[Home] [Databases] [World Law] [Multidatabase Search] [Help] [Feedback]
	England and Wales High Court (Commercial Court) Decisions
You are here: BAILII >> Databases >> England and Wales High Court (Commercial Court) Decisions >> BHP Billiton Petroleum Ltd & Ors v Dalmine SpA [2002] EWHC 970 (Comm) (16 May 2002) URL: http://www.bailii.org/ew/cases/EWHC/Comm/2002/970.html Cite as: [2002] EWHC 970 (Comm)

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

IN THE HIGH COURT OF JUSTICE
QUEENS BENCH DIVISION
COMMERCIAL COURT
[2002] EWHC 970 (Comm)

		Royal Courts of Justice Strand, London, WC2A 2LL
		16 May 2002

Mr Justice Cresswell:

INTRODUCTION

1. The first claimant (“the Operator”) is and was at all material times the Operator of the Liverpool Bay Development.
2. Each of the second to seventh claimants was at some point in time between the date of the first Joint Operating Agreement (“JOA”) and the Joint Development Agreement (“JDA”) a Co-Venturer in the Liverpool Bay Development.
3. The Development comprises facilities built to exploit four oil and/or gas fields in the Liverpool Bay area of the Irish Sea.
4. The fields are known as the Hamilton, Hamilton North, Douglas and Lennox fields. They are situated about 15-20 miles off the coast of North Wales, in relatively shallow tidal waters (about 25-28 metres in depth), to the west of the main shipping channel into Liverpool docks.
5. The Hamilton, Hamilton North and Lennox fields are served by unmanned satellite production facilities which are connected by sub-sea pipelines to a (manned) central production facility at Douglas.
6. Sales gas is transported by sub-sea pipeline from Douglas to an on-shore terminal sited at Point of Ayr, Clwyd, North Wales.
7. Produced oil is transported by sub-sea pipeline from Douglas to an oil storage installation located on block 110/8a of the Irish Sea.
8. Gas produced from the Lennox field (together with some Douglas off-gas) is injected back into the Lennox field’s gas cap via pipeline. This is to maintain/maximise the reservoir pressure to produce the Lennox oil, and also to provide a means of postponing the extraction of such gas.
9. The Development is mainly comprised within blocks 110/13 and 110/15 of the Irish Sea but also covers other offshore blocks and on-shore acreage.
10. The Development was built between 1993 and 1996. Gas was produced from the Development in January 1996, and oil was produced from April 1996.
11. The defendant (“Dalmine”) is an Italian steel-making company which produced 12” diameter steel pipes used for the construction of a sub-marine gas reinjection pipeline (“the pipeline”) laid between the Douglas and Lennox platforms.
12. These proceedings arise out of the failure of this pipeline.
13. British Steel plc supplied the Operator with the pipes incorporated in the pipeline, but the pipes had been manufactured by Dalmine pursuant to a sub-contract between British Steel and Dalmine.
14. The pipes should have been manufactured so as to comply with a Data Sheet and Specification which constituted the contractual specification for the pipes, both in the contract between the Operator and British Steel and in the sub-contract between British Steel and Dalmine.
15. Dalmine manufactured the pipes between October and December 1993.
16. Each batch of steel from which pipes were made is referred to as a “heat”. Dalmine tested each heat of steel for compliance with the Data Sheet and the Specification before producing pipes from that heat.
17. Dalmine also carried out product analyses on 2 pipes manufactured from each heat.
18. Dalmine produced

a) Inspection Reports for each heat which included statements of the result of analysis both of the heat prior to the production of pipes and of two pipes manufactured from that heat; and

b) A Certificate of Compliance signed by Mr Antonini (a senior member of Dalmine’s Quality Control Department) which certified that the pipes had been manufactured and tested in accordance with the Data Sheet and Specification.

19. The pipeline was laid between 30 April and 27 June 1994 but did not enter service until April 1996. The first discovery of a leak occurred in June 1996 when bubbles were observed on the surface of the Irish Sea.
20. When the whole of the pipeline had been checked, leaks were discovered at six failure sites (referred to as Leak Sites A – F).
21. When pipes at the Leak Sites were tested the claimants discovered that some of them failed to meet the requirements of the Data Sheet and Specification in that, in particular, the carbon equivalent value (“the CEV”) of the steel from which the pipes had been made was in excess of the contractual maximum.
22. One or more of the pipes at each of Leak Sites A – F failed to comply with the Data Sheet and Specification contrary to the relevant Inspection Reports and Certificate of Compliance.

The Proceedings

23. Proceedings were issued on 16 April 1998 in which claims were made against British Steel in contract for failing to produce pipes which complied with the Data Sheet and Specification and in tort for negligence. In the same proceedings, claims were made against Dalmine in tort, the claimants relying on allegations of negligence in the production and/or certification of the pipes.
24. The claims against British Steel were abandoned following the decision of the Court of Appeal on the construction of limitation and exclusion clauses in the contract between the Operator and British Steel: see BHP Petroleum Ltd. v British Steel PLC and Dalmine SpA (CA), [2000] 2 Lloyd's Rep 277.
25. Internal records first produced by Dalmine in the course of discovery on 24 May 1999 revealed discrepancies between the results of the testing they recorded, and the results which had been inserted in Dalmine’s Inspection Reports. The claimants amended their claim against Dalmine so as to introduce a claim in deceit alleging that Dalmine had fraudulently misrepresented (by the production of false Inspection Reports and a false Certificate of Compliance) that some of the pipes produced by Dalmine complied with the specified maximum for CEV and the requirements relating to elemental chemical composition, when in fact they did not.
26. According to the statements of case, two causes of action are pursued by the claimants: a claim in negligence and a claim in deceit.
27. This hearing is concerned with issues as to causation in relation to the case in deceit.
28. The defendant has admitted that it fraudulently misrepresented the CEV of some of the heats produced by it (see further below).
29. Issues of reliance and causation of alleged losses were to be tried as preliminary issues in February 2002. However, on 5 February 2002, the defendant admitted the claimants’ plea (as it stood prior to the last round of amendments) of reliance on the defendant’s fraudulent misrepresentations.
30. On 17 January 2002 the parties entered into a Memorandum of Agreement by which they each agreed to forgo certain arguments that might otherwise have been available to them. The existing statements of case have not been further amended to reflect that Memorandum of Agreement and they therefore need to be read together with that agreement.
31. On 17 January 2002 there was to be a PTR in this action. In the event, all matters which were to be addressed at the PTR were agreed prior thereto. In the circumstances Langley J did not require the attendance of the parties, and made an Order by consent.
32. It is, by the Memorandum of Agreement, agreed that if the remaining issue of causation is determined in the claimants’ favour, the claimants will be entitled to judgment on liability with damages to be assessed and they will not need to pursue their action in negligence.
33. The issue to be tried in this hearing is as follows. Did the incorporation of non-compliant pipe cause the pipeline to fail (as the claimants say) or would it have failed anyway (as the defendant says)?

Testing of the Pipes

34. Various tests were carried out on sample pipes prior to the construction of the pipeline and numerous tests have been carried out on samples of unused pipe and pipe recovered from the pipeline post-failure.
35. The experts instructed by the parties have agreed a detailed “Memorandum of Technical Facts” which is intended to record, among other things, summaries of various technical documents and reports. It does not, however, reflect any agreement between the parties as to the underlying technical assumptions, results and/or conclusions drawn in those technical documents and reports.

The Cracks

36. At each of Leak Sites A to F at least one of the pipes was manufactured from heats 935517, 935333, 935567 or 935571 and was found in post-failure testing to have a CEV which exceeded the maximum of 0.40% permitted by the Data Sheet and the Specification.

Dalmine’s Fraudulent Misrepresentations

37. Dalmine has admitted that

(1) by the Inspection Reports and the Certificate of Compliance, it represented that the pipes had, to the best of their knowledge, the chemical composition and CEVs therein set out and accordingly that the pipes’ chemical composition and CEVs complied with the Specification and Data Sheet; and

(2) that those representations were false in that some of the pipes were not within specification for individual chemical elements and that in some instances the CEV did not comply with the Specification and Data Sheet, contrary to what was set out in the Inspection Reports and the Certificate of Compliance.

38. Dalmine has further admitted that

(1) it knew that the representations were false or it was reckless as to whether they were false;

(2) it knew that Dalmine’s deceitful misrepresentations would be repeated to the first claimant; and

(3) it intended its deceitful misrepresentations to be relied upon.

39. So far as specific misrepresentations about particular heats (and pipes from such heats) are concerned:

(1) the results of heat and/or product analyses for heats 935517, 935567 and 935571 were altered so as to misrepresent that the chemical composition and/or CEV of the pipes manufactured from those heats complied with the Specification and Data Sheet. At each of Leak Sites A –E, one or both pipes either side of the weld came from one of these three heats.

(2) the parties agreed on 17 January 2002 that the trial of the causation issue should proceed on the basis that, if no false representations had been made, only compliant pipes would have been incorporated into the pipeline. This agreement is recorded in the Memorandum of Agreement.

Reliance

40. Reliance is no longer in issue. Dalmine conceded the “February reliance case” on 5 February 2002.
41. Prior to 5 February 2002 Dalmine had admitted that

(1) The Operator was at all material times acting as agent of the Co-Venturers who were party to the JOA or the JDA.

(2) Brown & Root (“B&R”), as representatives of the Operator, attended and inspected the carrying out of works at Dalmine’s steelworks, from time to time.

(3) Mr. Gerry Slater of B&R was responsible for ensuring that the information recorded in the Inspection Reports was consistent with the Specification and Data Sheet.

(4) Dalmine showed Mr. Slater Inspection Reports on heat analyses from the end of October 1993 onwards.

(5) Dalmine’s Certificate of Compliance was included in the Data Book submitted by Dalmine to British Steel by 10 February 1994 and passed on by British Steel to the Operator on 23 February 1994.

(6) Dalmine knew and intended that this Certificate of Compliance would be relied upon by B&R and by the Operator in verifying that the pipes were manufactured according to the Specification and the Data Sheet.

(7) Those of the claimants which were Co-Venturers at the time Dalmine’s deceitful misrepresentations were relied upon by the first claimant, would be the parties entitled to recover damages for deceit (if they suffered any loss as a result, which is denied by Dalmine).

42. Since 5 February 2002 the following elements of the claimants’ case are also now admitted by Dalmine:

(1) Dalmine knew and intended that Mr. Slater while acting as agent for the Operator and/or for the Co-Venturers for the time being, should rely on the Inspection Reports shown to him.

(2) The Uncoated Pipes were only released to UPCL from Dalmine after Mr Slater had checked the Inspection Reports in draft and had checked that they confirmed compliance with the Specification and Data Sheet.

