Thames Water Utilities Ltd. v London Regional Transport & Anor

The Claimants, Thames Water Utilities Limited (TW) supply water to most of London and the Southeast area. Part of the piped water supply system comprises the 36-inch cast iron mains such as that laid along the length of St. Thomas Street, Southwark in about 1925. It was laid to a depth of approximately 1.4 metres. Around it and above it in the succeeding 75 years, a variety of relevant events occurred, resulting in an accumulation of locked in strains within the rigid structure of the cast iron pipe. Some of these events were mundane and common place, such as the passing and re-passing of traffic above, indirectly transmitting forces below onto and into the grey cast iron pipe and others remarkable, and outside the ordinary, such as the direct hit by a bomb on the arch beneath London Bridge Station when 63 people were killed and the water main in St Thomas Street to the east was fractured and other bomb damage. A post war bomb damage map of the area indicates that numbers 21 to 27 St. Thomas Street were destroyed, along with the East Wing of Guys Hospital and buildings to the east of Great Maze Pond and Joiner Street.

On 24^th October 1999, the Claimant’s water main burst under St. Thomas Street slightly to the north west of the gateway to Guys Hospital in front of numbers 19/21 St Thomas Street. The burst caused substantial damage. The cost of remedial works and compensating third parties affected runs into many millions of pounds.

The Defendants (LUL) were engaged on the construction of the Jubilee Line extension. Contract 104 included the construction of London Bridge Station which involved extensive tunnelling, shaft sinking and compensation grouting work. The LUL works for all material purposes took place between about October 1995 and the end of 1997.

The Works were carried out pursuant to powers conferred under the London Underground Act 1992 and the London Underground (Jubilee) Act 1993. Costain Taylor Woodrow joint venture (CTW) were engaged to carry out the LUL works, and CTW in turn sub-contracted parts of the work to other companies.

Thames Water contend that LUL by their works caused the burst of the main in St. Thomas Street and are liable in consequence for the loss and damage which resulted from the failure of the main on 24^th October of 1999. It is evident that the failure of the main was sudden. There were fractures in the cast iron main at the spigot to socket joint with a large single piece at the top of the socket breaking out. This piece was bounded by two longitudinal fractures of length about 1.8 metres, these fractures being 1.3 metres apart at the socket end and about 0.6 metres apart at the transverse fracture joining the two longitudinal fractures. A diagram prepared on 12^th November of 1999 showing the arrangements of the broken pieces viewed from outside the pipe is agreed by the experts in this case. It was prepared by Dr T.J. Baker, a metallurgist from Imperial College, London on 12^th November of 1999. It was accompanied by a factual narrative description of what he found. The inference is warranted that had there been anything unusual, of metallurgical significance on the fracture surfaces he would have been in the best position to observe them.

T.W’s materials expert Prof. Burdekin at paragraph 51 of his first report observed:

“It is in controvertible that the main at St. Thomas Street did fail. It is therefore also self-evident that the applied stress levels must have exceeded the strength of the material. The stress levels from normal design conditions, including internal water pressure, external earth pressure and super imposed loading are low, of the order of 25MPa (equivalent to a strain of approximately 300 microstrain). The difference between the applied design stresses and those necessary to cause the observed failure must have come from additional extraneous loading”.

TW contend that the additional loading resulted in part from the LUL Works under Contract 104 and that under Section 42(10) of the London Transport Act 1963, LUL are strictly liable to:

“…make compensation to …..[TW] –

a)

for any damage caused to any apparatus…; and

b)

..any other expenses, loss, damages, penalty or costs incurred by the undertaker;

by reason or in consequence of the execution, maintenance, user or failure of any such works or otherwise by reason or in consequence of the exercise by the Board of the powers of this Act)”.

Mr Moxon-Browne QC on behalf of LUL contends that TW have not proved that the LUL works were a material cause of the pipe failure and “that TW have not demonstrated in a plausible way how its case on causation ‘technically/scientifically hangs together”.

LUL further contends that if the condition of the pipe at the time prior to the burst was that it was so stressed and thus so vulnerable that the slightest additional loading would cause bursting fractures that event properly would be characterised as ‘de minimis’ even though it was in time the proximate cause of the fracture, and to adopt Mr Moxon-Browne’s phrase merely “the straw that broke the camel’s back”. Whilst the loading would properly be described as causative, if any additional loading would have sufficed whatever its origin, whether point or indirectly transmitted such as from excessive road user, that would not be sufficient cause. To be causative in order to fix liability upon LUL any additional loading would have to be proved to be the result of the LUL works and a material cause, that is not de minimis.

The claim of Thames Water against the defendants is pleaded as three alternative causes of action, a claim for a statutory breach under section 42 of the London Transport Act 1963; a claim at common law in negligence for failure to use reasonable skill and care in carrying out and completion of the Works and in nuisance for removing and/or undermining ground support from the main as part of the construction of the Works.

There is sufficient authority for the proposition that there should be a single test of causation applicable to the same set of facts, irrespective of whether the case is pleaded in negligence, nuisance or for a statutory breach. In theory there may be situations where different tests of causation should apply in negligence and for a statutory breach as where the rules of liability regarding the scope and nature of the duty imposed on the defendant at common law and by a statute differ in such a way as to impose different requirements for a cause or connection that a claimant needs to establish. That would be determined by comparing the statutory duty to the relevant common law duty pleaded. In this case as a matter of construction, there is nothing in the wording of Section 42 of the London Transport Act of 1963, which would indicate that the defendants owed to Thames Water a statutory duty that in relevant and material respects differed from the obligations arising in the law of negligence and nuisance. Lord Reid in Bonnington Castings Ltd v Wardlaw [1956] AC 613, HL at page 620 said:

“It would seem obvious in principle that a pursuer or plaintiff must prove not only negligence or breach of duty, but also that such fault caused or materially contributed to his injury, and there is ample authority for that proposition both in Scotland and in England. I can find no reason nor authority for the rule being different where there is breach of a statutory duty… in my judgment, the (claimant) must in all cases prove his case by the ordinary standards of proof in civil actions: he must make it appear at least that on a balance of probabilities the breach of duty caused or materially contributed to his injury”.

The question of causation is a matter for factual enquiry by the court, which is traditionally undertaken by the application of the “but for” test. A helpful summary of this approach can be found in a passage from Clarke and Lindsell 17^th Edn. at 2-06:

“The first step in establishing causation is to eliminate irrelevant causes, and this is the purpose of the “but for” test. The courts are concerned, not to identify all of the possible causes of a particular incident, but with the effective cause of the resulting damage in order to assign responsibility for that damage. The “but for” test asks: would the damage of which the claimant complains have occurred “but for” the negligence (or other wrongdoing) of the defendant? Or to put it more accurately, can the claimant adduce evidence to show that it is more likely than not, more than 50 per cent probable, that “but for” the defendant’s wrongdoing the relevant damage would not have occurred. In other words, if the damage would have occurred in any event the defendant’s conduct is not a “but for” cause”.

