legs have been the subject of a simplified fatigue analysis. Since the fatigue analysis showed possible damage, it has been necessary for the legs to be re-inspected at Capetown for crack initiation in way of the six levels of 'pin holes' above and below the jack house top. Remedial work has been undertaken as found necessary."

I have already pointed out that the reference to a "simplified fatigue analysis" suggests that Noble Denton had not seen Viking's spectral fatigue analysis, with its much higher number.

71. On 24 October 2005 AVCAPE issued a certificate of repair, which was signed by Mr Badenhorst of AVCAPE, Mr Freeman of ABS and Mr Ooley for the claimants. These were essentially the people who had been at the repair site while the various email exchanges had been going on. On 28 October 2005, the tow departed Saldanha Bay with two tugs in attendance, the "Smit Amandla" having joined the "Atlantic Hickory" for the passage through the waters round the Cape in accordance with Noble Denton's original stipulation.

Discussion

72. The claimants' closing submissions invite the court to draw the following conclusion from the expert evidence: "Better repairs were practically possible at Cape Town. The legs could have been returned to at least as good a condition as they were in at Galveston, or to DNV C Class condition, or to "machined" condition, with progressively improved fatigue performance". Their "primary case" is that the "proximate cause of the loss was (in shorthand) inadequate repairs and the fortuitous failure to ensure that the legs were put in a condition such that they would survive the passage."
73. By way of explanation, DNV C Class is a relatively high performance standard (as compared to welds), but is the lowest for non-welded details given in the DNV RP. The defendant has tagged this "good as new" (though the claimants point out that the starting condition of a DNV C Class detail could envisage cracks of up to 0.1mm). The term "machine" finish comes from the evidence of Dr Andrews, the purpose, according to the claimants, being to "answer the question whether, with an ideal repair, the fatigue life behaviour of the legs could be improved and to demonstrate the effect of surface finish on fatigue life. The essential conclusion is that an ideal repair would improve fatigue performance… The essential differences between the "machined" finish and a DNV C Class finish are the DNV C Class permits some rust pitting, surface roughness and cracking after some service life, whereas Dr Andrews' "machined" finish has that latitude removed. Small differences in surface finish can have a large effect on fatigue life". The defendant has tagged this "machined finish" as "better than new".
74. Before coming to the detail, the claimants' case is that "better repairs were practically possible at Cape Town", or to quote from the amendments to the Reply, that "the repairs at Saldanha Bay were "inadequately performed and/or inadequately engineered". This case requires some careful consideration. The argument that these repairs were the proximate cause of the loss came (as I have said) after the expert evidence was exchanged, which must have made it clear that the claimants' originally pleaded case was not tenable. The parties then discussed the form of the proposed amendments in draft, and the current version came after confirmation by the claimants that they did not allege negligence. It is not the claimants' case therefore that the repairers at Saldanha Bay negligently carried out the repairs in question. Better repairs may or may not have been practically possible, but the prior question must be in my view as to the repairs which the claimants intended or attempted to carry out. A theoretical resetting of the fatigue life of the legs is (in my view) irrelevant to the issues that I have to decide.
75. This question is a factual one, and not one to be decided on the expert evidence. The evidence in this regard was primarily Mr Ooley's oral evidence. He was asked in cross examination:

Q. There was no discussion and no intention to cut away or grind any of the pinholes that weren't showing cracks?

A. No, there was no discussion about that.

…

A. There was no discussion about looking at -- doing any type of remedial action to pinholes that didn't have any cracking in it.

MR PARSONS: The approach is the same as at Galveston. If you saw a crack you would repair it, but you weren't going to grind pinholes that weren't showing cracks. You weren't trying to reset a fatigue life everywhere?

A. No, that's correct.

76. He was asked in re-examination:

Q. What was the intention of the repairs at Cape Town?

A. The intention was to make the cargo safe for the balance of the voyage.

Q. And you were asked, was there any discussion of remedial action at pinholes where there was no cracking, and you confirmed that there was not. Who or what determined what repairs were carried out at Cape Town?

A. Noble Denton.

Q. And what would have been your attitude if it had been said to you that you should do something to pinholes where no cracking was found?