(3) Dalmine knew and intended that the Certificate of Compliance would be relied upon by B&R and by the Operator as agent of the Co-Venturers (the second, third, sixth and seventh claimants) in verifying that the pipes were manufactured according to the Specification and Data Sheet.

(4) The Co-Venturers for the time being, by their agents the Operator and/or B&R, relied upon the Inspection Reports and the Certificate of Compliance:

a) in the case of the Inspection Reports when allowing Uncoated Pipe to be released from Dalmine’s factory and by using the data recorded in them to carry out weld procedure qualification tests between 7 December 1993 and 4 January 1994; and

b) in the case of the Inspection Reports and the Certificate of Compliance by allowing and requiring McDermott to weld the Incorporated Pipes together and incorporate them into the pipeline between 30 April 1994 and 27 June 1994.

43. As from 5 February 2002 the following elements of the claimants’ case are also now admitted by Dalmine:

(1) In reliance upon Dalmine’s deceitful misrepresentations, the Operator and/or its agents acting on behalf of the Co-Venturers for the time being, and in particular for the second, third, sixth and seventh claimants, accepted the pipes and utilised them by incorporating them into the pipeline believing that those pipes complied with the Specification and Data Sheet when in fact they did not and/or believing that the Inspection Reports could be relied on generally when in fact they could not.

(2) Had the Operator and/or its agents (including B&R) known that any of the pipes did not comply with the Specification or the Data Sheet, they would have been rejected and they would not have been incorporated into the pipeline.

44. The claimants had also alleged, additionally or alternatively, that had the Operator and/or its agents known that the Inspection Reports could not be relied on generally, the Operator would either have rejected all the pipes manufactured by Dalmine or would have insisted on comprehensive and independent re-testing of those pipes before any of them were incorporated into the pipeline. However, the claimants agreed, by the Memorandum of Agreement dated 17 January 2002, not to pursue any such argument (although the Particulars of Claim have not been amended to reflect this agreement).

Causation

45. The pipeline had, at the time that service was suspended, failed only where one or more of the incorporated pipes either side of the weld had a CEV in excess of the specified maximum of 0.40%.
46. Dalmine contends that the welding procedure caused excessive hardness at the weld roots between pipes. The claimants do not accept that the welding procedure was the sole cause of the excessive hardness in any of the weld roots (in particular at Leak Sites A to F) although they admit that it was one of the causes of excessive hardness in some of the welds (in particular those at Leak Sites A to F). The claimants contend that the high CEV of pipe material either side of any weld contributed to any excessive hardness in the weld root.
47. The claimants say that the reliance by the second, third, sixth and seventh claimants (through their agents, the Operator and/or B&R) on Dalmine’s deceitful representations caused non-compliant pipes to be incorporated into the pipeline in that

(1) had the Operator or its agents been shown the Internal Records relating to the heats for which results had been obtained, which did not comply with the Specification and/or the Data Sheet (including in particular 935517, 935567 or 935571), pipes from those heats would have been rejected and would not have been incorporated; and/or

(2) had the heats been retested, all pipe from heats failing the retest would have been rejected, and these would have included heat 935333 in addition to those identified in (1) above. The claimants rely on the results of post-failure testing.

48. The parties agreed on 17 January 2002 that the February trial of the causation issue should proceed on the basis that, if no false representations had been made, only compliant pipes would have been incorporated into the pipeline.
49. After the agreement contained in the Memorandum of Agreement had been made, a second supplementary report of Dr. Baker was served on 14 February 2002. A dispute as to the construction of the Memorandum of Agreement has arisen, it being contended by the claimants that Dr. Baker’s probability analysis in support of Dalmine’s case on causation takes into account “non-compliant” pipes (namely, pipes from heats from which one or more pipes that have been tested have been measured as out of specification as to CEV), and that Dalmine should not be entitled to rely upon it in so far as it does.
50. The claimants contend that the incorporation of non-compliant pipes caused the pipeline to fail relying on the facts that

(1) at each of Leak Sites A to F, one or more of the pipes either side of the weld came from one of heats 935333, 935517, 935567 or 935571 and did not comply with the Specification and/or the Data Sheet in that it had a CEV in excess of 0.40%; and

(2) the higher the CEV of any particular pipe, the less resistant it is to the propagation of cracks initiated by sulphide stress corrosion cracking (“SSCC”), and the greater the probability that failure will occur.

51. Dalmine denies the claimants’ contention that the incorporation of the non-compliant pipes caused the pipeline to fail.

Loss

52. Dalmine admits that the claimants have replaced the pipeline and carried out certain associated exercises and further admits that the cracking in the pipeline prompted the claimants to take these steps. However, Dalmine denies that it is liable for the costs of taking those steps and avers that the pipeline would have had to be replaced regardless of anything done or not done by it.
53. Dalmine does not admit the amount of any loss or damage suffered by any of the claimants.
54. If the claimants are successful on the preliminary issue of causation, quantum is to be tried separately.

STATEMENT OF AGREED ISSUES AND AREAS OF DISPUTE

55. It is common ground between the parties that:

(1) The pipes were intended to be and were subjected to sour service conditions in that they were used to carry fluids containing both hydrogen sulphide and water.

(2) Steel with excessive hardness which is subjected to sour service conditions is prone to the initiation of cracks through SSCC.

(3) The conditions in the Failure Zone of the pipeline were such as to promote SSCC. In particular, there was: (i) intermittent flow rate; (ii) injection of corrosion inhibitor was not continuous as had been intended; (iii) there was no batch inhibition or sphering, which further reduced the effectiveness of the corrosion inhibitor; (iv) there was water drop out in the relevant section of pipeline without which there would have been no corrosion and no SSCC; (v) there was a higher level of hydrogen sulphide than had been anticipated (and than was permitted by the pipeline Works Authorisation). The environment was what is described as “Domain 3”.

(4) The Failure Zone of the pipeline contained 144 pipes between suspected leak site 18 (pipe number 62792 / weld DJ01089 at the Lennox end) and actual leak site F (pipe number 61583 / weld ML02127 at the Douglas end).

(5) Dalmine does not rely on any pipes outside the Failure Zone in support of its case.

(6) Limits were placed on the hardness both of the steel and the weld metal in the Data Sheet and the Specification.

(7) The cracks which led to failure of the pipeline each initiated in welds between pipes, and not in the pipes themselves.

(8) Cracks initiated at points in the weld root where the hardness of the weld root material (an alloy of steel from the pipes either side of the weld and weld filler metal) exceeded the design limit.

(9) Excessive hardness in the weld root was not caused by excessive hardness in the parent metal.

(10) SSCC was the cause of the initiation of cracks in weld roots. Cracks initiated in the root of some (but not all) welds in the pipeline through SSCC.

(11) No cracks initiated in the pipes themselves.

(12) Excessively hard spots in the weld root made particular welds susceptible to SSCC. It is not known at what hardness welds in the pipeline became susceptible to SSCC. BS4515 prescribed a hardness limit in the weld root of 250HV10.

(13) (Per the claimants) a cause (per the defendant) at least a cause of the excessively hard spots in the weld roots was the weld procedure employed.

(14) Cracks initiated in welds between compliant pipes as well as between welds on one or both sides of which there was non-compliant pipe.

(15) Certain cracks which had initiated in welds, on one or both sides of which there was/were non-compliant pipe(s), propagated through the thickness of one or both of the pipes on either side of the weld to cause the pipeline to leak. Thus the pipeline failed.

(16) Those pipes which came from heats in respect of which Dalmine has admitted that it fraudulently mis-stated the CEV, did not comply with the specification because their CEVs exceeded the maximum permitted percentage of 0.40%.

(17) A steel’s resistance to crack propagation generally decreases as its CEV increases .

(18) Pipe with a CEV over 0.40% (i.e. non-compliant pipe) would generally be expected to be less resistant to crack propagation than compliant pipe, although the extent to which that is so is in issue.

(19) At each of the six Leak Sites A to F, one or both pipes either side of the weld were non-compliant pipes.

(20) If Dalmine had not been fraudulent, none of the pipes from the fraudulent heats and only compliant pipes would have been incorporated into the pipeline.

(21) Even had the pipes all been compliant as to CEV, cracks would have initiated in some of the weld roots.

A SUMMARY OF THE ISSUES

56. The central issue is did the incorporation of non-compliant pipe cause the pipeline to fail, or would it have failed anyway? Thus it is necessary to consider whether or not any crack which might have initiated in a weld between compliant pipes, is likely to have been or become large enough to overcome the resistance to crack propagation of pipe which complied with the specification as to CEV - with the result that the pipeline would have failed anyway, even if Dalmine had not been fraudulent and had supplied only compliant pipe?
57. There are issues of law between the parties as to the burden of proof in this case.
58. The parties have identified the following issues which arise:

(1) (Had it not been for Dalmine’s fraud) how resistant to crack propagation would compliant pipes have been within the Failure Zone?

(2) What range (as regards depth) of cracks was present, or was likely to appear, in welds between compliant pipes within the Failure Zone and what was their likely propensity to propagate?

(3) Can the Court assess the probability of the combination of factors required to produce the situation in which a crack would develop in a weld between compliant pipes within the Failure Zone, which would be sufficiently large to overcome the resistance offered by (compliant) pipe either side of the weld, to the propagation of that crack through the thickness of the pipe material? If so, how should it go about doing so?

59. Dalmine contends that the pipeline would have had to have been replaced at a later stage in any event, even if it had been built entirely of compliant pipe. The claimants dispute this.

I. The Resistance of Compliant Pipes within the Failure Zone to Crack

Propagation

60. Issue 1: How resistant to crack propagation would compliant pipes have been within the Failure Zone?
61. Sub-Issue 1.1: What was the upper limit of the CEV of “compliant pipes” according to the Specification and Data Sheet?
62. Sub-Issue 1.2: What was the actual CEV of each of the pipes within the Failure Zone?
63. Sub-Issue 1.3: Which heats fall to be excluded from consideration in the light of the parties’ agreement of 17 January 2002?
64. Sub-Issue 1.4: What, if any, figure should the Court take as being the CEV of pipes which Dalmine would have supplied (instead of those pipes which it would not have supplied had it not been fraudulent) or which fall to be excluded as a result of the agreement of 17 January 2002?
65. Sub-Issue 1.5: On the assumption that only compliant pipes had been incorporated into the pipeline, what was the resistance to propagation of compliant pipe within the Failure Zone?

II. Cracks in Welds in the Failure Zone.

66. Issue 2: What range (as regards depth) of cracks was present, or was likely to appear, in welds between compliant pipes within the Failure Zone and what was their likely propensity to propagate?
67. Sub-Issue 2.1: What size cracks had initiated and/or propagated pre-failure and what was their distribution between the welds?
68. Sub-Issue 2.2: What scope was there for the further initiation and/or propagation of cracks had the pipeline not been de-pressurised, given further time and/or changes in environmental conditions?
69. Sub-Issue 2.3: What was the stress intensity factor likely to have been at the tip of the crack of the size which Dalmine contends would be sufficiently large to cause failure of compliant pipes?