However, at the end of the passage, the editors of C&L add the following caveat:

“It is worth bearing in mind that the “but for” test functions as an exclusionary test, i.e. its purpose is to exclude from consideration irrelevant causes. The fact that the defendant’s conduct is found to be a cause, applying the “but for” test, is not conclusive as to whether he should be held responsible in law since the function of the causal enquiry in law is to determine which causes have significance for the purpose of attributing legal responsibility. It is sometimes said that the law seeks the causa causans (effective factor) rather than the causa sine qua non (factor(s) without which damage could not have occurred).”

There has been considerable judicial discussion as to the limitations of the “but for” test in establishing necessary causal connections between the defendant’s conduct and the claimant’s loss. See Dean v Dean and others (2000) 80 P&CR 457, Lexis UK Property, 1975 CA, at para [31] where Peter Gibson LJ observed that it is rarely a sufficient test for ascertaining whether to that defendant is to be attributed the sole effective cause of loss in a case of tort.

In Banque Bruxelles Lambert SA v Eagle Star Insurance Co Ltd and others (unreported, Court of Appeal, 22 March 1996), an appeal on an aspect of the trial that ended in the House of Lords decision in Banque Bruxelles Lambert SA v Eagle Star Insurance Co. Ltd and others [1995] QB 375, Saville LJ said:

“These two findings of the judge undoubtedly satisfy a “but for” test of causation. But so too, of course, would an infinity of other facts. The real test however, is, in my judgment, much more pragmatic and based simply on common sense rather than philosophical or metaphysical considerations. It is whether the negligence was an effective cause of the loss. That test has been repeated many times in our courts.”

Otton LJ elaborated on the common sense approach:

“It is not necessary to identify the source of those expressions, but they include, “an effective cause”, “present in the mind and influenced”; “contributory”; “an inducing cause”; “operated upon the mind”; “relied upon, in a broad or narrow sense”; “gives weight to his decision”; “motivates him”; “encourages him”; “is one of the factors”.

Apart from the philosophical and metaphysical considerations, to which my Lord has referred, it is not helpful to consider any linguistic distinctions between these expressions. In the context of this case they would sit comfortably in a judicial thesaurus. The expression “but for” does not, in my view, add anything and I am sceptical that except in a general sense it has much value as a test for causation. To paraphrase, which I do gratefully, the words of McHugh JA in Alexander v Cambridge Credit Corporation Ltd (1987) 9 NSWLR 310 at p 359 ”the common law champions the common sense notion of causation”. It is trite to say that causation is essentially a question of fact in each case”.

In March v. E. & M.H. Stramare Pty. Ltd. (1991) 171 CLR 506, in the High Court of Australia, four of the five members of the High Court of Australia took the view that the “but for” test was not a definitive test of causation in tort. In the judgment of Mason CJ at 515:

“The common law tradition is that what was the cause of a particular occurrence is a question of fact which ‘must be determined by applying common sense to the facts of each particular case’, in the words of Lord Reid: Stapley v Gypsum Mines Ltd. [1953] A.C. 663, 681… It is beyond question that in many situations the question whether Y is a consequence of X is a question of fact. And, prior to the introduction of the legislation providing for apportionment of liability, the need to identify what was the ‘effective cause’ of the relevant damage reinforced the notion that a question of causation was one of fact and, as such, to be resolved by the application of common sense. Commentators subdivide the issue of causation in a given case into two questions: the question of causation in fact – to be determined by the application of the ‘but for’ test – and the further question whether a defendant is in law responsible for damage which his or her negligence has played some part in producing …… It is said that, in determining this second question, considerations of policy have a prominent part to play, as do accepted value judgments ….. However, this approach to the issue of causation (a) places rather too much weight on the ‘but for’ test to the exclusion of the ‘common sense’ approach which the common law has always favoured; and (b) implies, or seems to imply, that value judgment has, or should have, no part to play in resolving causation as an issue of fact. As Dixon CJ Fullagar and Kitto JJ remarked in Fitzgerald v Penn (1954) 91 C.L.R. 268, 277 ‘it is all ultimately a matter of common sense’ and ‘in truth the conception in question (i.e. causation) is not susceptible of reduction to a satisfactory formula’”.

This approach was endorsed, as the correct approach to causation, in Fairchild v Glenhaven Funeral Services Ltd and others [2003] 1 AC 32, HL, Lord Bingham said that Mason CJ did not at p.508 (supra)

“accept the ‘but for’ (causa sine qua non) test ever was or now should become the exclusive test of causation in negligence cases”,

and then referred to the passage from the Australian decision at 516:

“The ‘but for’ test gives rise to a well known difficulty in cases where there are two or more acts or events which would each be sufficient to bring about the plaintiff’s injury. The application of the test ‘gives the result, contrary to common sense, that neither is a cause’: Winfield & Jolowicz on Tort, 13^th ed (1989), p 134. In truth, the application of the test proves to be either inadequate or troublesome in various situations in which there are multiple acts or events leading to the plaintiff’s injury: see e g Chapman v Hearse, Baker v Willoughby [1970] AC 467; McGhee v National Coal Board; M’Kew v Holland & Hannen & Cubitts (Scotland) Ltd 1970 SC(HL) 20 (to which I shall shortly refer in some detail). The cases demonstrate the lesson of experience, namely, that the test, applied as an exclusive criterion of causation, yields unacceptable results and that the results which it yields must be tempered by the making of value judgments and the infusion of policy considerations.”

In Fairchild v Glenhaven, the House of Lords considered whether, and in which circumstances, the ‘but for’ test of causation could be relaxed. By reference to the specific facts of the case, Lord Bingham formulated the principles determining when the causal test for the purposes of establishing liability could be relaxed so as to allow the claimant to succeed in the circumstances when the claimant for reasons of scientific impossibility could not satisfy the ‘but for’ requirements. At para. 40, Lord Bingham set out the factual circumstances and the question raised by the appeals:

“The essential question underlying the appeals may be accurately expressed in this way. If (1) C was employed at different times and for differing periods by both A and B, and (2) A and B were both subject to a duty to take reasonable care or to take all practicable measures to prevent C inhaling asbestos dust because of the known risk that asbestos dust (if inhaled) might cause a mesothelioma, and (3) both A and B were in breach of that duty in relation to C during the periods of C’s employment by each of them with the result that during both periods C inhaled excessive quantities of asbestos dust, and (4) C is found to be suffering from a mesothelioma, and (5) any cause of C’s mesothelioma other than the inhalation of asbestos dust at work can be effectively discounted, but (6) C cannot (because of the current limits of human science) prove, on the balance of probabilities, that his mesothelioma was the result of his inhaling asbestos dust during his employment by A or during his employment by B or during his employment by A and B taken together, is C entitled to recover damages against either A or B or against both A and B? To this question (not formulated in these terms) the Court of Appeal (Brooke, Latham and Kay LJJ), in a reserved judgment of the court reported at [2002] 1 WLR 1052, gave a negative answer. It did so because applying the conventional ‘but for’ test of tortious liability, it could not be held that C had proved against A that this mesothelioma would probably not have occurred but for the breach of duty by A, nor against B that his mesothelioma would probably not have occurred but for the breach of duty by B, nor against A and B that his mesothelioma would probably not have occurred but for the breach of duty by both A and B together. So C failed against both A and B.