A. We would have been obliged to follow the requirements of the warranty survey.

77. Of course I accept that in general terms, the intention was to make the cargo safe for the balance of the voyage, and that if Noble Denton had required it, something would have been done even to pinholes where no cracking was found. There was plainly considerable concern about the state of the rig, and the factual account set out above shows that the inspection and repair process at Saldanha Bay was in effect gone through twice. But I consider that the defendant was correct to submit that the evidence shows that there was no intention or attempt to re-set the fatigue life at Saldanha Bay.
78. There is a point in this respect which I have already flagged. Mr Ooley believes that the preliminary inspection encompassed the 18 ft level at Saldanha, but could not give first hand evidence in that regard. The documentary evidence suggests otherwise. The repair matrix contains no reference to the 18 ft level. The ABS report of the inspection is to the effect that no inspection at that level was carried out, saying "No request made for NDT examination as previously carried out". Mr Ooley confirmed in his oral evidence that this was a reference to what was done in Galveston before the tow left. On the balance of probabilities, I find that no inspection was carried out at the 18 ft level at Saldanha. The experts' agreed memorandum noting the absence of records of inspection goes further, saying that there might have been cracks at this level upon departure. In any case, I further find that repairs were only carried out to those pin-hole corners where a crack had actually initiated and been detected.

Conclusions as to the re-set argument

79. Taking these points together, I find that there was no intention or attempt to effect a "fatigue life re-set" to DNV C class at Saldanha Bay, and a fortiori no intention or attempt to effect an even better "machined curve" fatigue life re-set of the kind posited by Dr Andrews. I agree with the defendant that this is an end of the matter on this part of the claimants' case. The fact that such repairs may have been theoretically possible is beside the point.
80. The defendant goes further, submitting that such repairs were not in any case practically possible. This has to do in part with the limitations on what was possible on a stopover half way through the tow and in the environment of Saldanha Bay. The claimants say that a fatigue life resetting repair involves removing enough steel to remove all cracks and then removing a safety margin to allow for the limits of the detection equipment, and that there was no practical obstacle to conducting such a repair given that flame cutting and grinding was being done in any event. I think they accepted that there would have been problems in repairing the inside surfaces of the pinholes, though it was suggested that someone might have been able physically to squeeze through into the interior of the legs. Dr Andrews' view is that whilst there would have been practical problems in repairing the corners, particularly the internal edges, these were not "insurmountable". But he accepted in cross examination that his "machined curve" was unrealistic. That scenario can be eliminated therefore, and (with respect) I cannot really see why it was ever raised. Dr Andrews however continued to maintain that the repairs could have been "better than C-class". Dr Aston's view was that the legs "could" have been repaired to a fatigue life re-set condition as represented by the DNV C-Class endurance curve and that would have offered a "reasonable" chance of the legs surviving the second stage. Significantly he added, "even though it would not have satisfied normal fatigue design criteria for marine operations". (In his supplemental report he substituted "would not" for "might not" as it had appeared in his original report.)
81. Dr Sykes noted that to remove all of the cracks from the corner of the pinholes, it would be necessary to first cut out the cracked material using the technique of flame cutting and then to carry out abrasive grinding to produce a smooth rounded surface and to remove any quench cracks. Whilst in theory such a procedure might be expected to generate crack-free corners, he thought that there would be considerable practical difficulties in ensuring that all of the cracked material was removed. It is not in dispute that it would have been a good idea to protect the pinholes from the potentially corrosive marine environment. So far as any cracking problem as regards paint is concerned, Dr. Andrews accepted that his suggested use of wax or grease was not a traditional repair method.
82. Finally, Mr Colman drew attention to the number of pinholes that would have to be reset, saying that there was no possibility of this scale of effort being applied. In that regard, however I accept the claimants' submission that repairs would only be necessary in the danger zone at the lower levels where the highest stress was experienced. The evidence was that the upper parts of the legs were not in practice going to cause problems. Dr Sykes said (and I accept) that you would keep going up from the 30-foot level until you found two rows clear of cracks. But even with that qualification, there would have been a significant number of levels involved in a fatigue life reset. I am satisfied on the evidence that although repairs of the type now adumbrated by the claimants might not have been impossible to achieve at Saldanha Bay, they would in practice have been difficult to achieve.
83. I have mentioned Dr Aston's view that the legs repaired to a fatigue life re-set condition as represented by the DNV C-Class endurance curve would have offered a "reasonable" chance of the legs surviving the second stage. But this proposition bears further examination. On the basis that each sector of the voyage would consume about the same fatigue life, Dr Aston's evidence was that a reset would give a fatigue life of about 1.5 based on the original voyage, or 1.1 on the basis of the routing by the Mozambique Channel. Each of these it is to be noted is above unity. As I have said already, I prefer the evidence of Mr Colman that much more of the fatigue life would be consumed in the sector of the voyage after Saldanha Bay. His rough and ready proportion of about 1 to 2 brings even Dr Aston's figures up considerably. This part of the case has to be seen against the more exacting conditions that the tow would experience as it rounded the Cape. Dr Andrews in effect accepted that low cycle fatigue would have been encountered in these waters, and I am satisfied that it would have been. The defendant says that the motion of the barge would have been such as to impart stresses in excess of twice the yield strength. I have already noted that failure due to low cycle fatigue occurs over tens of thousands of cycles, rather than millions of cycles as in the case of high cycle fatigue. Under these conditions, it is clear that the fatigue life would have been eaten up very quickly. I am satisfied that in general terms Mr Colman was correct to say that, "I don't think that these legs were ever going to make it round the Cape". I consider that to be the reality of the situation in this case, whatever repairs were or could have been carried out in Saldanha Bay.