III. Approach to Probability

70. Issue 3: Can the Court assess the probability of the combination of factors required to produce the situation in which a crack would develop in a weld between compliant pipes within the Failure Zone, which would be sufficiently large to overcome the resistance offered by (compliant) pipe either side of the weld, to the propagation of that crack through the thickness of the pipe material? If so, how should it go about doing so?

THE EXPERT EVIDENCE

Expert Witnesses called/instructed by the Claimants

Mr. John Marlow

71. Mr. Marlow is a Chartered Engineer with 36 years industrial experience, the last 22 of which have been specifically related to metallurgy, corrosion, welding engineering, construction, non-destructive examination and quality assurance/control for both on and off shore oil and gas development projects.
72. Mr. Marlow provided two reports. The first was dated 29 June 2001. In his first report Mr. Marlow discussed welding procedure, the relationship between CEV and hardness, and set out his conclusions. The supplemental report was dated 3 December 2001. In his supplemental report Mr. Marlow discussed weld metal root hardness, parent pipe hardness and the welding procedure, and again set out conclusions.
73. Mr. Marlow struck me as a witness who was careful to give a balanced opinion to the Court.
74. In the executive summary to his first report Mr. Marlow stated that the pipeline failed due to SSCC. The welding procedure used to weld the pipeline tended to produce, in some cases, hard zones in the root of some welds joining both “in specification” and “out of specification” pipe. These hard zones were susceptible to SSCC. Mr. Marlow explained that where one or both of the pipes joined was out of specification with respect to CEV (but not otherwise), the sulphide stress corrosion cracks propagated out of the weld roots into the heat affected zone and from there into the material of the pipe itself due to the “dramatically reduced” fracture toughness of the out of specification line pipe in sour service conditions. This cracking continued through the thickness of the pipeline in six locations resulting in a leak at each location.

Professor Frederick Burdekin

75. Professor Burdekin took up his appointment as a Professor of Civil and Structural Engineering at the University of Manchester Institute of Science and Technology (“UMIST”) in 1977 following a period of 7 years in industrial research, and 8 years with professional Consulting Engineers. His specific area of technical expertise is in the field of steel structures and engineering components, particularly in respect of the application of fracture mechanics to fracture, fatigue and stress corrosion failures and implications for reliability and non-destructive testing. In July 1987 Professor Burdekin was elected to the Fellowship of Engineering (now The Royal Academy of Engineering). From 1990 to 1993 he served as a member of The Engineering Council and in March 1993 was elected a Fellow of The Royal Society.
76. Professor Burdekin provided three reports. The first was dated 28 June 2001. In his first report Professor Burdekin discussed fracture mechanics and stress corrosion cracking, test results, severity of applied conditions at cracks, static fracture toughness tests by TWI, and set out his conclusions. In a supplemental report dated 14 December 2001, Professor Burdekin commented on and responded to reports prepared by Dr. Baker (dated 2 November 2001 and 7 December 2001) and Dr. Prosser. He also summarised the major differences between the views of Dr. Baker and himself. In his second supplemental report dated 18 February 2002, Professor Burdekin clarified the values used for stresses and explained the reasons for certain changes in the calculations leading to revised results for stress intensity factors for cracks of different sizes. In his third supplemental report dated 21 February 2002 Professor Burdekin commented on Dr. Baker’s second supplemental report dated 15 February 2002, in which Dr. Baker attempted to carry out a probabilistic analysis to estimate the likelihood of pipes, which complied with the CEV specification requirement, failing by SSCC. Professor Burdekin amended his (first) report and first supplemental report principally (but not exclusively) to reflect the matters set out in his second supplemental report.
77. In my view Professor Burdekin was an extremely impressive witness who was careful to give a balanced opinion to the Court.

Dr. Frank Golightly

78. Dr. Golightly is the Technical Director of CAPCIS Ltd which is an internationally recognised independent consultancy organisation specialising in corrosion engineering, materials testing, failure investigation and contractual research and development. CAPCIS was established by UMIST in 1973. Dr. Golightly’s specialist fields include metallurgy and sour service.
79. Dr. Golightly provided two reports. The first was dated 28 June 2001. In his first report Dr. Golightly provided an introduction to sour service and cracking mechanisms, discussed pipeline design, pipe Specification and Data Sheet, CAPCIS full ring testing, pipeline failure, crack initiation - location, cause of cracking of the weld root, crack propagation into the parent pipes, chemical composition of the pipes, and set out his conclusions. In a supplemental report dated 3 December 2001 Dr. Golightly considered those areas of the reports prepared by Dr. Baker (2 November 2001) and Dr. Prosser (2 November 2001) which fell within his own expertise, and the results of further work carried out by The Test House on behalf of the defendant.
80. Regrettably Dr. Golightly was taken ill in the course of the hearing and had to go into hospital. By agreement his reports were admitted under the Civil Evidence Act. For this reason I have had to be careful as to the amount of weight I give to the contents of his reports, because his evidence was not tested in cross-examination.

Expert Witnesses called by the Defendant

Dr. Kevin Prosser

81. In 1971 Dr. Prosser was appointed as a Research Fellow in Metal Joining in the Department of Industrial Metallurgy at the University of Birmingham after obtaining a PhD in the same department. In 1976 he joined British Gas Engineering Research Station as a Senior Engineer in the Materials Department. He was promoted to Principal Engineer in 1981. In 1989 he was given responsibility for two laboratories at the Engineering Research Station Metallography and Welding. In 1996 he left British Gas and helped form the engineering consultancy company MACAW Engineering Ltd, as Director.
82. Dr. Prosser’s expertise lies in the field of material selection and welding matters.
83. Dr. Prosser said that in considering his evidence the Court should not take into account any comment he made in the context of fracture mechanics, but should have regard to Professor Burdekin and Dr. Baker. In so far as any question arises as to whether cracks were in fact likely to propagate, that was not a matter in respect of which he wished his evidence to be taken into account. Thus all aspects of his report where he expressed views as to the likelihood of cracks propagating from welds between compliant pipes should be disregarded.
84. Dr. Prosser provided one report dated 2 November 2001. He withdrew part of paragraph 3.6.2(a).
85. When cross-examined Dr. Prosser provided valuable evidence as to how an honest and reasonable manufacturer would proceed in the case of a maximum for CEV of 0.40%. Such a manufacturer would set his target at a mean in the order of 0.37%. A range of ( or (.02% might be expected either side of 0.37% (i.e. 0.35% to 0.39%). The distribution of carbon equivalent values would normally be such that one would expect a smaller number of pipes at the extremes, the lower end and the higher end. Different techniques are employed by different laboratories to measure CEV. Different laboratories would measure the same pipe at different places. On the second day of his evidence Dr. Prosser said that he had appreciated overnight that the implications of his evidence (as above) were damaging to Dalmine’s case, but he confirmed that the honest and reasonable manufacturer would aim for 0.37%. He added that variation of CEV within a heat would be larger than variation within a single pipe. Dr. Prosser said that pipes with higher CEV can lead to higher hardness in the HAZ and higher hardness in the weld root (particularly the former).
86. As to yield strength Dr. Prosser said that he would tend to take the average of 438, 542, 587 and 612 i.e. 544.7 N/mm².
87. In my view Dr. Prosser gave frank answers when cross-examined on the matters set out above (albeit that these were damaging to the defendant’s case and in particular to Dr. Baker’s probabilistic analysis). I accept his evidence to the extent indicated in this judgment.

Dr. Timothy Baker

88. Dr. Baker is a professional metallurgist. He was awarded PhD in 1971 (research on fracture toughness of structural steels). Between 1970 and 1975 he was Research Investigator and subsequently Project Manager, Fulmer Research Institute, Slough. Between 1975 and 1986 he was a Lecturer and between 1986 and 1995 Senior Lecturer at the Department of Materials, Imperial College of Science, Technology & Medicine, University of London. Between 1995 and 2001 he was part-time Senior Lecturer. He has acted as an independent consulting engineer since 1995. Dr. Baker has given expert evidence in three public inquiries and approximately 50 court cases and arbitrations.
89. Dr. Baker’s first report was dated 2 November 2001. In his first report Dr. Baker discussed mechanism of failure, and set out conclusions. He stated at paragraph 3.16 “I have little doubt that over a period of weeks, continued crack development would have occurred leading to many more leakages developing, both from welds associated with pipes which exceeded the CE limit of 0.40% and from welds joining pipes which complied with the carbon equivalent requirement.”
90. In his first supplemental report dated 7 December 2001 Dr. Baker discussed K_1SCC testing, K_1SCC test results, significance of K_1SCC data and set out conclusions.
91. Dr. Baker’s second supplemental report was served on 15 February 2002. This report contained what purported to be a probabilistic analysis. At paragraph 13 Dr. Baker wrote “If as suggested... above, approximately 20% of these pipes had contained a weld root crack having a depth of at least 4.5mm, there would have been two pipes which would be expected to have developed through wall cracks”. The service of this report on 15 February was contrary to the overriding objective, contrary to the Court’s directions as to the time for service of experts’ reports and contrary to the proper conduct of commercial litigation. It is most regrettable that Dr. Baker was instructed to prepare the report as late as (I was informed) 12 February.
92. Dr. Baker’s third supplemental report dated 20 February was by way of response to Professor Burdekin’s second supplemental report dated 18 February.
93. Dr. Baker’s fourth supplemental report was dated 21 February but the service of this report did not comply with the Court’s directions and was withdrawn. Dr. Baker’s revised second supplemental report was dated 22 February.
94. Dr. Baker is a highly intelligent man. I regret that in my judgment on this occasion Dr. Baker’s evidence and approach did not accord in a number of respects with the guidelines set out in the Ikarian Reefer [1993] 2 Lloyd’s Rep. 68 at 81 and [1995] 1 Lloyd’s Rep. 455 at 496, CA. At times Dr. Baker was inclined to assume the role of an advocate. Numerous points and arguments were raised in his reports served on and after 15 February which should have been included in his earlier reports. Dr. Baker did not always make it clear when a particular question or issue fell outside his expertise. Dr. Baker’s evidence was inconsistent in a number of material respects. Dr. Baker has no professional qualifications as a statistician and is not an acknowledged expert in research into probabilistic fracture mechanics or risk assessment. His probabilistic analysis in his second supplemental report/revised second supplemental report was flawed for the reasons set out below. When the claimants’ solicitors raised proper and necessary questions in correspondence (following service of the second supplemental report) certain answers provided by Dr. Baker (as he acknowledged in evidence) were inadequate and incomplete, and did not address the question. Dr. Baker accepted that he should have indicated at the start of his oral evidence that he had changed his view in one respect.
95. In fairness to Dr. Baker I should point out that he was put in a difficult position by being instructed to prepare a probabilistic analysis in the week immediately before the trial. Had he thought such analysis appropriate he would have undertaken it earlier and would have sought advice from a statistician.
96. For the reasons set out above and for the further reasons set out below, Dr. Baker’s evidence in this case should in my judgment be approached with considerable caution.