The crucial issue on appeal is whether, in the special circumstances of such a case, principle, authority or policy requires or justifies a modified approach to proof of causation.”

“To the question posed in paragraph 2 of this opinion I would answer that where conditions (1)–(6) are satisfied C is entitled to recover against both A and B. That conclusion is in my opinion consistent with principle, and also with authority (properly understood). Where those conditions are satisfied, it seems to me just and in accordance with common sense to treat the conduct of A and B in exposing C to a risk to which he should not have been exposed as making a material contribution to the contracting by C of a condition against which it was the duty of A and B to protect him. I consider that this conclusion is fortified by the wider jurisprudence reviewed above. Policy considerations weigh in favour of such a conclusion. It is a conclusion which follows even if either A or B is not before the court. It was not suggested in argument that C’s entitlement against either A or B should be for any sum less than the full compensation to which C is entitled, although A and B could of course seek contribution against each other or any other employer liable in respect of the same damage in the ordinary way. No argument on apportionment was addressed to the House. I would in conclusion emphasise that my opinion is directed to cases in which each of the conditions specified in (1)-(6) of paragraph 2 above is satisfied and to no other case.”

Lords Nicolls and Hoffman arrived at the same view holding that exceptionally a lesser degree of causal connection may suffice than the ‘but for’ test of causal connection.

The ‘but for’ test clearly continues to be the normal route to considering questions of causation. It is not determinative as to the question of causation. It may be a weighty ingredient when all the factual elements are evaluated not least whether the liability lies within the scope of duty to the claimant imposed on the Defendant by law. See Regina v Immigration appeal Tribunal ex parte Shah 1959 AC page 629

Similarly in Banque Bruxelles Lambert SA v Eagle Star Insurance Co. Ltd and others sub nom. South Australia Asset Management Corporation v York Montague Ltd [1997] AC 191. a decision on the issue of the measure of damages, Lord Hoffmann considered the following example that is illuminating on the legal principles of causation as well:

“A mountaineer about to undertake a difficult climb is concerned about the fitness of his knee. He goes to a doctor who negligently makes a superficial examination and pronounces the knee fit. The climber goes on the expedition, which he would not have undertaken if the doctor had told him the true state of his knee. He suffers an injury which is an entirely foreseeable consequence of mountaineering but has nothing to do with his knee.

On the Court of appeal’s principle, the doctor is responsible for the injury suffered by the mountaineer because it is damage, which would not have occurred if he had been given correct information about his knee. He would not have gone on the expedition and would have suffered no injury. On what I have suggested is the more usual principle, the doctor is not liable. The injury has not been caused by the doctor’s bad advice because it would have occurred even if the advice had been correct.

(…) Your Lordships might, I would suggest, think that there was something wrong with a principle, which, in the example, which I have given, produced the result that the doctor was liable. What is the reason for this feeling? I think that the Court of Appeal’s principle offends common sense because it makes the doctor responsible for consequences which, though in general terms foreseeable, do not appear to have a sufficient causal connection with the subject matter of the duty. The doctor was asked for information on only one of the considerations which might affect the safety of the mountaineer on the expedition. There seems no reason of policy which requires that the negligence of the doctor should require the transfer to him of all the foreseeable risks of the expedition.”

From the decisions that I refer to above, the following considerations apply to questions of causation in this case.

T.W contends that the pipe failed due to a combination of increased loadings caused by a number of factors. These included movements caused by the LUL Works, normal operation at loads such as the internal water pressure in the pipe, loads from the ground in which the pipe was located and traffic above, together with locked strains and strain concentration effects.

T.W contend that the fragmentation pattern of the pipe indicated a bursting at the joint due to radial forces from within the socket. The condition of the fracture surfaces and the location of the fragmented pieces indicated that the top of the pipe was prised or sheered upwards as a direct result of joint rotation and the relative displacement of the adjacent pipe sections. The distribution of strain within the pipe was very complex as was the variation of strength within the iron of the pipe, however, it is evident that no identifiable weaknesses in the iron, participated in the failure.

TW contends that ground movements caused by the LUL works gave rise to differential settlement in the area immediately surrounding the point of failure. Thus taken together, the measured differential settlements, the evidence of at least 2mm joint rotation at the spigot end of the failed pipe section and the fragmentation pattern indicate the causal mechanism.

It is TW’s case that the evidence of differential settlement is well tested and reliable and is confirmed by the monitoring data provided by LUL and indeed, the location of failure occurs at the point in St Thomas Street where on any view of the evidence the greatest differential ground movements were measured following the LUL works.

The ground conditions at the site demonstrate a complex geological structure which has the potential to promote both long and short term localised differential settlements. In similar ground conditions on the adjacent JLE Contract 105, unusual ground movements have been observed. TW contend that they were similar as to cause and effect.

TW argue that the timing of the failure is explained by the progressive undermining of the pipe that steadily increased the strain in the pipe to a critical value and/or by the localised collapse of metastable ground de-stabilised by the consolidation processes. It further relies upon the results of the full scale testing of a similar pipe joint at the Transport Research Laboratory which served to identify the strains in the socket due to the joint rotation caused by ground movement. The testing conditions were not designed to reflect the operational loads or the locked in stresses of such a pipe in the ground. An attempt at modelling the state of the pipe in the ground subject to such loads and stresses by finite element analysis was abandoned because of its inherent complexity.