The law

84. As I said at the beginning of this judgment, this is a case about causation. In this part of the law, the question is framed in terms of "proximate cause": see s. 55 Marine Insurance Act 1906. The word "proximate" reflects the familiar observation that many matters may contribute factually to a particular outcome. The choice of proximate cause is the process by which the legally relevant cause is selected. The parties agree that the question for the court is as to the proximate cause of the loss of the legs of this oil rig. They are also in agreement that the test to be applied is essentially one based on the common sense of the ordinary businessman or seafarer. As Bingham LJ put it in T.M Noten B.V. v. Harding [1990] Lloyd's Rep 283 at 286-287, "Unchallenged and unchallengeable authority shows that this is a question to be answered applying the common sense of a business or seafaring man". In applying the test, I shall consider first the defendant's contention that the loss of the legs was inevitable, and go on to consider the two rival contentions on the basis that the loss was accidental, namely that the cause of the loss was inadequate repairs at Saldanha Bay ( to adopt the same convenient shorthand adopted by the claimants), or that it was inherent vice.

Inevitability

85. Underwriters can rely upon inevitability of loss to turn aside a claim because the whole concept of insurance is about risks, not certainties (Soya GmbH Mainz Kommandgesellschaft v. White [1982] 1 Lloyd's Rep 136 at.149, per Donaldson LJ). The leading case is British and Foreign Marine Insurance Co Ltd v. Gaunt [1921] 2 AC 41 which concerned a cargo of wool shipped from Chile to England and insured, as in the present case, under an "all risks" policy. As Lord Birkenhead LC put it at p.46, "These words cannot, of course, be held to cover all damage however caused, for such damage as is inevitable from ordinary wear and tear and inevitable depreciation is not within the policies".
86. I accept Ms Blanchard's submission that the question of inevitability has to be judged as at the time when the policy incepted. I also agree with her that to show that the loss in question was fortuitous (in the sense that it was not inevitable) represents a low hurdle for the insured. As Lord Sumner explained in Gaunt at p.58:

"When he [the insured] avers loss by some risk coming within "all risks," as used in this policy, he need only give evidence reasonably showing that the loss was due to a casualty, not to a certainty or to inherent vice or to wear and tear. That is easily done. I do not think he has to go further and pick out one of the multitude of risks covered, so as to show exactly how his loss was caused. If he did so, he would not bring it any the more within the policy. These considerations answer the appellants' complaint, that the plaintiffs were meagre with their proof. So they were, but it was enough for them to prove some casualty insured against. Rowlatt J., as I venture to think, attached too much importance to the absence of any kind of marine disaster. If the casualty was fortuitous, it needed not to be a calamity."