THE CLAIMANTS’ CASE

97. Mr. Howard QC for the claimants submitted as follows.
98. There is no scope for any dispute that the incorporation of non-compliant pipe caused the pipeline to fail at the six leak sites.
99. The issue is whether Dalmine has satisfied the Court that the pipeline was doomed to fail in any event at one or more welds between compliant pipes.
100. In considering this question, it is necessary to be clear as to the context in which it arises.

(i) The pipeline leaked at the 6 leak sites.

(ii) Those leakages were caused by cracks in the welds of the order of 2-3 mm (per Dr. Baker) and 3 mm (per Professor Burdekin).

(iii) Such cracks would not have affected compliant pipe (per Dr. Baker and Professor Burdekin).

(iv) The pipeline failed because of the leakages at the six leak sites.

(v) However, Dalmine contend that notwithstanding the above, the pipeline would have failed at another unidentified pipe or pipes in the Failure Zone, even if the pipeline entirely comprised compliant pipe.

101. The case has proceeded on the basis that there are two relevant stages in SSCC:

(i) Stage 1-initiation in the weld root;

(ii) Stage 2-propagation out of the weld root.

102. Surprisingly, Dr. Baker, in the course of cross examination, sought to depart from the agreed basis on which the case was to be conducted and to introduce some variant whereby stage 1 extended beyond the weld root. Mr. Hancock QC intimated on the same occasion that Dalmine would seek to argue such a point. This thesis is inconsistent with the basis on which the experts have consistently approached the case. It is not open to Dalmine now to suggest some new mechanism not previously raised.
103. Because Dalmine cannot point to any particular pipe or pipes that would have failed, they sought to prove a case of “doomed to fail” in two ways, both through Dr. Baker. There are important differences between the two approaches. One involves a question of risk assessment and the acceptance of risk. It is not addressing the question of probability of failure and so is irrelevant, and certainly does not support a “doomed to fail” argument. The other involves probabilistic calculations and is hopelessly flawed.

The BS7910 Approach

104. Dr Baker’s first approach was based on BS 7910. However, BS7910 is concerned with design and determining acceptable risk. It is not concerned with assessing probability of failure. If the consequences of failure are unacceptable, then a very low risk of failure will be tolerated. But it does not follow that because something might have an unacceptable risk of failure, that it would probably fail. BS 7910 is addressing a wholly different question.
105. In the course of cross-examination, Dr. Baker accepted that his approach by reference to the acceptability of the consequences of failure was wholly irrelevant to the issue before the court, which is not concerned with risk assessment but with the question whether a compliant pipeline would have failed.

The Probabilistic Approach

106. Dr Baker’s probabilistic analysis was introduced without prior warning on the eve of the trial. He told the Court that had he considered such an approach to be appropriate, he would have attempted it earlier and would have sought advice from a statistician. He was asked to attempt such an analysis on the Tuesday of the week before the trial.
107. A failure of a pipe in the Failure Zone (assuming it was entirely comprised of compliant pipe) depends upon it being established that in at least one joint a crack would have initiated in the weld root which possessed a sufficient K_applied that it could overcome the resistance, the K_1SCC, of the (compliant) parent pipe into which it sought to propagate.
108. The first and obvious flaw in Dr. Baker’s approach is that it does not predict what in fact occurred when applied to the conditions that the pipeline experienced. There were no failures at 95 bar, and yet he maintained in cross-examination that he would predict the failure of 4 or 5 joints.
109. Dr. Baker sought to explain away the absence of failures in compliant pipe by saying that any such pipe had not been exposed for a sufficient period of time. Assuming that the Failure Zone had experienced conditions conducive to SSCC for a period close to 30 days, Dr Baker expressed the opinion that within a few more weeks (3 or 4), failures in welds between compliant pipes would have occurred. The fact is that the period of exposure to the corrosive conditions was far longer than Dr. Baker was allowing. It was in fact of the order of 75 days (Dr. Prosser) or 83 days (Dr. Golightly). Unlike Dr. Baker, both these witnesses were giving evidence about something within their expertise. On Dr. Baker’s own analysis, there had been sufficient time for larger cracks to initiate and propagate. That is also the view of Dr. Golightly.
110. The fact that Dr. Baker’s analysis does not fit the facts indicates that it must be wrong. That may be as to size and incidence of cracks, K_applied, numbers of susceptible pipes, K_1SCC of pipes or any or all of these.
111. In any event, Dr. Baker’s analysis is dependent upon the following:

(i) determining what type of cracks are of concern and why;

(ii) determining what type of pipes would be susceptible to such cracks;

(iii) determining what is the likelihood of cracks of the type in question;

(iv) determining what is the population of pipes that would be susceptible to failure.

112. The difficulty with the analysis is that it has proceeded on an erroneous basis in many respects and it is fundamentally riddled with uncertainty, none of which was recognised by Dr. Baker or reflected in the analysis. The truth is that such an analysis cannot be done on the “simplistic” basis put forward by Dr. Baker and a proper analysis is not within his expertise. As he eventually appeared to acknowledge, such an analysis requires the expertise of a statistician.
113. Professor Burdekin, who has much experience in the field of risk assessment, considered that Dr. Baker’s probabilistic analysis was completely unreliable and could not be used to provide any indication of the expected performance in the pipeline. Indeed, Dr. Baker accepted in the course of his oral evidence, although not previously, that he ought to have considered the various elements described by Professor Burdekin which rendered his analysis uncertain.
114. There are considerable flaws in Dr. Baker’s “simplistic” approach. In particular, the uncertainties are so great at a number of stages, particularly as to crack incidence and carbon equivalent values and hence K_1SCC values, that, even on its own terms, no reliance can be placed on the analysis. Furthermore, the analysis leads nowhere if (a) the correct K_applied values are used (b) effect is given to the parties’ agreement and (c) the matter is approached on the premise that Dalmine had sought to perform their contract. Professor Burdekin was correct in his analysis that it was unlikely that a pipeline comprised of compliant pipe would have failed. Dr. Baker accepted that his view that the pipeline would have failed was dependent on his probability analysis. If that analysis is shown to be of little value due to its flaws (as it is), he accepted that the basis of his opinion would fall away.

THE DEFENDANT’S CASE

115. Mr. Hancock QC for the defendant submitted as follows.
116. Dalmine admit that they fraudulently misrepresented that the pipes in heats 935517, 935567 and 935571 complied with the CE requirement of 0.40% when in fact they did not.
117. Dalmine admit that BHP relied on those fraudulent misrepresentations, and that they incorporated those pipes into the pipeline, when they would not otherwise have done so.
118. The issue for the Court is whether that reliance caused the losses pleaded, which all arise out of the fact that the pipeline had to be replaced. The question is whether the pipeline would have had to be replaced had BHP not incorporated the pipes from those fraudulent heats.
119. The burden of proving that those losses were caused by the breach complained of is on BHP.

(1) BHP suggest that the only burden on them is to prove that the actual leaks would not have occurred, had it not been for the breach. Even were this the only burden on BHP, they do not discharge it.

(2) However, BHP’s suggestion is wrong in law. They must show that the loss they claim would not have occurred had it not been for the breach of duty. That loss is the loss caused by the failure of the pipeline. When and where the pipeline failed is irrelevant; they must show that it would not have failed at all (in its 30 year life).

Technical issues

The two stage process of cracking

120. The cracking process involved two stages, stage 1 and stage 2. Stage 1 is slow cracking, which is not K controlled. It has been referred to as “initiation”. Stage 2 is faster cracking, which is K controlled. It has been referred to as “propagation”.

Stage 1 cracking: “Initiation”.

121. Cracks began in the weld roots, in welds between compliant as well as non-compliant pipes.
122. The CEV of the parent pipes had minimal influence on the hardness of the weld root and HAZ. The cooling rates achieved when the pipes were welded on the laybarge were the important influence on this factor (and on weld strength). Thus, had the parent pipes been compliant, the cooling rates in fact achieved would have led to hard zones in the weld root and HAZ, and consequent cracks.
123. The conditions for stage 1 cracking other than hardness (i.e. continued corrosion, ineffective inhibition, hydrogen sulphide and CO₂ content of the gas, pipeline pressure and weld stress) would remain constant or worsen, so that the size of the cracks which would have developed via stage 1 cracking was determined by the size of the hard zones in the weld root and HAZ. The maximum depth of crack would be about 5mm (the size of the weld root); the width of the crack would be determined by growth through the weld root and into the HAZ. The cracks would develop towards the semicircular shape, through the HAZ.

Stage 2 cracking: propagation.

124. Stage 2 cracking (“Propagation”) depends on the comparison between the K_applied of the crack and the K_1SCC of the adjacent material. When the K_applied exceeds the K_1SCC, the crack will propagate and fast cracking will ensue.
125. The K_applied of the crack depends on:- residual stress (itself dependent on weld strength); pipeline pressure; crack shape; crack depth; and crack location. Only surface emergent cracks have been considered as relevant in this case.
126. Here:-

(1) The appropriate pipeline pressure is 128 bar;

(2) The appropriate weld strength is 587, and the appropriate residual stress is at least 402;

(3) The appropriate crack shape is between semi-circular (a=c) and circular (2a=c).

127. The appropriate K_applied figure for a 4.5mm crack is between 1455 (at the very lowest) and 1780 (at the highest).
128. The K_1SCC of the pipes is affected by the CEV of the pipes. The trend is as set out in the graphs annexed to Professor Burdekin’s supplementary report. Dr Baker’s graph line is to be preferred. The value for 0.40 is about 1500Nmm-3/2.
129. The likely CEVs of the pipes in the Failure Zone is as set out in Dr Baker’s revised supplementary reports. The effect of the January agreement is then that such of the pipes as have been found on the facts to have a CEV of more than 0.4049% must be deemed not to form part of the pipeline. This is the only effect of the agreement.
130. The question is thus whether the cracks which would have initiated in the weld root and HAZ would have reached a K_applied figure sufficient to overcome the K_1SCC associated with the CEV of compliant pipes.