At the planning stage, before LUL undertook their tunnelling and associated works they were in close contact with TW. It was obvious to both parties that by tunnelling under TW’s substantial pipe work system that there was a risk of some damage occurring. TW had identified the main in St. Thomas Street as being exposed to the real risk of damage from ground movements during the proposed LUL works. TW’s senior engineer on 2^nd June of 1995 in relation to contract 102 wrote to LUL engineers saying:

“I was asked to provide the maximum allowable settlement and change of slope to be tolerated by the 30 inch main and other mains. The amount of strain that can be accommodated within a cast iron pipe before failure is microscopic. Any change of slope can only be sustained if that movement is successfully transferred to a joint. Typical joints on cast iron mains are run lead. They are semi-flexible only on account of the yielding of the lead by cold flow under load. Manufacturers tolerances are intended for accommodating variations during laying. Any subsequent rotary movement will, if continued, shear the iron, splitting and rupturing the pipe. Any rotation of the pipe joint is therefore undesirable.”

This was equally valid in relation to the mains in St. Thomas St. in contract 104, and elsewhere.

LUL’s contractors took elaborate and sophisticated measures to minimise the risk to damage to some of the structures above the ground and within the ground in the vicinity of the tunnelling which were perceived to be at risk of damage. Firstly they sought to identify the extent of the ground that would be affected by the tunnelling process and the ancillary works. This was based upon their previous experience of tunnelling works in the London clays. A risk evaluation entitled “Prediction of surface settlement due to tunnelling” was carried out for LUL by the Consulting Engineers Mott MacDonald. The Contractor for the Works, Costain Taylor Woodrow (CTW) also carried out a ground movement predication study similar to that prepared by Mott MacDonald. Buildings thought likely to be influenced by LUL works were fitted with metal pins that could be regularly surveyed by precise levelling techniques. The buildings immediately adjacent to the location of the failure Mary Sheridan House at No.19 St Thomas Street and Nos. 21 to 27 St. Thomas Street were fitted with such measuring pins. What is apparent from the evidence is that the so called ‘predictions’ related only to one aspect of settlement consequent upon tunnelling, that is the vertical or tabular ground movement, resulting from the loss of ground due to the tunnelling. It is common ground between the experts that whilst the predicted settlements were calculated competently using accepted methods, they did not include the longer term movements resulting from subsequent consolidation and other long term ground movements which affect a greater extent of ground and more remote from the tunnelling or other physical works than the vertical or tabular displacements.

A comprehensive monitoring system was set up to measure the extent of the anticipated ground movements and the data comprising such measurements is relevant to the evaluation of the degree and location of ground movements in St Thomas Street that may be relevant to this pipe fracture. Such evidence is contentious and I deal with it below.

Cast iron mains were identified as being at risk even with minimal ground movements, but no special measures were put in place by LUL to prevent or minimise such expected movement relative to this main. Comprehensive protection measures were employed however in relation to adjacent buildings and to some pipeline at the western end of St Thomas Street. The protection involved the use of compensation grouting delivered by Tubes a Manchette (TAMS). The process of inserting and deploying such tubes itself can have some effect upon the ground into which they are inserted and adjacent areas, by causing displacement and heave.

Professor Clayton in his impressive and agreed report described the original geology of the area. Between Roman times and 1750 when Westminster Bridge was constructed, London Bridge was the only Thames crossing in the Central London area. Developments south of the Thames occurred from early times. This combined with the rise in sea levels estimated at 4.5 metres since Roman times resulted in a significant rise in ground levels in the area. Current ground surface is at about +4.0m to +4.1m above Ordnance datum. In Quaternary times the area south of the Thames was low lying, with much of the area being below water at high tide.

Archaeological evidence has shown that the Roman settlement at North Southwark was initially constructed on the alluvial sand capped gravel islands or ‘eyots’ that lay within the much wider River Thames. River channels existed between these eyots.

The site of the burst lies on the eastern edge of one such eyot. The Museum of London Archaeological Service Report and sketches for an archaeological dig undertaken at 21/27 St Thomas Street indicate a reasonable ground in the form of ‘clear water lain silts’ at a fairly consistent +0.25m AOD.

“A typical sequence in the area assessed by plotting two sections of the British Geological Survey boreholes in the area and the burst point and extracting levels from other nearby boreholes was found as follows:

Main Ground. Ground level to 0m AOD

Alluvium present in pockets, up to two metres thick

Terrace Gravel +1m AOD to +6m AOD

London Clay –6m AOD to –26 AOD”

These materials are described as spatially variable in terms of composition and properties. At the lowest part, in the London clay the basal beds where the tunnelling activity took place, are dominated by silty and sandy layers becoming clay, silty sand at depth. Above this is ‘typically a very stiff, thinly laminated, very closely fissured, dark grey and grey-brown clay of very low to medium compressibility’.

The Terrace Gravels are described as medium to dense with frequent zones of finer grain material such as clay and silty sand and even occasional organic deposits. The description for the East Vent Tunnel connecting the East Vent Shaft with the main station concourse area and which is nearest to the burst point describes the occurrence of ‘occasional soft casts of clay and occasional large pockets up to one metre thick of thick coarse sand in the gravels’.

The alluvium is highly variable in composition and can consist of very soft-to-soft organic clay, very soft peat or fine to medium sand. Alluvial clays and organic materials are identified in the borehole records in the area.

At para. 4.1 of his report Prof. Clayton describes the made up ground:

The construction records for the 36” main given in the Metropolitan Water Board Record Plan, No.6 of 1925 show the Main to have passed through an old cellar near the point of burst and to have encountered an old well. The walls of the cellar were recorded in the Transport Research Laboratory (TRL) Report dated 2001.

The parish map of St Thomas, Southwark of 1792 shows plans for the realignment of a number of streets in the area including St Thomas Street. A comparison with a more modern Ordnance Survey Map TQ 3380SW shows a new alignment of St Thomas Street over the bases of previous housing in the area of the water pipe failure.

There is also considerable evidence of Roman and Mediaeval activity in the vicinity of the failure recorded in various publications of the Museum of London Archaeological Service and various other publications. Archaeological reports and sketches for the dig undertaken at 21 to 27 St Thomas Street indicate stake holes and deep pits cut to levels of between – 0.25m and – 0.75m OD.

The development beyond the original area of the eyots required channelling of the Thames and construction over the soft ground and so much of the original alluvium was excavated and placed elsewhere as made ground.

“The presence of significant thicknesses of clay Made Ground have been found close to the area of the failure.”

A post war bomb damage map of the area indicates that 21 to 27 St Thomas Street were destroyed along with the West wing of Guys Hospital and buildings to the east of Great Maze Pond and Joiner Street. The East Wing of Guys Hospital was rebuilt and the redevelopment of 21-27 St Thomas Street took place in 1989. The original basement walls were left in situ following internal propping after re-excavation.