87. For the defendant insurer, Mr Parsons QC rightly accepted that "inevitability" in this context means that the loss in question is certain to happen. The consequence is that probability, however high, does not bring a case within the ambit of inevitability. The evidence in the present case provides an illustration of that point. I accept Mr Colman's view that the rig was "very lucky" to have survived the first sector of the voyage, as indeed the speed with which it failed during the second sector showed. Nevertheless, a developed crack would not, on its own, have been sufficient to cause one of the legs to come off. That required in addition a "leg breaking" or "final straw" stress that finally fractured the weakened steel. As Mr Colman put it, "you've got to catch it just right, if you want to make it actually fail all the way round". I have quoted above his summary of the various factors in play in that regard and need not repeat it. The defendant's case on inevitability was largely based on the probability of one or more of the many pinhole corners failing calculated by reference to the fatigue damage ratios given in the various fatigue analyses. But these analyses, as indeed the defendant emphasised, are predictions dependent on a number of imponderables, including the assumed sea state. The actual sea state encountered during the voyage may be different. There is no question in this case where the probabilities lie, but it does not follow that the case is one of inevitability. My conclusion is that the failure of the legs as this rig was towed around the Cape was very probable, but it was not inevitable.
88. That is sufficient to decide this issue in favour of the claimants. I need not decide further whether the claimants are correct to submit that a policy will respond to an inevitable loss unless, in addition to being inevitable, it was also known to the claimants to be inevitable when the policy was concluded or the risk incepted. This proposition has the support of Donaldson LJ in Soya GmbH v. White, ibid, as p. 145, who says that he "would … prefer to use the term 'known certainty' instead of 'inevitability'". Ms Blanchard submits that this approach makes commercial sense because the insured takes out cover against what it sees as the potential risks facing the venture, and the alternative is that the question of inevitability is judged with the benefit of hindsight.
89. The defendant submits that loss without fortuity is not within the scope of a voyage policy, regardless of whether or not the assured knew that the loss was certain. This proposition is supported by Arnould's Law of Marine Insurance, 17^th edn, which states at paragraph 22-24 that "inevitability of loss should probably be regarded as a ground of defence (on the principle that insurance is about risks not certainties) even where the fact that the loss was certain to occur was wholly unknown to the parties. It must, however, ultimately be a question of construction, whether the policy wording extends to cover inevitable damage, which was not a known certainty. It will only be in rare cases that the point is likely to be open to debate, since it is clear that 'all risks' in a cargo policy does not include inevitable losses …". I would accept the law as stated in Arnould. It is consistent with Soya GmbH v. White at page 140 where Waller LJ makes no distinction between known and unknown certainty of loss. But on the basis of my finding that the loss was not in fact inevitable, the question does not arise for decision in the present case.

Inherent vice

90. The policy incorporated the ICC(A) terms by which the insurance covered all risks of loss of or damage to the subject matter insured, subject to the exclusion that, "In no case shall this insurance cover … loss damage or expense caused by inherent vice or nature of the subject matter covered". It is not in dispute that damage may be caused by inherent vice without it being inevitable (e.g. Noten B.V. v Harding, ibid, at 289), so it is not an answer to a plea based on inherent vice that the loss was fortuitous in the sense of not being factually inevitable. The burden is however on the defendant insurer to make out the exclusion.
91. The disagreement between the parties centred round the meaning of the term "inherent vice" as it applies in this case. The claimants submit that it refers to the natural behaviour of the insured cargo without (and they emphasise this) external intervention. Their argument has focused on the decision in Mayban General Insurance Bhd v. Alstom Power Plants Ltd [2004] 2 Lloyd's Rep 609. Insofar as that case decides that if the sea is no more severe than could reasonably have been expected, the conclusion must be that the real cause of the loss is inherent vice, it is submitted that the case was wrongly decided. The claimants refer in support of their submission to an article by Professor Howard Bennett, Fortuity in the Law of Marine Insurance, LMCLQ [2007] 315 (particularly at p. 348). The significance of the point so far as the facts of the present case are concerned, is that the claimants accept that the weather experienced during the voyage was within the range that could reasonably have been contemplated. So they are understandably concerned to persuade the court that it in no way follows from this that the inherent vice defence succeeds.
92. I begin with Soya GmbH v. White (cited above in connection with the inevitability issue). In that case, a cargo of soya beans was insured against the risk of heat, and one of the issues was whether the underwriters could avoid liability on the basis that the proximate cause of the deterioration of the cargo was inherent vice. As Arnould puts it at para 22-25, this concept was "defined in somewhat different terms" at various stages of the appeal process. In the Court of Appeal [1982] 1 Lloyd's Rep 136 at p.149-150, Donaldson LJ (citing Lord Sumner in Gaunt, ibid, at p.57, and Scrutton LJ in R.D. Sassoon & Co Ltd v. Yorkshire Insurance Co (1923) 16 Ll. L Rep 129 at p.132) said as follows:

"… in my judgment a loss is proximately caused by inherent vice if the natural behaviour of the goods is such that they suffer a loss in the circumstances in which they are expected to be carried. This is the test under a contract of affreightment and the shipowner in this case could have pleaded inherent vice in answer to a claim for damage to cargo."

93. On appeal to the House of Lords, and on the assumption that the proximate cause of the loss was inherent vice, it was held that the policy nevertheless covered the loss. The provisions of the policy took the case outside s. 55(2)(c) Marine Insurance Act 1906 by which the insurer is not liable for inherent vice "unless the policy otherwise provides". The leading judgment was given by Lord Diplock (with whom the other judges agreed), who said at [1983] 2 Lloyd's Rep 122 at p.126 that the phrase "inherent vice or nature of the subject matter insured" where it is used in s. 55(2)(c) of the 1906 Act—

"… refers to a peril by which a loss is proximately caused; it is not descriptive of the loss itself. It means the risk of deterioration of the goods shipped as a result of their natural behaviour in the ordinary course of the contemplated voyage without the intervention of any fortuitous external accident or casualty. Prima facie, this risk is excluded from a policy of marine insurance unless the policy otherwise provides, either expressly or by necessary implication …"

94. It is not (as I understand it) contended by either party that these formulations are different in substance, nor are they: both passages were cited by Phillips J at first instance in Noten B.V. v. Harding [1989] 2 Lloyd's Rep 527 at pp.529-530. Based on Lord Diplock's judgment however, Ms Blanchard submits that inherent vice means the natural behaviour of the insured cargo without (as she emphasises) external intervention. The insured, she points out, whether under a hull or a cargo policy, need not prove that the weather was exceptional to claim for a loss by perils of the sea. This latter proposition is supported by authorities that show that a loss may be said to be caused by perils of the sea even where the seastate is that which could reasonably have been expected: see The Xantho (1887) LR 12 App. Cas. 503, Canada Rice Mills Ltd v. Union Marine & General Insurance Co [1941] AC 55, and The Miss Jay Jay [1985] 1 Lloyd's Rep 264, and [1987] 1 Lloyd's Rep 32. At first instance in the latter case (p. 271) for example, Mustill J said, "The cases make it quite plain that if the action of the wind or sea is the immediate cause of the loss, a claim lies under the policy notwithstanding that the conditions were within the range which could reasonably have been anticipated".
95. The defendant of course accepts this proposition as correct. But the claimants' further submission, namely that the decision in the Mayban case undermines the proposition, is not accepted by the defendants as correct. Given the course of the argument, I must consider the case with some care. In Mayban, an electrical transformer was shipped from Ellesmere Port to Rotterdam where it was transferred to a container vessel for onward carriage to Lumut (by coincidence the destination of the oil rig in the present case). The vessels carrying the transformer encountered severe winter weather shortly after leaving port, but not such as a commercial person would regard as falling outside the range of what could reasonably be expected in the seas in question at that time of year. When it arrived at site, it was found to be seriously damaged and a claim was made under the policy of insurance. However, the insurers considered that the damage had resulted from the transformer's inability to withstand the ordinary incidents of carriage by sea from the United Kingdom to Malaysia during the winter months, and rejected the claim on the grounds that the loss was caused by inherent vice.
96. The policy was an "all risks" one, as in the present case. At [7], Moore-Bick J says as follows.