Approaches to determining the case on the balance of probabilities

131. Dalmine rely on a number of matters, alternatively and cumulatively.
132. First, Dalmine say that it is clear that cracks had in fact initiated in compliant pipe with a sufficient K_applied to overcome the K_1SCC of a pipe with a CEV of 0.4049%, which it is accepted was compliant pipe. Had all the pipeline been made of compliant pipe, the pipeline would have failed.
133. Secondly, Dalmine say that, had an ECA been performed in accordance with BS 7910, this pipeline would have been found unfit for its purpose, i.e. likely to fail.
134. Thirdly, Dalmine rely on:-

(1) The view of Dr Prosser that the specification was non-conservative;

(2) The fact that the replacement pipeline had a different specification in material respects, along with the fact that a different welding procedure was employed for the replacement pipeline.

135. Fourthly, Dalmine rely on the probability analysis of Dr Baker, in which he concludes that there would be four leaks. Although it is accepted that there are factors of uncertainty, the conclusion is sufficiently robust to mean that these factors of uncertainty do not operate to alter it.
136. Fifthly, Dalmine rely on the fact that Professor Burdekin has put forward no sensible support for a conclusion that the pipeline would probably not have failed.
137. For all of these reasons, Dalmine say that the pipeline would probably have failed even had there been no fraud.
138. Finally, Dalmine submit that (if this is a relevant inquiry, which Dalmine say it is not) there is no evidence to support the conclusion that, had the out of specification pipes at the leak sites been replaced with in specification pipes, the pipeline would not have failed.

Conclusions

139. In conclusion, Dalmine submitted that:-

(1) The Court should conclude that the pipeline would have failed in any event, on the balance of probabilities; and/or

(2) The Court should conclude:-

(a) That there is no evidence in support of the proposition that the pipeline would not have failed at the leak sites, if the out of specification pipes had been replaced with in specification pipes, so that the claim fails.

(b) That the claimants cannot establish that, on the balance of probabilities, the pipeline would not have failed, so that the claim fails.

ANALYSIS AND CONCLUSIONS

Sulphide stress corrosion cracking

140. The cracking mechanism involved in this case was SSCC.
141. Precautions can be taken to minimise the risk of SSCC occurring, and these include controlling the chemical composition of the steel so as to limit the CEV of the pipe material and its hardness.
142. Dr Golightly explained the nature of sour service and the susceptibility of carbon and low alloy steels to SSCC in section 3 of his first report. In short, a product of corrosion of steel in the presence of hydrogen sulphide (H₂S) is hydrogen atoms (H). While hydrogen atoms can combine to produce hydrogen molecules (H₂), the combination of hydrogen atoms is retarded, and before gas molecules are formed which would have been carried away down the pipeline, hydrogen atoms have sufficient time to work their way inside the atomic lattice of the steel (a process known as diffusion).
143. Dr Golightly also produced a helpful illustration of the development of an SSCC crack. It is important to note that the cracks at the failure sites all propagated in the longitudinal direction of the pipe, as represented on Dr Golightly’s illustration.
144. The experts agree that there are two relevant stages in SSCC:

(1) Stage 1 – Crack Initiation; and

(2) Stage 2 – Crack Propagation.

145. It is common ground that regions of high hardness in excess of the BS 4515 limit of 250HV10 made the weld root of some welds susceptible to crack initiation (Stage 1).
146. Once an initiated crack has formed at the weld root, it will only spread if it is so large that the material surrounding it is not sufficiently resistant to prevent it from propagating (Stage 2).
147. It is common ground that resistance to crack propagation decreases with increasing CEV.
148. At the leak sites, a crack initiated in the weld root and then propagated into the HAZ, and then through the thickness of the pipe, to cause failure of the pipeline.

Welds / Pipes / Heats

149. The four stages of the development of the welding procedure were as follows:

(1) Weldability Testing – carried out by Dalmine;

(2) Welding Procedure Qualification Testing (“WPQT”) – carried out by the welding contractor, McDermott-ETPM (“MET”);

(3) The Welding Specification was produced based on the WPQT, and provided detailed instructions to the welder as to the welding process, weld joint configuration and the welding parameters to be applied in order to control the welding process used on the laybarge; and

(4) Actual Welding.

150. I refer to the following explanations and diagrams:

(i) The description of the Saturnax mechanised Gas Metal Arc Welding pipeline welding system employed by MET contained in Mr. Marlow’s first report and the diagrams of the torch welding system and the weld pass sequence.

(ii) The three dimensional representation of a girth weld found in Dr. Golightly’s first report, identifying the weld cap and the weld root.

(iii) The definition of “Heat-Affected Zone” in NACE MR0175-93 as “That portion of the base metal that was not melted during brazing, cutting or welding, but whose microstructure and properties were altered by the heat of those processes.”

(iv) The representation of a simulated weld in Mr. Marlow’s report and Dr. Golightly’s illustration of the development of SSCC.

151. As to the numbering of Heats, Pipes and Welds:-

Each heat of steel produced by Dalmine was given a unique reference number at the Mill: a three figure number prefixed with “935” to make a six figure number. Each pipe was given its own unique sequential reference number starting from 60000. Section 11 of the Specification required the pipes to be marked: “the unique pipe number shall be die stamped on the end faces of the pipe”. Dalmine kept Tally Sheets recording the pipes which had been manufactured from each heat. A total of 2617 pipes produced by Dalmine were used to make a pipeline 31.5 km long. Not every pipe from each of the 54 heats supplied was incorporated into the pipeline. Some, but not many, were left over as “spare”. There were a total of 2616 welded joints in the pipeline. The pipeline was welded together and laid from a laybarge which proceeded from Lennox to Douglas. Section 13 of the Memorandum of Technical Facts describes the process of “pipeline construction”. There were three welding locations on the laybarge – two for welding pairs of pipes together end to end to produce a so-called double joint (“DJ”) and one for welding DJs into the Main Line (“ML”). Each welded joint was given its own unique reference number. There was a series of sequential reference numbers, including both odd and even numbers and beginning with the letters DJ, for double joint welds, and a separate series of sequential reference numbers, all of which were odd numbers and beginning with the letters ML, for main line welds. A record was kept of the sequence of welds in the pipeline. A record was also kept of the pipes either side of each weld. There is a record of the location of each pipe incorporated into the pipeline: see the ‘12” Lennox – Douglas Laying Ramp Pipe Reports – As Built Records’.

The leak sites

152. Following the discovery of the first leak in the pipeline on 7 June 1996, the pipeline was surveyed, and by 22 June 1996 the six leak sites had been located. No further leaks were discovered by the time of the end of the second survey on 8 July 1996.
153. The Memorandum of Technical Facts describes the commissioning of the pipeline, the period and conditions of its operation, and the discovery of its failure in Sections 15, 16 and 17 respectively.
154. Although when the pipeline was first surveyed following discovery of a leak, 18 suspected leak sites were identified, it was subsequently determined that the pipeline had failed only at 6 locations.
155. The “ROV Pipeline Inspection Survey Report” contains a diagram of the pipeline and the identification of the suspected leak sites.
156. The numbering of the leak sites is explained by the fact that the Remote Operated Vehicle (“ROV”) began its survey at the first identified leak site (1) and moved from there to the Douglas end finding suspected leak sites (2) to (15), before returning to (1) and surveying the longer length of the rest of the pipeline from there to Lennox finding only three further suspected leak sites (16) to (18).
157. Confirmed Leak Sites A to F were at suspected leak sites (1), (4), (7), (8), (12) and (14), respectively: see the table at the end of section 19.2 of the Memorandum of Technical Facts.
158. The leak sites were clustered in a section of the pipeline closer to the Douglas than the Lennox end. The pipeline has an overall length of 31.7 km. The section of the pipeline between leaks A and F measures 1.15 km.
159. The failed welds comprised:-

(1) 2 double joints: A DJ01114

B DJ01115; and

(2) 4 main line welds: C ML02043;

D ML02089;

E ML02101; and

F ML02127.

Fracture Mechanics

160. The condition necessary for crack propagation to occur is that the forces applied to the surrounding material at the tip of a crack exceed the strength of the material surrounding the crack. The force applied by the crack is “the applied stress intensity factor” denoted by K (or K_applied). The material property which is a measure of its strength or resistance to crack propagation is referred to as the “threshold stress intensity factor” and is denoted by K_1SCC.
161. Fracture mechanics analysis predicts that a crack will grow if the applied stress intensity factor (K, or K_applied) associated with the weld root crack exceeds the threshold stress intensity factor (K_1SCC) of the materials through which the crack is to propagate.

K – Applied Stress Intensity Factor

162. It is common ground that for a crack of a given shape and location (geometry), K_applied will increase as the crack gets bigger.

K_1SCC – Threshold Stress Intensity Factor

163. Attempts were made to measure the material resistance property (K_1SCC) of materials with various CEVs using the Double Cantilever Beam tests described by (and first carried out by CAPCIS under the supervision of) Professor Burdekin.
164. These are tests carried out in acidic NACE solution to simulate a corrosive environment using prepared cracks to simulate cracks that might have initiated in the weld root. In short, a value for the resistance of the pipe material to crack propagation (K_1SCC) is obtained on observing how far the crack propagates through the material.
165. Dr Baker was critical of some of the very low results of the first set of tests obtained for high CEV pipe material. His criticisms were accepted by Professor Burdekin who applied a technique referred to as “finite element analysis” to recalculate a modified (higher) K_1SCC value for those results which had fallen outside the validity limits of NACE TMO177-96 because of more extensive crack growth. The revised figures are set out in a table annexed to Professor Burdekin’s (amended) second report.
166. Two additional sets of tests were conducted (one for Dalmine and one for the claimants) by The Test House, the results of which were reported in December 2001. These tests and the results are summarised by Dr Baker in his first supplemental report.
167. The results of the tests show that there is a degree of scatter in the measurements of K_1SCC for pipe material with a given CEV.

Points of agreement by expert witnesses

168. A meeting of the experts took place at the offices of CAPCIS Ltd., Manchester on 18, 19 and 20 December 2001 to prepare a ‘Statement of Points of Agreement and Issues’ for the benefit of the Court. The following matters were agreed.

Introduction

169. Cracking was caused by SSCC.
170. When crack initiation occurred, it did so in weld metal at the weld root.
171. Through-thickness cracks developed by a two-stage mechanism. Stage 1 involved crack initiation in the weld root. Stage 2 involved crack propagation through the weld HAZ and parent material of the pipe.
172. Most Stage 1 cracks did not grow out of the weld root.
173. Crack propagation through the pipe wall (Stage 2) leading to leakage of gas occurred at six failure sites designated A to F.
174. There were no other known failure sites.
175. At each of the 6 failure sites at least one of the pipes had a CEV which did not meet the Specification requirements, i.e. the CEV was greater than 0.40%. The average CE of out of specification pipes at Leak Sites ranged from 0.419 to 0.447%.
176. The initiation of cracking in the weld root (Stage 1) was caused by hydrogen embrittlement (SSCC) resulting from corrosive attack by the wet and sour environment within the pipe.
177. Stage 1 cracks occurred in the root of some welds joining compliant pipe materials and in some welds where one or both pipes had a CEV above the permitted maximum.