Professor Clayton says that: local ground water conditions were affected by a number of major activities associated with the LUL works:

“….These included the construction of a 4.2 metre thick mass concrete underpinning raft beneath the British Telecommunication Building in London Bridge Street; the drilling of holes to allow the installation of tube-a-manchette pipes from the Long Subway Tunnel that ran beneath London Bridge Street; excavation of a compensational grouting shaft; conditioning grouting, prior to tunnel excavation; compensation grouting, before, during and after tunnel excavation; back-filling of the CSLR pedestrian access tunnel; diaphragm walling for the East Vent Shaft; sinking of caissons for the lower part of the East Vent Shaft; excavation of a single phase shotcrete-lined East Vent adit; excavation of the East bound and West bound platform pilot tunnels and the concourse tunnel using sprayed concrete liming; subsequent enlargement of the platform of concourse tunnels to their four sections; installation and grouting as spheroidal graphite iron (SGI) bolted tunnel lining segments and the grouting de-watering and pumping of water from sumps during the construction of the works.”

The changes in local ground conditions induced by the tunnelling and associated works not only give rise to the type of settlement envisaged in LUL’s contractors ‘prediction’ studies, they also give rise to significant changes in ground conditions inducing both vertical settlement and horizontal movement. These changes arise as a result of consolidation over a period of time, as for example where moisture in the pores of homogenous material such as clay moves. Where the material is not homogenous, such as in the London Clay, the position is more complicated. Where there are lenses of other material or voids, horizontal movements may be associated with the vertical settlement and thus give rise to differential settlement.

It is evident from the evidence of Professor Clayton which I accept, that the effects of consolidation following the significant tunnelling work undertaken by LUL, gave rise to a more radical degree of settlement, both vertical and differential and over a much wider area than predicted by the studies undertaken by Mott McDonald and CTW.

The ground water conditions in the area of the site, suggest that the gravel layer overlying the relatively impermeable London Clay and adjacent to the river would be a shallow water aquifer fed by the River Thames. However, in reality the gravels are typically only 6 metres thick and contain significant quantities of less permeable material. The site is about 250 metres from the Thames and the lower part of the London Clay is relatively sandy and permeable. I accept that the borehole records do not provide a reliable guide to near surface ground water conditions because the boreholes have been cased in the Made Ground, and in the alluvium and terraced gravel in order to prevent their collapse, and water will have been added to assist boring in granular materials. Where rotary coring has been used, drilling fluid will be added. The view is expressed by Professor Clayton, is that whilst the bore hole records, although likely to have over estimated the depth of water, nonetheless show that the water table in the area is high.

It is difficult to generalise about the state of the ground water in the shallow deposits alongside the Thames, but the conclusion is justified that the gravels alongside the Thames are in hydrologic continuity with it. There is flow through the gravels and connection with the Thames, albeit restricted by the thinness of the gravel layer, the distance from the river and the obstructions presented by foundations, tunnel service runs and backfill. Professor Clayton states that the terrace gravel layer is not of uniform coarse granular composition, it contains significant sand lenses and pockets of clay. The London Clay similarly is not of uniformly low mass permeability since it contains zones with significant fissuring and permeable fabric. The permeability of the London Clay is likely to have been increased by the stress relief imposed by the tunnelling activity.

A study was made by Drivers Jonas in May of 1990 in connection with the projected tunnelling works noting that the planned excavations may release water from the strata and thus cause a draw down in ground water levels. Nonetheless, having identified a potentially important environmental impact, there is no evidence that any measurements were made to assess the draw down of ground water in fact caused by the JLE works.

In relation to settlement LUL issues relied upon the expert evidence of Mr David Harris, director of the Geo-technical Consulting Group who has specialised in settlement issues associated with tunnelling in London by LUL over a period of 13 years. Mr Harris’ impressive curriculum vitae indicates that he was responsible for the implementation on site of the first application of compensation grouting in the United Kingdom protecting the Victory Arch and the Waterloo and City line from settlement related damage associated with a construction of the new escalator providing access to the Bakerloo and Northern lines at Waterloo Station for London Underground. He also has impressive experience derived from his work at Westminster and the stabilisation of ‘Big Ben’ and the nearby buildings affected by the construction of the Jubilee Line under and through the existing station and close to those buildings.

In his report of 1^st February, he recognised that for multiple tunnels in an urban environment, the calculated settlement from volume loss movements does not in fact represent the prediction of the actual movements, which can be expected. He said that a conservative assessment of potential damage can be made by extending the simple empirical method for a single tunnel to a more complex situation. The method is described as conservative since it is the differential settlements and curvatures associated with volume loss movements as opposed to consolidation settlements that are generally the most onerous values. The parameters which determine the magnitude and extent of volume loss settlement are determined on case history data.

He accepted that the ‘zone of influence’ of tunnelling works was conventionally deemed to be the one millimetre settlement contours, determined from the short-term volume loss calculations which are inaccurately referred to as ‘predictions’ for settlement, because the ‘effects of consolidation must be added to those of ground (volume) loss in order to predict the extent and degree of any consolidation settlement’. Uncontroversially, Mr Harris observed that the settlement trough produced by long term, consolidation movement is known to have a wider, shallower form than for volume loss movement, and that published case history data on long-term settlement was far more limited than for volume loss movement, despite the numerous tunnelling projects in London. He accepted that no generally applicable empirical method exists to predict consolidation movements.

At 3.9 of his report he says “complex numerical analysis is the only method of predicting consolidation movements. The difficulty in determining the appropriate parameters, particularly those which govern the flow of water through the ground (which include the permeability of the ground and its spatial variation), limits the value of such calculations.”

Finite element analyses were undertaken to examine the effects of different construction sequences at Waterloo and Westminster Station. Waterloo Station, it appears, has a similar layout to that at London Bridge. The analyses give some indication of the magnitude and extent of the long-term consolidation settlement, which could be expected. According to Mr Harris, the analysis results show that both the magnitude and rate of consolidation settlement are sensitive to small changes in the relative permeability of different layers within the soil stratigraphy. The magnitude and changes in permeability are ‘small’ compared to the precision with which they can be determined by currently available site investigation techniques.

He concludes:- “although the analysis results show that consolidation settlement can be very variable in rate and magnitude, such movement consistently occurs over a much wider area than volume loss settlement. The maximum movement occurs above the tunnels and can be as great as 100 millimetres over a complex station and 50 millimetres over twin running tunnels. Settlement in between 12mm and 35mm are indicated by the analyses at the 1mm volume loss settlement contour with the full magnitude of movement developing over between 3 and 10 years”.

Professor Clayton’s evidence was that whilst deeper deposits vary significantly in terms of their permeability, they are relatively incompressible. On the other hand, soft cohesive near surface materials (alluvium and Made Ground) have high compressibility, whilst granular near surface materials have low compressibility. Because of the spatial variability of the alluvium and Made Ground, differential settlements would therefore be expected to be particularly pronounced where shallow ground was affected by the JLE works and might be expected to be of the same order of magnitude as any induced total settlements.