"The expression "all risks" is used in policies of insurance as a convenient way of encompassing all insurable risks to which the property in question may be exposed without attempting to identify them individually. The contract nonetheless remains one under which the insurer accepts the risk of loss occurring through the occurrence of some peril acting on the property insured. A number of consequences follow from this. The first is that in order to recover under the policy the insured must prove that the loss was caused by an accident or casualty of some kind. Insurers accept the risk, but not the certainty, of loss. The second is that although the insured must prove a loss by an accident of some kind, it is not necessary for him to go further and establish the exact nature of the accident by which it occurred. The third is that the policy does not cover the insured against loss due to wear and tear or the inherent vice of the thing insured, whether that loss was bound to occur or was fortuitous in the sense that its occurrence depended on the particular circumstances to which the goods happened to be exposed in the course of the voyage.

He then goes on to cite the passage from Lord Diplock's judgment in Soya GmbH v. White which I have set out above. It was not in dispute that inadequate packing, where packing is required to enable the goods to withstand the ordinary incidents of the voyage, can properly be regarded as an aspect of inherent vice (see [19], and Bennett, The Law of Marine Insurance, 2^nd edn, paragraph 15.55).

97. Having analysed the facts, Moore-Bick J said at [21] that the immediate cause of the damage to the transformer was the violent movement of the vessel due to the actions of the wind and sea: "These in themselves were certainly events of a fortuitous nature and they were external to the cargo, but were they the real cause of the loss?" There then follows the passage in the judgment to which the claimants take exception:

"The action of the winds and waves is, of course, an inevitable incident of any voyage and is therefore a hazard to which all goods carried by sea are necessarily exposed. Goods tendered for shipment must therefore be capable of withstanding the forces that they can ordinarily be expected to encounter in the course of the voyage and these may vary greatly depending on the route and the time of year. In a case such as the present, therefore, the competing causes, namely, perils of the sea and inherent vice, are to a large extent opposite sides of the same coin. If the conditions encountered by the vessel were more severe than could reasonably have been expected, it is likely that the loss will have been caused by perils of the sea (though even then there might be evidence that the goods would have suffered the same degree of damage under normal conditions). If, however, the conditions encountered by the vessel were no more severe than could reasonably have been expected, the conclusion must be that the real cause of the loss was the inherent inability of the goods to withstand the ordinary incidents of the voyage."