Corrosive Environment

178. The pipeline operating conditions were such that the H₂S levels in service were greater than those anticipated at the design stage and greater than those permitted by the pipeline Works Authorisation.
179. The combination of intermittent operation of the pipeline and no batch inhibition or sphere pigging would reduce the effectiveness of the inhibitor in preventing corrosion in the pipeline.

Welding and Weld Root Hardness

180. The mechanised welding equipment was typical of that used for welding of pipelines on laybarges offshore.
181. Regions of high hardness in excess of the BS 4515 limit of 250HV made the weld root susceptible to crack initiation.
182. Hardness in excess of 250HV was observed in the weld metal at the root of welds joining pipes which complied with the CE limit of 0.40% and in welds where one or both pipes had a CEV greater than the permitted maximum.
183. Both the average and maximum hardness of the weld metal in the weld root were influenced (increased), but only marginally, by the CE of the parent pipes as shown by the hardness traverses made in accordance with BS 4515. This can be explained by the relatively small increase in the weld root CE for welds joining low CE pipes and welds joining high CE pipes due to the effect of dilution. If there were no variables in the welding procedure, metallurgical principles dictate that the weld and HAZ hardness would increase as parent pipe CE increases.
184. The scatter in measured weld root hardness between different welds is attributed to differences in cooling rate, composition and the method of measurement.
185. Post failure hardness tests in the HAZ at the weld root, on samples examined in accordance with BS 4515, gave some results which were in excess of the BS 4515 limit of 250HV. Such results were found in pipes which did and did not satisfy the specified CE requirement of 0.40% maximum.
186. The mean root HAZ hardness increased with increasing CE, consistent with metallurgical principles.
187. The variation in HAZ hardness on post failure specimens, measured in accordance with BS 4515, for a particular CEV can be attributed to welding procedure variables and local variations in hardness within the HAZ.
188. The acceptance of a Welding Procedure Specification with a minimum specified preheat of 50⁰C was outwith the requirements of BS 4515 since it was qualified with a minimum preheat of 60⁰C.

Crack Growth and Fracture Mechanics Issues

189. Where propagation of the root crack occurred (Stage 2), it did so through the weld HAZ and through the wall thickness of the pipe by SSCC.
190. Fracture mechanics is an appropriate means of assessing the potential for weld root cracks to propagate through the wall thickness of the pipe.
191. Crack propagation (Stage 2) occurs when the stress intensity factor (K) associated with the weld root crack exceeds the threshold stress intensity factor (K_ISCC) of the materials through which the crack has to propagate.
192. The applied stress intensity factor (K) depends on the total stress acting in way of the initiating crack (i.e. pressure-induced hoop stress and welding residual stress), the crack depth, shape and location.
193. At the weld, the welding residual stress is the dominant stress. The near surface hoop residual stress levels at the weld root found by TWI were 342 and 399 N/mm² (average value 83% of specified minimum yield strength). The pressure stresses at 95 bar and 128 bar are 112 N/mm² (25% SMYS) and 152 N/mm² (34% SMYS) respectively.
194. The level of residual stress was a consequence of the strength of the weld consumable used for the pipeline.
195. A meeting of the experts took place on 21 February 2002 concerning the determination of stress intensity factors and K_1SCC test values. The following matters were agreed.

CALCULATION OF K_APPLIED

Weld Strength

196. Results of tensile tests on four all-weld metal samples are available. These include one weld procedure test and three joints recovered from the failed line. The initial level of residual stress after welding depends on the weld metal yield strength.

Pressure

197. The hoop stresses in the pipe at a pressure of 95 bar are 112N/mm² and at 128 bar are 150N/mm², based on the internal diameter.

Proof Load Relaxation

198. Some relaxation of residual stress occurs as a result of the proof pressure test. The amount of relaxation depends on the initial level after welding and the stress level in the proof pressure test.
199. The optional procedure for calculating the relaxation in the TWI computer program Crackwise 2 uses an equation which assumes that a crack is present at the time of the proof pressure test. This is not appropriate for the present case.

Crack Shape

200. Stage 1 cracks may develop with a range of depths and widths. The crack shape depends upon the stress field and the material characteristics in the region concerned.

Values of K_applied

201. The TWI computer program Crackwise 2 is an acceptable basis for calculating K_applied.
202. The calculations for K_applied in Table 3 of Professor Burdekin’s first report and in Table 1 of his second supplemental report are based on a crack shape for which the depth and width are equal.
203. For a semi-circular surface crack (i.e. length on the surface twice the depth) the K_applied will be higher than for a crack with equal depth and width.

Use of Finite Element Analysis by Professor Burdekin

204. As to the K_1SCC data estimated from the FE results (whereas Dr. Baker accepted the validity of the FE analysis but questioned the validity of the K_1SCC results derived in this way and Professor Burdekin considered that these were reasonable estimates of what the K_1SCC would have been and noted that results for valid specimens are broadly consistent with the approach adopted) neither expert considered this to be of great importance with respect to the influence of CE on K_1SCC.

The issues identified by the parties

205. I turn to consider the issues identified by the parties. In doing so it is important to remember Professor Burdekin’s warning (which I accept) that the results of fracture mechanics analyses should be used only to indicate general trends and not to predict precise behaviour in the pipeline. There are a number of assumptions made in the analyses which, whilst they are best estimates of expected behaviour, will be subject to some uncertainty or variability in the pipeline itself.

I. The Resistance of Compliant Pipes within the Failure Zone to Crack

Propagation

206. Issue 1: How resistant to crack propagation would compliant pipes have been within the Failure Zone?
207. Sub-Issue 1.1: What was the upper limit of the CEV of “compliant pipes” according to the Specification and Data Sheet?
208. The claimants conceded (for the purposes of this case) that the specification of the maximum CEV of 0.40% includes pipe with a CEV of up to 0.4049%.
209. I emphasise that the composition of the pipe should have been within the specification at each end and at all points in between.
210. Sub-Issue 1.2: What was the actual CEV of each of the pipes within the Failure Zone?
211. The Failure Zone contained 144 pipes. The full results of post failure analyses (“PFAs”) are set out in the Memorandum of Technical Facts. Such data as is available in respect of the CEV of Dr. Baker’s blue (0.39% CE) and red (0.40% CE) pipes (in his second supplemental report/revised second supplemental report) is reproduced in the claimants’ Schedule entitled “Post-Failure Analyses (Red/Blue)” (“the PFA Schedule”). Ten laboratories carried out analyses of the chemical composition of pipes post-failure and calculated CEVs based on their measurements. No pipe was tested by every laboratory. The number and combination of laboratories testing each pipe varied. I do not consider that there is any justification in the present case for disregarding the CEV measurements of any of the ten laboratories. By way of example:-

(1) STL measured pipe 61385 from heat 935340 at 0.396% whereas Wilan measured the same pipe at 0.421%. In the case of pipe 62363 from heat 935581, STL measured the pipe at 0.401% and Wilan measured it at 0.369%.

(2) STL measured pipe 61583 from heat 935328 lower than TWI, but measured pipe 61355 from heat 935332 higher than TWI.

(3) STL’s measurements from the two ends of pipe 61034 from heat 935339 showed a variation within that pipe of 0.402% and 0.375%.

Tested Pipes

212. Of the 36 blue/red pipes relied on by Dalmine, only 10 were tested post-failure. These are identified in the second column of the PFA Schedule. So far as the actual CEV of the tested pipes is concerned, regard should be had to each PFA that is available. A pipe which was measured as out of specification by one or more of the laboratories carrying out PFAs, should be considered to be non-compliant. It follows that, although the average analyses of pipes 61621 and 61623 (from heat 935516) and pipe 63299 (from heat 935557) are all within specification, because they were each found to be non-compliant by particular laboratories, they should be excluded. STL measured all of them to be out of specification. STL’s view was confirmed in the case of pipe 61623 by a measurement taken by ATC.

Untested Pipes

213. Something was fundamentally wrong with the manufacturing process of the pipes. There is no evidence as to what that was. Pipes were supplied with a CEV above the specified limit, but there is no evidence as to why that was. When cross-examined Dr. Prosser provided valuable evidence as to how an honest and reasonable manufacturer would proceed in the case of a CE limit of 0.40%. Such a manufacturer would set his target at a mean in the order of 0.37%. A range of ( or (.02% might be expected either side of 0.37% (i.e. 0.35% to 0.39%). The distribution of CEVs would normally be such that one would expect a smaller number of pipes at the extremes, the lower end and the higher end. But Dalmine was not an honest and reasonable manufacturer. In all the circumstances of the present case I do not consider that it is possible to answer the question posed in respect of the untested pipes save to say:

(1) Given the very serious nature of Dalmine’s conduct and the extent of Dalmine’s failings as set out in this judgment (a) it is not possible to determine from the very limited data the mean CEV for any particular heat; and (b) it should not be assumed that all pipes in a heat would have been in a range of plus or minus 0.02% of any mean figure.

(2) The very limited data suggests that a significant proportion of the untested pipes were probably out of specification.