It is also evident that the Made Ground in the area of the Main with its old basements and soft organic clays incorporate material within it, which is likely to be metastable. Collapse of materials such as old foundations supported on rotting timber piles, or loose soil and lateral movement of voids, can be triggered by quite small events, for example, by changes in groundwater levels or flows, or ground movements from other sources such as tunnelling shaft excavation or the compensation grouting.

Mr Harris accepted that the make-up of the Made Up ground in which the main was laid was in part metastable and accepted that even slow settlement related to differential settlement could trigger a metastable event as well as fast settlement.

The ground in which the pipe was laid was susceptible to both volume settlement and consolidation settlement. Both could give rise to differential settlement in the Made Up ground in which the pipe was laid, by reason of the settlement itself/or by triggering a metastable event, or by a combination of both.

Mr Harris said in evidence but not in his report that a metastable event could be triggered through transmission of a dynamic force through the over burden from traffic above, such as ambulance traffic. He was not able to support this opinion with any calculations or data. Indeed there was no evidence that the road was used regularly by ambulances. It was at best a theoretical possibility that must be considered both as to the likelihood of its occurrence, and any effect it might have had upon the main in terms of causation of the fracture.

The adjacent JLE contract works (105) in my judgment serves to illustrate the potential for localised differential settlement in similar ground conditions to those at London Bridge. There was significant London clay cover at both tunnelling locations. There were Terrace gravels, Alluviums and Made Ground overlying both the tunnels in Contracts 104 and 105. At Millstream Road (105) a small volume loss within the limits of the so-called ‘prediction’ settlement following tunnelling gave rise to a 60mm drop over a wide area following a tunnel drive. The two areas covered by the contracts are immediately adjacent. The movements were reported in a Paper by M. Friedman. The Paper given in 2003 was entitled “Tunnel-induced disturbance of near-surface alluvium, its effect on overlying structures and remedial ground works treatments”. It was a case history study from JLEP London, published by CIRIA SP199.

The events described in that paper were commented upon by both the tunnelling experts, Dr Barry New (Thames Water) and Mr Harris (LUL), and Professor Clayton. There are clearly points of difference between the two locations. It is not suggested that they are exactly similar. It is evident that events at Millstream Road illustrate that the state of knowledge as to the effects of tunnelling on structures in and above the ground was swiftly evolving, project to project. This was further demonstrated in the lessons learnt as to the effects of preventative and compensation grouting derived from Westminster and Waterloo projects reported by Mr Harris.

I am satisfied that what happened at Millstream Road illustrates that conditions existed for potential, though localised ground movement at London Bridge because there were sufficient points of similarity in the geology, ground water and type of Made Up ground.

The ground conditions directly beneath the failed section of the St. Thomas water main were investigated by the excavation of a trial pit by the Transport Research Laboratory (TRL). Their report describes four cross sections taken and details the materials found. At the base of each section soft organic clay (alluvium) was discovered just below the invert of the pipe extending, in one place where it was tested, for a distance of at least 1.2 metres downwards.

Professor Clayton in his report emphasised that this material was highly susceptible to changes in volume as ground water conditions change.

I do not accept Mr Harris’ view that the conditions at Millstream Road were wholly dissimilar and cannot be of any assistance in interpreting the events in St Thomas Street. It is apparent from Mr Harris’ evidence that he did not attach sufficient weight to the effect of change to the ground water regime which could affect the localised ground conditions in St. Thomas’ Street as a result of the tunnelling works and the sinking of the East Ventilation Shaft. It is perhaps because he did not attach much importance to the potential effect of the ground water conditions that no real attempt was made by LUL by way of disclosure to collate such recorded evidence as may be in existence relating to the water extraction or diversion measures to protect the works. Such records would indicate the volume and extent of any changes to the ground water regime in St Thomas’ Street in the vicinity of the burst.

I was satisfied that the state of the ground conditions at the point of burst was such that relatively small volume loss caused by tunnelling, shaft sinking and associated works would have led to relatively significant consolidation settlement. The extent of such settlement depending upon the make-up of the ground could have directly caused settlement to occur differentially. Alternatively, it could have triggered a metastable event, also leading to differential settlement. It could be a combination of both mechanisms.

I do not accept Mr Harris’ views, subscribed to by Mr Lance that consolidation settlement would primarily be a vertical or tabular settlement. It would tend to be so only if the clay or gravel or other material were homogenous. Such a view ignores the significance of ground water changes.

I prefer the evidence of Prof. Clayton and Dr. New in this respect.

Mr Kimmance of LUL wrote an internal document dated 2^nd February 1993 identifying the first 70 metres of the 36-inch main in St. Thomas Street as being ‘at risk of failure due to tunnelling settlement’. He made reference to Mott MacDonald’s report ‘Prediction of Surface Settlement due to tunnelling – RevD’, saying that it classified the St. Thomas Street Road areas in Rankin Risk categories 3 and 4, i.e. as potentially damaging to rigid pipelines… “It is therefore recommended that the first 70 to 80 metres of the existing cast iron pipe is either protected by compensation grouting or replaced with a steel variety in order to reduce the risk of failure and potential flooding of the proposed excavations.”

LUL’s internal engineers clearly appreciated the obvious risk of settlement damage on the conservative basis of volume loss alone.

On 28^th July of 1993 the LUL Utilities Engineer, Mr Boyden, sent an extract of the proposed contract for Contract 104 to Thames Water which contained proposals that the ground movement that could affect Thames Water’s services should be monitored (Clause 1.10.3) and, under clause 1.10.5 that:

“The Contractor shall immediately notify the Engineer should the results of monitoring indicate any of the following:

On 7^th October of 1993 Thames Water wrote to LUL referring to these proposals:

“Turning now to the document supplied with your letter, the clauses therein are clearly a matter for agreement between JLEP and its contractors. Thames Water will hold London Underground Ltd responsible for any damage to or failure of our water mains and apparatus from JLEP project however caused”.

By a letter dated 8^th July of 1994, Costain Taylor Woodrow the Contractor sent a copy of the contour predicted settlement drawing relating to Contract 104 to Thames Water indicating a ‘predicted’ settlement around the point of failure in the region of 0-1mm.

On 12^th July 1994 Thames Water made it clear that they considered all of the mains within the settlement zone to be at risk and commented that London Underground was supposed to be conducting a risk assessment with proposed actions such as re-routing or replacement of mains. This was emphasised in subsequent correspondence, which culminated in discussions about the assurances as to LUL tunnelling works and the risk to Thames Water’s mains expressed by Thames Water’s senior engineer to LUL’s Mr Steel on 2^nd June of 1995.