98. The judge went on at [26] to find that the relatively short periods of high wind encountered on the passage were neither extreme nor even unusual in the sense that they are encountered often enough for mariners to regard them as a normal hazard, concluding that a cargo that could not withstand exposure to conditions of that kind could not be regarded as fit for the voyage. In the result, he was satisfied the loss was caused by the inability of the transformer to withstand the ordinary conditions of the voyage rather than by the occurrence of conditions which it could not reasonably have been expected to encounter.
99. In her well argued submissions, Ms Blanchard submits that it is wrong to say that if the conditions encountered by the vessel are no more severe than could reasonably have been expected, the conclusion "must" be that the real cause of the loss is the inherent inability of the goods to withstand the ordinary incidents of the voyage. All her arguments (as I have understood them) really go to that point. The decision it is submitted is in "flat contradiction" to NE Neter & Co Ltd v. Licenses and General Insurance Co Ltd [1944] 4 All ER 341, Tucker J, where it was said (at p.343G) that "it is clearly erroneous to say that, because the weather was such as might reasonably be anticipated, there can be no peril of the sea". The Mayban case, it is said, is relied upon by the defendant as effectively creating a rule of evidence that, absent exceptional weather being shown to have occurred, the loss must be attributed to inherent vice. The effect is to reverse the burden of proof, by taking the burden of proof off the insurer to prove inherent vice and requiring the insured to prove extraordinary weather. It is said that the consequence is to significantly reduce the scope of cover afforded for perils of the sea in a cargo policy. Finally, it is said that the Mayban case wrongly equated the concept of fitness to withstand the ordinary incidence of the voyage as it is typically applied in a contract of carriage by sea with the concept of inherent vice in an insurance contract.
100. In my opinion, this is a mistaken analysis of the Mayban case. Taking the last point first, the view that the exception against "inherent vice" is the same in the context both of carriage by sea and marine insurance is supported by the judgment of Donaldson LJ in Soya GmbH v. White at p.149-150 that I have quoted above. I agree with Mr Parsons QC for the defendant that one would expect the test to be the same. As Arnould says in paragraph 22-25, although Lord Diplock did not refer to these observations, there is no reason to suppose that he disagreed with Donaldson LJ's approach, or that he intended to give the concept of inherent vice a narrower meaning than had been indicated in the Court of Appeal. At paragraph 22-26 Arnould says that after Soya v. White "inability to withstand the ordinary incidents of the voyage is clearly an appropriate test of inherent vice". With the addition of the word "inherent" before "inability", I accept this as an accurate statement of the law. It was the approach that Moore-Bick J adopted in the Mayban case, and I respectfully consider that he was correct to do so. Contrary to the claimants' submission, there is no contradiction with the NE Neter case, in which Tucker J made it clear that he was contemplating a situation in which the goods were properly stowed and in good condition when loaded (ibid, at p.343H, and see the headnote at p.341).
101. I do agree with Ms Blanchard that were it to be submitted that Mayban effectively created a rule of evidence that, absent exceptional weather being shown to have occurred, the loss must be attributed to inherent vice, such a submission would be contrary to authority, and wrong. But that is not the defendant's submission, nor is it what Mayban decided. The defendant says rightly in my view that the enquiry in Mayban was as to the proximate cause of the damage to the transformer. The wind and the weather were identified as a cause, as was the inherent nature of the transformer being shipped. On the facts, Moore-Bick J decided that inherent vice was the proximate cause. He did not suggest that, in order to qualify as a peril of the sea, the weather has to be extraordinary. On the contrary, he expressly contemplated that ordinary weather conditions could qualify as a peril of the sea. The real question in the case, as in the present case, was as to the proximate cause of the loss.
102. There are two other points to deal with before coming to the application of the law to the facts of the case. Although it is correct that inherent vice is concerned with something inherent to the subject cargo itself, I agree with the defendant that the claimants do not put the test quite correctly in their submission that inherent vice is the natural behaviour of the insured cargo "without external intervention". This paraphrase does not accurately reproduce the careful language that Lord Diplock used in Soya GmbH v. White (who said, "… without the intervention of any fortuitous external accident or casualty"). As the defendant submits, whereas the "vice" must be internal, the damage, being the consequence of that vice, can and often will develop with the assistance of an external circumstance, typically the weather. That indeed is what happened in this case, as the evidence set out at paragraphs 48 and 49 above demonstrates.
103. The defendant further submits that in judging the "ordinary incidents of the voyage", the Court considers the voyage as intended, and therefore does not take into account purely theoretical possibilities. In general terms, I accept this proposition. That is one reason for leaving out of account repairs that could theoretically have been undertaken mid tow, but which were not in fact intended or attempted. But it does not follow that the court has to leave out of account the alternative Mozambique Channel route as the defendant suggests. As explained above, although the tow would still have had to navigate the rough waters around the Cape, this would have been more sheltered than the direct route which had been planned across the Indian Ocean. On the stress calculations, such a change would have had brought the estimated fatigue life expenditure down somewhat. I have found that it would have made sense to proceed by the Mozambique Channel given all the concerns about the legs, and I am satisfied that there was a change of plan at some point. I accept the defendant's submission that it is not an answer to contend that the voyage could in theory have been differently routed, and that the rig could in theory have survived such a different voyage. But I do not accept that this means that the court must exclude consideration of the causal effect of a predictable and reasonable change of plan of this nature. However, for the reasons I have set out above, on the facts the intended change of route was not such as to make a practical difference so far as the fate of the legs was concerned.