214. Sub-Issue 1.3: Which heats fall to be excluded from consideration in the light of the parties’ agreement of 17 January 2002?
215. Dalmine had not at the time of the Memorandum of Agreement raised the case advanced by Dr. Baker for the first time in his second supplemental report served on 15 February 2002. I repeat that the service of the report on 15 February was contrary to the overriding objective, contrary to the Court’s directions as to time for service of experts’ reports and contrary to the proper conduct of commercial litigation. It is most regrettable that Dr. Baker was instructed to prepare the report as late as (I was informed) 12 February. It is further most regrettable that the service of the report (with the defendant’s opening skeleton argument) should occasion a dispute as to the construction and effect of the Memorandum of Agreement made between the parties (represented by counsel and solicitors) as recently as 17 January. The parties (and particularly for these purposes the claimants) did not contemplate that it would or might govern what was an unforeseen situation. By virtue of the Memorandum of Agreement I must consider the question of causation on the premise that no non-compliant pipe would have been incorporated anywhere in the pipeline. The analysis of and the answer to sub-issue 1.2 is repeated. No non-compliant pipe would have been incorporated into the pipeline if Dalmine had set its target at a mean in the order of 0.37% and had achieved the expected range of plus or minus 0.02% either side of 0.37% (i.e. 0.35% to 0.39%). The distribution of CEVs would then normally have been such that there would have been a smaller number of pipes at the extremes, the lower end and the higher end. But Dalmine was not an honest and reasonable manufacturer. I repeat that I must consider this case on the premise that no non-compliant pipe would have been incorporated anywhere in the pipeline. Given this premise there is force in the claimants’ submission that the Court should exclude from consideration all pipes from any heat shown by post-failure testing to contain one or more non-compliant pipes, being (on the assumption that a pipe with any PFA measurement by any laboratory in excess of 0.40% should be regarded as non-compliant) the following heats: 935340, 935516, 935328, 935557, 935332, 935334, 935339, 935584 and 935581. If (contrary to the above) any of these heats should not be excluded, at the least very considerable caution should be exercised before making any of the assumptions that Dr. Baker sought to make in respect of pipes from these heats within the Failure Zone.
216. Sub-Issue 1.4: What, if any, figure should the Court take as being the CEV of pipes which Dalmine would have supplied (instead of those pipes which it would not have supplied had it not been fraudulent) or which fall to be excluded as a result of the agreement of 17 January 2002?
217. I find that if Dalmine had behaved honestly and reasonably it would have been concerned to avoid the risk of supplying pipes which did not comply with the Specification, and would accordingly have taken all necessary steps to produce pipes with a CEV at a target mean of about 0.37%. A range of plus or minus 0.02% might have been expected with a smaller number of pipes at the extremes. Dalmine’s suggestion that any replacement pipe would have matched the profile of the pipes supplied (and not excluded from consideration) is, I find, unrealistic.
218. Sub-Issue 1.5: On the assumption that only compliant pipes had been incorporated into the pipeline, what was the resistance to propagation of compliant pipe within the Failure Zone?
219. There is limited data as to the resistance of pipes to the propagation of cracks (K_1SCC). In order to compare the results obtained from testing each of a limited number of samples, the CEVs of each sample were measured by TWI and the results were plotted against the CEV as measured by TWI.
220. Measurements of K_ISCC indicate that resistance to crack propagation decreases as CE increases. There was broad agreement between Dr. Baker and Professor Burdekin regarding interpretation of the empirical data in terms of average K_ISCC values as a function of CE as set out in Table 1 below.
TABLE 1

CE (%)	0.36	0.37	0.38	0.39	0.40	0.41	0.42	0.43
BURDEKIN Mean K1SCC (N-mm-3/2)	2235	2050	1870	1685	1500	1320	1133	950
BAKER Mean K1SCC (N-mm-3/2)	2227	2059	1890	1720	1550	1390	1220	1060

II. Cracks in Welds in the Failure Zone.

221. Issue 2: What range (as regards depth) of cracks was present, or was likely to appear, in welds between compliant pipes within the Failure Zone and what was their likely propensity to propagate?
222. Sub-Issue 2.1: What size cracks had initiated and/or propagated pre-failure and what was their distribution between the welds?
223. The results of post-failure assessment of the size of cracks in welds recovered from the pipeline are set out in the table attached to Dr. Baker’s second supplemental report (now revised). Post-failure examinations revealed only three welds with cracks deeper than 4mm. Dr. Baker counted the following welds as having cracks of at least 4.5mm: ML2011 (4.8mm), DJ1114 (>4mm) and ML2089 (5mm). DJ1114 was at Leak Site A (two non-compliant pipes); ML2089 was at Leak Site D (one non-compliant pipe).
224. There is no basis for Dr. Baker’s suggestion in his probabilistic analysis of a 20% incidence rate of cracks of 4.5mm. It is based on a small non-representative sample which includes two joints from leak sites containing one or two non-compliant pipes.
225. The claimants accept that the specified minimum pre-heat of 50C for the weld procedure to be used on the lay barge was insufficient and therefore was a cause of hardness in the welds. But it does not follow that all welds were excessively hard. The claimants also accept that an increase in cooling rate will have some effect on the hardness in the weld, but they submit that the CEV of the pipes either side of the weld had an effect on the hardness in the weld.
226. Professor Burdekin referred to the TWI ECA Report and a comparison between 12 inch line material and 20 inch line material (different steel, but low CE). As to the latter he said “You do not get the hard zones in the weld root. Those welds were made with the same welding consumables, same equipment, same procedure and you get very little evidence of hard zones in the weld root, because the CE of the parent steel is much lower”.
227. I accept Professor Burdekin’s view that on the basis of general metallurgical principles it is more likely that larger zones of higher hardness would form in the weld root (for the same welding conditions) with higher CE pipes and hence, that if initial cracks formed, they would be larger in the weld roots joining higher CE pipes under exposure to wet sour gas conditions, even though there is insufficient empirical evidence in this case. The contribution of this effect should not however be overstated.
228. Sub-Issue 2.2: What scope was there for the further initiation and/or propagation of cracks had the pipeline not been de-pressurised, given further time and/or changes in environmental conditions?
229. To the extent that Dr. Baker advanced a new theory of interim crack development in cross-examination, it should be noted that he did not set this out in clear terms in his reports. I prefer Professor Burdekin’s analysis of this aspect of the case.
230. The claimants accept that the pipeline had only experienced pressures of 95 bar for any sustained period of time, and that increasing the operational pressure of the pipeline to its maximum of 128 bar would have increased the stress intensity factor (K_applied) of existing cracks so as to increase their propensity to propagate.
231. In his first supplemental report Dr. Golightly (who has greater expertise in the field of corrosion than any of the other expert witnesses) wrote

“Dr. Baker observes that gas injection commenced on 13 April 1996 and the first leakage was reported to have been detected on 7 June 1996, an intervening period of 55 days.

I believe that exposure of the pipeline to the sour gas environment was considerably longer than 55 days.

Gas was introduced into the pipeline when pipeline pressurisation commenced on 8 April 1996.

The discovery of the pipeline leakage was as follows:

On 7 June 1996 the Training Ship Winston Churchill reported gas rising to the sea surface.

On 13 June 1996 the Grampian Supporter reported “big bubbles”.

On 14 June 1996 pipeline depressurisation was commenced as a precaution.

On 18 June 1996 it was confirmed that the Douglas to Lennox gas reinjection pipeline was leaking and an ROV survey from the vessel Ocean Stephaniturm was begun. This was completed on 22 June 1996.

Depressurisation of the pipeline, which commenced on 14 June 1996 was a gradual process and pipeline pressure did not fall below 60 bar until 30 June 1996 whereafter rapid depressurisation to approximately 4 bar occurred over the following 3 days.

The total time of exposure of the pipeline to the pressurised sour gas environment was from 8 April to 30 June 1996, a period of 83 days.

Thus, contrary to Dr. Baker’s suggestion that the exposure period “...was not dissimilar...” to laboratory test durations, this exposure period was considerably longer than the duration of tests (30 days) designed to establish the resistance of materials to SSCC specified in published Standards (NACE TM0177, EFC 16) and the duration of the full ring tests (30 days) which were carried out.

The significance of this exposure period relates to the time required for crack initiation and growth to occur”

232. Regrettably Dr. Golightly was taken ill in the course of the hearing and had to go into hospital. By agreement his reports were admitted under the Civil Evidence Act. For this reason I have had to be careful as to the amount of weight I give to the contents of his reports, because his evidence was not tested in cross-examination.
233. While recording the above, when the evidence is looked at in the round, I consider that if further crack growth was going to occur, it would probably have already taken place within the period during which the pipeline was in service. In reaching this conclusion I have had regard to all the points made by the defendant’s experts to the contrary. On balance I prefer the analysis and reasoning of the experts called/instructed by the claimants. The NACE Standard testing procedure for determining K_1SCC requires specimens to be exposed to the full NACE solution environment for a period of 14 days. The length to which the crack has grown in that period is used as the basis for determination of K_1SCC, i.e. the value of the stress intensity factor below which no further crack growth will occur for the same stress level and environment however long the sample remains exposed. Other stress corrosion related tests use periods of up to one month to determine susceptibility. Thus a period of exposure is required to evaluate resistance to SSCC and this is recognised by standards such as NACE, and TM0177 which specifies a test period of 30 days for stress corrosion cracking tests. A similar test duration is specified by EFC 16. I take into account the fact that the pipeline did not experience the maximum operational pressure of 128 bar. Despite this the actual period of exposure of the pipeline to the very aggressive environment was such that all regions of all weld roots which were exposed to this environment and which were susceptible to SSCC would probably have cracked. Failures occurred only at welds where at least one of the pipes joined had a CE value greater than 0.40%. No leakage occurred at a compliant joint.
234. Sub-Issue 2.3: What was the stress intensity factor likely to have been at the tip of the crack of the size which Dalmine contends would be sufficiently large to cause failure of compliant pipes?
235. In general, the stress intensity factor (“K_applied”) at the tip of any crack increases as the size of a crack increases.
236. The calculation of K_applied depends on:

(1) the residual and applied stresses in the welds.

(2) the size of the crack in the weld.

(3) the shape of the crack in the weld.

(4) the location of the crack in the weld. (K_applied for “surface-emergent” cracks is greater than for “embedded” cracks.)

(Pipeline pressure is part of/contributes to residual and applied stress).

237. The appropriate applied stress for the Court to consider is that which results from a pipeline operating pressure of 128 bar.
238. As to yield strength, measurements of the strength of as-deposited weld metal are available from four welds.

(1) at the WPQT, a yield strength of 438 N/mm² was measured;

(2) ML1571 was measured at 542;

(3) ML2089 (Leak Site D) was measured at 587;

(4) DJ1115 (Leak Site B) was measured at 612.