“I was asked to provide maximum allowable settlements and change of slope to be tolerated by the 30 inch main and other mains. The amount of strain that can be accommodated within the cast iron pipe before failure is microscopic. Any change of slope can only be sustained if that movement is successfully transferred to a joint. Typical joints on cast iron mains are run lead. They are semi-flexible only on account of the yielding of the lead by cold flow under load. Manufacturers tolerances are intended to accommodating variations during laying. Any subsequent rotary movement will, if continued, shear the iron, splitting and rupturing the pipe. Any rotation of the pipe joint is therefore undesirable.”

(my emphasis)

An extensive system of monitoring was set up. Monitoring points included both road pins and building pins. There were some 250,000 settlement readings made during the relevant period covering pre and post work activity.

It was only in the course of these proceedings that certain road pin data relating to readings of pins close to the site of the burst became available and was therefore disclosed. The building settlement data had already been disclosed.

In relation to the buildings adjacent to the point of failure, numbers 19, Mary Sheridan House, and 21/27 St Thomas Street were fitted with pins. It is clear that the movements recorded far exceeded the ‘predictions’ made by CTW. Nonetheless the risk analyses and schemes for mitigation works to protect the utility apparatus were not revised by the Design Contractor Mott MacDonald in the light of these actual settlements.

Structures thought to be at risk were to be protected by compensation grouting. These included the buildings referred to above. Tubes-a-Machette (TAMS) were installed so that grouting could be injected to compensate and reduce the settlement experienced by the building. The TAMS ran beneath the pipeline 22metres west of the failure point and some 14.8 metres below its invert level. The placement of the TAMS was to protect Mary Sheridan House.

There was no attempt by LUL to protect the Main by compensation grouting anywhere along St Thomas Street although it was known to be at high risk. The road pin settlement data recorded for pin 1071 closest to the point of failure exhibited substantial differential settlement at its location, relative to the adjacent road pins and those on the nearest structures. The extant data for pin 1072, the next close road pin to the point of failure also showed differential movement compared with the settlement on the adjacent buildings.

The reliability of the road pin data is in issue. Mr Harris on behalf of LUL sought to discount it completely, arguing that it was anomalous and should not be relied upon since only data that shows clearly discernible trends and which had been checked by further measurements should be relied upon. Caution should be exercised in respect of data that has not. He said that the movement of pin 1071 did not appear to benefit from either source of reassurance and pin 1072 does not show a clear trend prior to the step settlement, although the change in level was confirmed by several further measurements.

Dr New argued that it was important not to confuse good practice in other scientific and engineering work with what is necessary in the forensic investigation. It is not good practice for an engineer to discard any data point unless there is clear evidence as to why it in an error.

The removal of a data point because it does not fit into a pre-supposed pattern known to the reviewer is unacceptable because it is often the presence and recognition of such an anomaly that illuminates the reason for an unusual or otherwise explained event. It was clearly not appreciated then what the effects of deep excavations on the near surface could be. The read pin data from pins. 1071 and 1972 in the period December 1994 to October 1995 shows that the readings of both pins were stable to within the accuracy of the survey measurements, that is plus/minus 2mm. Similarly the settlements of the nearby buildings showed no sign of movement until the LUL works commenced. Thus for a period of at least 11 months immediately proceeding the works, there was no significant settlement or heave in the immediate vicinity of the failure as defined by the nearest pin locations. The fact that the pins were stable shows that they were clearly not affected by traffic or other disturbances.

Mr Harris proffered a number of potential reasons for the anomalous behaviour of the pins. But first he had to accept the possibility that the movement recorded could be real. Stoutly he argued that such differential movement measurement would represent only a localised shallow movement, which could not be generated by deep-seated movements associated with tunnelling.

In the light of the matters disclosed in the Friedman paper, that is an assertion which standing alone would be bold but in the light of the evidence of Professor Clayton and Dr New, in my judgment the assertion is clearly unfounded.

He went on to suggest that a credible alternative source of localised movement could be identified. He said that point 1072 was located close to a shallow trench excavated from the corner of the JLE compound. That is an assertion, not founded on any evidence. Indeed, the memorandum of agreement resulting from the experts’ meetings on 17^th December of 2003 and 22^nd December of 2003, at paragraph 8 states that there were no road openings in the immediate vicinity at the time of the failure. Had there been any relevant road openings which Mr Harris then thought had significance he would have drawn them to the attention of this to his fellow experts.

Dr New in his report gave evidence that he had carried out investigations to discover what road opening activities had taken place in St Thomas Street. He obtained from the London Borough of Southwark details of all road openings in St Thomas Street in the 12 months prior to the failure. These comprised the Street Works Notices required under the New Road and Street Works Act. He searched to see if anything might have happened which might have caused either a direct impact to or crushing of the pipe or some movement to the ground. Although the visual evidence did not support either impact damage or crushing, he considered it prudent to investigate those matters.

The Street Works Record is an exhaustive record of all works. He gave evidence that there was no roadwork activity as could cause the pipe failure and concluded ‘I am fortified in my conclusions by the agreement of the other experts that roadworks did not cause the failure’.

In this respect and many others, Dr New exhibited an exemplary fairness and thoroughness in his investigations. I accept his evidence in this respect and reject the assertion of Mr Harris. It is evident that Mr Harris did not fully appreciate his duties as an expert both to his colleagues and to the court and indicated a willingness to express an opinion that was not founded on any sound evidential basis.

He went on to offer a further reason for impugning the data evidence. He speculated that the data point had been replaced and did not have its point reading adjusted. The absence of a reading in December of 1995 when both of the adjacent points were read, suggested this to him as ‘a possibility’. He drew the inference that the point must have been damaged and deemed unrepresentative or was destroyed or became inaccessible asserting that ‘this’ was the most probable reason since its position at the corner of the work site would render pin 1072 particularly susceptible to damage. He then concluded

His conclusion that 1072 movements are not relevant to the water main is based upon two unwarranted assumptions.

The senior members of the team responsible for collating the pin data, Mr Sharrocks, Mrs McDonnell and Mr Allan were called to give evidence. It is evident that they did not at the time of the reading of the measurement from pin 1071 consider it and then go on to reject it as being unreliable.

Mrs McDonnell had no recollection of any conclusions reached on pin 1071 at the time when the measurements were taken and reviewed. Mrs McDonnell, an impressive and precise witness gave evidence that all readings were included in the documentation. The totality of gross errors were the 12 identified by Mr Harris, that is to say, 12 out of 250,000 readings. Mr Allan confirmed in his evidence that if the reading of 1071 was thought to be wrong at the time, then the pin would have been re-read. Mr Allen regarded levelling accuracies to be minimal and errors kept to a practical minimum and stated so in his reports.