Conclusion

104. My conclusion so far as the inevitability defence is concerned is that although the failure of the legs as the rig was towed around the Cape of Good Hope was very probable, it was not inevitable. I decide this issue therefore in favour of the claimants.
105. That leaves for decision the question as to the proximate cause of the loss of the legs of the rig. As Moore-Bick J said in the Mayban case at [20], issues of causation should normally be determined by applying common sense rather than too refined an analysis. With that in mind, the choices are the claimants' case that the proximate cause was inadequate repairs carried out mid-tow at Saldanha Bay, and the defendant's case that the proximate cause was the inability of the legs to withstand the ordinary incidents of the voyage, and thus the loss was caused by inherent defect.
106. In stating my conclusion, I highlight some of the features of the evidence discussed above, starting with the history of the legs of this type of rig failing in transit, including on this very rig some nine years previously. What happens as the barge pitches and rolls in the sea with the legs sticking into the air over three hundred feet high is that cracks initiate at the corners of the pinholes in the legs. The cracks propagate until they reach a critical size when, with the application of sufficient stress in the form of waves, the leg fails altogether. That is what happened in this case, and may explain why the predominant jack-up design is now a lattice-type structure.
107. Then one comes to the particular circumstances of this tow. Prior to departure, Swire advised that the legs should be shortened for transit. That was because they thought that the stress limits within the legs would be exceeded in the rough weather conditions around the Cape of Good Hope. Although considerable attention was paid to the arrangements for the transit, that recommendation was not followed. In due course, on the basis of a stress analysis which estimated a fatigue life expenditure for the voyage at an acceptable 0.26, the tow was certified by the marine surveyors. There is no dispute that this figure was too low. After the tow sailed, a more sophisticated spectral fatigue analysis was produced giving a number of 2.13.
108. By the time the rig arrived half way through the voyage at Saldanha Bay just north of Cape Town, there was extensive cracking at the pinhole corners. Considerable work was done by way of repairs, but although it would have been theoretically possible to have repaired the legs so as to reset the fatigue life, I am satisfied that in practice that would have been difficult to achieve. In any case, there was no intention or attempt on the claimants' part to do so, and this possibility was ultimately a theoretical one. At all events, the marine surveyors duly certified the tow to continue on its way—but they had apparently not been provided with the spectral fatigue analysis with its greatly elevated number.
109. In fact it is common ground that the figure of 2.13 is itself too low. The claimants' expert's estimate for the consumption of fatigue life was 2.90 (though admittedly a maximum figure). I have found for reasons set out above that even this is probably too low. He himself as a warranty surveyor would have not approved this tow. The claimants' other expert, when asked what his reaction would have been if shown a spectral analysis giving a damage ratio of 2.00 said, "I would have had to say, don't cast off, stop".
110. It is not in issue that the weather experienced was within the range that could reasonably have been contemplated (albeit the claimants' expert puts it at the upper end). Nevertheless, the legs broke off some seven days after the tow ventured out into the waters of the Cape.
111. The weather is plainly an important part of the evidence in the case, but it is only part. In arriving at a conclusion I have kept in mind the claimants' submissions on the law, and have tried to take account of the evidence as a whole. Nevertheless, I cannot accept that the proximate cause of the loss of the legs was inadequate repairs at Saldanha Bay. There was no intention or attempt on the claimants' part to reset the fatigue life of the legs. To state that, is not to criticise the claimants' conduct in any way. But the tow was midway through the voyage, and it would in practice have been difficult to achieve repairs to that standard. I have given my reasons above and shall not repeat them. As a matter of common sense, the legs failed not because of the repairs, but despite them. The real problem lay with the inherent inability of the legs to withstand the normal incidents of the voyage. In that regard, I felt that the defendant's expert Mr Colman came closest to expressing the view of a "business or seafaring man" (Noten B.V. v. Harding, ibid, per Bingham LJ). As he put it—"I don't think that these legs were ever going to make it round the Cape". That in my opinion is the reality of this case. Taking the evidence as a whole, I am satisfied that the defendant insurers have proved that the proximate cause of the loss was the fact that the legs were not capable of withstanding the normal incidents of the insured voyage from Galveston to Lumut, including the weather reasonably to be expected.
112. It follows that in my view, the proximate cause of the loss was inherent vice, and the claim must be dismissed. I am grateful to both parties for their assistance.