239. I prefer the evidence and reasoning and approach of Mr. Marlow and Professor Burdekin (and to the extent that he agreed with them, Dr. Prosser) all of whom were content with a yield strength of around 542/544, to the evidence of Dr. Baker.
240. As Professor Burdekin pointed out one would want to examine the situation where the CE was within specification. The only sample where the CE was within specification was at the WPQT (yield strength of 438 measured). Professor Burdekin accepted that in practice there is evidence that the results will be higher than the weld procedure test. He said he thought he made a reasonable judgment in going to 542.
241. As to the relaxation formula, Dr. Baker produced calculations using both the TWI formula preferred by Professor Burdekin and BS 7910 formula. Both formulae are conservative - the TWI formula is less conservative than the BS 7910 formula. Professor Burdekin preferred the TWI formula because in his opinion it was more appropriate when considering what was likely to happen. I prefer Professor Burdekin’s evidence in this connection. Dr. Baker accepted that for the purposes of the exercise on which the Court is engaged, it was reasonable to use the TWI formula.
242. As to shape of cracks, I prefer the evidence of Professor Burdekin to the evidence of Dr. Baker.
243. Where leakage occurred at welded joints in the pipeline this involved a two-stage cracking mechanism. The first stage involved the initiation of a crack in weld metal at the root of the joint. The second stage involved propagation of this crack through the heat-affected zone (HAZ) into the parent material leading to complete penetration of the pipe wall at the six locations of pipeline failure.
244. Some welded joints recovered from the pipeline contained weld root cracks which had not cracked through the wall thickness of the pipe. The great majority of root cracks found were confined to weld metal. In joint ML 2011 (which was not associated with a leak site) a crack was measured to have a height of 4.8 mm. At joint ML 2011, both pipes complied with the CE requirement of 0.40% maximum. I prefer Professor Burdekin’s opinion that this crack had arrested because the applied K₁ value was less than the material K_1SCC resistance, although some propagation had occurred in order for the crack to have grown from the weld root and into the HAZ. This indicates that the initial crack dimensions had been less than the 4.5 mm height now observed for this crack.
245. For pipe material having a CEV of 0.40% and the mean K_1SCC for that material, Dr. Baker considered that at the operating pressure of 95 bar, the depth of surface defect required to cause crack propagation would be 4 to 5 mm (depending on crack shape and weld strength). At the maximum operating pressure of 128 bar, the depth of surface defect required to cause crack propagation would be 3 to 5 mm.
246. Professor Burdekin considered that at a pressure of both 95 bar and 128 bar the depth of embedded crack required to cause crack propagation would be more than 8 mm. For surface emergent defects the corresponding defect heights would be of the order of 6 mm at 95 bar and 5 mm at 128 bar. In relation to the weld root crack in joint ML2011, Professor Burdekin considered that the crack had arrested because the applied K value associated with the crack depth of 4.5 mm to 5 mm would have had a value of about 1450 N-mm^-3/2 which is less than an estimated mean value of K_1SCC of about 1870 N-mm^-3/2 for pipe material having a CE of 0.38%.
247. I prefer the evidence and reasoning of Professor Burdekin to that of Dr. Baker.

III. Approach to Probability

248. Issue 3: Can the Court assess the probability of the combination of factors required to produce the situation in which a crack would develop in a weld between compliant pipes within the Failure Zone, which would be sufficiently large to overcome the resistance offered by (compliant) pipe either side of the weld, to the propagation of that crack through the thickness of the pipe material? If so, how should it go about doing so?

Dr. Baker’s first supplemental report

249. In his first supplemental report Dr. Baker wrote:-

“The K_1SCC data obtained in the present tests, and the previous data which satisfied the specified validity criteria, when combined with the results of the metallurgical studies relating to the size of cracks which could be expected in the weld root, indicate that if the pipeline had not been depressurised, further cracking would have developed from several joints in which both pipes satisfy the CE requirement. Similarly, if the pipeline had contained pipes having CE values up to, but not in excess of the specification limit, crack propagation leading to multiple leakage sites would still have occurred. Consequently, it would still have been necessary to replace the pipeline.”

250. Dr. Baker’s evidence as to the “metallurgical studies” referred to was inconsistent and unsatisfactory. I prefer Professor Burdekin’s evidence as to the issue of the applicability of BS 7910. The passage quoted above referred to the pipeline as a whole. It should be remembered that Dalmine confirmed at the hearing that it does not rely on any pipes outside the Failure Zone in support of its case.

Dr. Baker’s probabilistic analysis

251. Dr. Baker’s probabilistic analysis was produced without prior warning on the eve of the trial. It is contained in Dr. Baker’s second supplemental report but was amended through his fourth supplemental report (which was withdrawn), and his revised second supplemental report.
252. Dr. Baker was put in a difficult position by being instructed to prepare a probabilistic analysis in the week immediately before the trial. Had he thought such an analysis appropriate, he would have undertaken it earlier and would have sought advice from a statistician. Dr. Baker’s probabilistic analysis did not comply with any of the following requirements to be expected in any expert’s report:-

(1) It did not summarise the facts and instructions given to Dr. Baker which were material to the opinions expressed in the report or upon which those opinions were based and did not set out any assumptions made and any uncertainties.

(2) It did not set out a summary of the range of opinion and the reasons for Dr. Baker’s own opinion.

(3) It did not express any qualification of or reservation to the opinion.

253. When the claimants’ solicitors raised proper and necessary questions in correspondence (following service of the second supplemental report) certain answers provided by Dr. Baker (as he acknowledged in evidence) were inadequate and incomplete, and did not address the question.
254. Dr. Baker’s analysis did not satisfy the criterion of meeting observed behaviour.
255. I refer to and accept the many criticisms that Professor Burdekin made of Dr. Baker’s probabilistic analysis in the Professor’s third supplemental report dated 21 February and in his oral evidence.
256. When cross-examined Dr. Baker accepted several of the criticisms that had been made by Professor Burdekin. As Dr. Baker eventually appeared to acknowledge, such an analysis requires the expertise of a statistician. Dr. Baker is not an expert statistician.
257. I turn to make the following particular points in relation to Dr. Baker’s probabilistic analysis (without prejudice to all the other criticisms that Professor Burdekin made).
258. There is no basis for Dr. Baker’s suggestion in his probabilistic analysis of a 20% incidence rate of cracks of 4.5mm. It is based on a small non-representative sample which includes two joints from leak sites containing one or two non-compliant pipes. A cardinal principle of sampling is that you should have random samples. Leak sites were not part of a random sample. Further the size of the sample was inadequate.
259. As Professor Burdekin observed “Probabilistic calculations are extremely difficult things to do properly and fully. The whole field of probability is a minefield for not getting all the right parts into the model and not getting all the data right. I am doubtful whether it could be done properly... I am certain it cannot be done with any high degree of confidence, with the data... available... I am surprised... that there is no indication in the probability analyses... of any confidence limits or uncertainties in the analysis.”
260. As to K_applied, Dr. Baker failed to reflect uncertainties about K_applied in his analysis. As to K_1SCC values Dr. Baker is not an expert as to material specification or purchasing or manufacture. As to manufacture, I observed Dr. Baker expressing concern during Dr. Prosser’s evidence. He should have consulted Dr. Prosser in relation to matters within Dr. Prosser’s expertise (and outside his own) and reflected Dr. Prosser’s contribution in assumptions recorded in his reports. Since the majority of pipes have not been tested at all (let alone by TWI) this gives rise to very considerable uncertainties. Dr. Baker’s analysis was dependent upon treating any untested pipe as if it had a CEV equivalent to what he asserted was the mean of the heat.
261. Dr. Baker did not adequately address the question - What would have been the position if Dalmine had performed the contract honestly and reasonably? Further Dr. Baker’s approach ignored the fact that (a) there is considerable uncertainty about the mean value of any particular heat since it was derived from a small sample, in many cases of one pipe; and (b) there is considerable uncertainty about the deviation from the mean; and (c) there is considerable uncertainty about the spread within a heat. K_1SCC tests were carried out on samples for which a TWI CE result was available. If an attempt is to be made to estimate K_1SCC values from CE values, one needs to know what a TWI result would be.
262. In my judgment Dr. Baker’s probabilistic analysis in his second supplemental report/revised second supplemental report was flawed. As to approach to probability, I prefer, accept and rely on the evidence of Professor Burdekin.

The pipeline failed at six sites. At each site at least one pipe did not meet specification requirements.

263. The pipeline failed at six sites. At each site at least one pipe did not meet specification requirements.
264. The failure of the pipeline with leakage at the six sites observed occurred because of the presence of material (at the welded joints concerned) which was out of specification in respect of CE.
265. No leakage occurred at a joint between compliant pipes.

K_1SCC values as a function of CE

266. Measurements of K_ISCC indicate that resistance to crack propagation decreases as CE increases. There was broad agreement between Dr. Baker and Professor Burdekin regarding interpretation of the empirical data in terms of average K_ISCC values as a function of CE as set out in Table 1 above.
267. According to Professor Burdekin for a carbon equivalent of 0.40%, the trend of the results would give an average value of K_1SCC of about 1500 N-mm^-3/2 whilst at a CE of 0.44% the average K_1SCC would be about 770 N-mm^-3/2 and at a CE of 0.38% the average K_1SCC would be about 1870 N-mm^-3/2.

Improbability of failure between compliant pipes

268. I repeat the analysis set out above.
269. When carrying out a fracture assessment at the design stage, the factors which contribute to the applied K_I (i.e. the applied and residual stress conditions and crack size) are chosen to represent an upper bound and the material resistance properties are then chosen to be above a lower bound so that the probability of failure is extremely remote.
270. The overall probability of failure in this case depends on the interaction between the probability of occurrence and distribution of sizes of any initial cracks in hard zones, the applied and residual stress levels and the actual distribution of values K_1SCC for each pipe and weld.
271. The evidence from service experience is that the pipeline did fail at welded joints where at least one of the pipes was out of specification, and did not fail at any joints where both of the pipes complied.
272. For material with a CE value of 0.40% and an average K_1SCC of 1500 to 1550 N-mm^-3/2 the height of an initial sub surface defect which would extend at the operating pressure of 128 bar is in excess of 8 mm and the height of surface defect is about 5 mm. For a CE value of 0.44% with an average K_1SCC of 770N-mm^-3/2 the corresponding heights of initial sub surface and surface defects are about 3 mm and 1.5 mm respectively. The probability of failure increases rapidly as the CE exceeds 0.40%.
273. For the reasons explained above the resistance of the pipes with CE values outside the Specification to propagation of sulphide stress corrosion cracks decreased progressively with increasing CE (“the first effect”).
274. It is more likely that larger zones of higher hardness would form in the weld root (for the same welding conditions) with higher CE pipes and hence, that if initial cracks formed, they would be larger in the weld roots joining higher CE pipes under exposure to wet sour gas conditions (“the second effect”).
275. While the contribution of the second effect should not be overstated, the combination of the two effects made the probability of failure significantly greater in the out of specification pipe material than in material which complied with the Specification.
276. Although small cracks might have formed in the root of welds between in specification pipes it is unlikely that they would have propagated extensively in the parent pipe to cause leakage, whereas such cracks in welds with out of specification pipes would be likely to propagate to cause leakage.
277. I do not consider it likely that leaks would have formed at joints between pipes which met the Specification given a longer period of service and exposure to wet sour conditions.
278. The failure of the pipeline with leakage at the six sites observed occurred because of the presence of material (at the welded joints concerned) which was out of specification in respect of CE.
279. Failures only occurred at welds associated with non-compliant material in the pipeline. A failure of the pipeline would probably not have occurred if all the material had complied with the Specification requirements.
280. The issue to be tried in this hearing is as follows. Did the incorporation of non-compliant pipe cause the pipeline to fail (as the claimants say) or would it have failed anyway (as the defendant says)? For the reasons set out above I find that on the balance of probabilities the incorporation of non-compliant pipe caused the pipeline to fail. For the reasons set out above I find on the balance of probabilities that the pipeline would not have failed anyway.