Mr Harris’ reaction to this evidence was to say that since Mr Allan was responsible for surveying he would, of course, say that he kept the inaccuracies to a practical minimum. Mr Harris then further criticised the format of Mr Allan’s ‘close out’ reports on measurement data, saying they were ‘cut and paste’ when in fact they were not. There seemed to be no warrant for diminishing the value of Mr Allan’s reports on either basis.

Finally Mr Harris sought to reject the readings on pin 1072 on the basis that there was a temporary benchmark problem.

Mr Sharrocks in evidence said that he ‘…had no particular concerns about 1072’.

In relation to the alleged benchmark problem Mr Allan was taken to the figures appended to Mr Harris’ expert report at figure 3 which showed the movements recorded in relation to pins 1745, 1748 and 1749 on the façade of the buildings in St. Thomas Street. There is an apparent downward trend from October/November of 1995 and the lowest point is reached on 12^th January 1996. Then there is disclosed an upwards trend to 12^th April 1996. Mr Allan confirmed the record of both the trending down and then the trending upwards.

Mr Harris’ explanation that this pattern of movement could be attributed to a temporary benchmark problem was considered by Mr Allan ‘a most improbable explanation’ for the pattern of behaviour recorded on Mr Harris’ figure 3. I agree with him. Over a period of time the benchmark would have had to trend both upwards and downwards.

Mr Harris showed himself to be resourceful and partisan in urging the Court to reject the pin data evidence. He is clearly a gifted and innovative settlement engineer but in this case he has shown himself as a witness too close to his de facto employer’s cause to be objective.

In my judgment the road pin data is reliable and properly should be taken account of in considering the cause of the Main burst. I am satisfied that Dr New was right to do so.

According to Mr Lance, there are 400,000 kilometres of cast iron piping carrying various utility services in the United Kingdom. The failed pipe in St Thomas Street was 73 years old when the burst occurred. None the less the distinguished experts in this case covering metallurgy and materials, engineering, soil mechanics and geology at the outset in their agreement concluded:-

“The specific problem of analysing the cast iron strains and stresses induced by longitudinal relative rotation of a large diameter pipe with a lead run joint is not addressed in the technical engineering literature. Although the mechanism is well known to cause failure, particularly in brittle materials, it has proved to be a complex problem from a theoretical standpoint, since it is very difficult to allow for uncertainties caused by the geometry of the ‘as built’ joint and the effective stiffness of the seal.

(The specific problem is not dealt with in the literature by any established numerical, close form or analytical methods.)”

Dr New and Mr Lance are at the forefront in their field as to the effects of tunnelling. Dr New has particular interest in the effects upon structures such as utility pipe work. Mr Harris for some years has been closely associated with LUL and the development of the Jubilee Line and the development of compensation grouting to alleviate the effects of tunnelling on structures. Professor Burdekin is a metallurgist of great experience and an authority in relation to fracture mechanics morphology. Mr Nicholas Glover is a Metallurgical consultant with the well-known consulting engineers Sandberg and had a special interest in grey iron and experience in its use in marine engineering and piston behaviour though not of pipework. Their expertise and judgments upon which their opinions are based are very much informed by their general experience and learning and upon published material, one such paper delivered in Canada during the currency of this trial. The Friedman paper on ground movements in relation to Contract 105 was of significance as earlier noted.

Reference was made in the course of the evidence both in the written reports and orally to the important publications by Attewell, Yeates & Selby [1986] entitled “Soil Movement induced by Tunnelling and their effects on Pipelines and Structures” and Attewell P.B. and R.K. Taylor [1984] Ground Movements and their effects on Structures.

These publications by Attewell and others enjoy considerable authority as text books relating to pipework of 12 inches and under. I accept Dr New’s evidence that the observed behaviour of joints in smaller diameter pipes is of limited assistance in considering the performance of a 36 inch cast grey iron main joint with lead infill. While some larger diameter pipes are referred to in relation to particular types of joint, those references are primarily concerned with leakage and the transmission of axial stresses rather than concerned with the potential effects of rotational movement in a 36 inch grey cast iron pipe. I accept the evidence of Dr New in this relation, implicitly confirmed as it is by the expert’s agreement.

Water main bursts occur from time to time and Thames Water have a database recording bursts in large diameter cast iron mains. LUL adduced evidence of these. It was evident that the majority of the bursts recorded were caused by direct impact during excavation and that the investigation of burst causes was secondary to the imperative of repairing the breach by replacement and stemming the effects of damage caused by the burst main. I am satisfied having heard the evidence of Mr Greenwood an employee of TW who was carefully examined by Mr Moxon-Browne QC and whose evidence was tested in cross-examination by Mr Tavener QC that surprisingly there is no reliable body of empirical evidence to be derived from this source as would throw any helpful light upon how and why the main at St Thomas Street burst.

Dr New was retained by Thames Water shortly after the burst to investigate its cause. He is a Civil Engineering expert and has expertise in relation to the impact of tunnelling beneath urban areas, particularly including the effects of ground movements on cast iron pipes, and cast iron lined underground tunnels. His evidence includes the results of his impressive and painstaking investigations. He eliminated a number of possibilities. He was right to do so as his fellow experts, subject to what I have to say below in their expert agreement, agreed with him. The main was not faulty or otherwise not fit for its purpose. There was no corrosion by graphitisation or tuberculation or manufacturing faults, stress concentrations or fatigue. There was no evidence that the main had been over-pressurised and a careful enquiry was made of records of water pressure from the Battersea Pumping Station and locally at Snowsfield. The main was protected against pressure transients by surge vessels, and the records of water pressure showed nothing unusual or excessive with regard to the water pressure in the main.

There were no roadwork openings or roadworks activities, which could have caused the failure. Dr New also ruled out unusual surface loadings as a principal mechanism of failure; in part because of the lack of hinge join joints at fracture site and because there was no evidence of high contact loads in the vicinity of the failure surface.

The failure fragmentation was not consistent with his wide field experience or ‘textbook’ knowledge as to the effect of crushing failure or a failure by highly localised external loading. These conclusions were subjected to unheralded criticism in the evidence of Mr Nicholas Glover and Mr Harris.

These criticisms in my judgment were not based on any sound evidential foundation. The assertions were not developed and supported by calculations or other scientific reasoning. They were not discussed with other experts. If they were not worth canvassing with fellow experts and exposed to investigation by them, it may signal to the court the measure of consideration that has been given to them and the weight that may be placed upon them. I am satisfied that the main was in fact designed to sustain the type and nature of road traffic using such a thoroughfare as St Thomas Street. Mr Lance LUL’s Civil Engineering expert agreed that this was so in his evidence as to the design loadings.

Having in my judgment properly eliminated the other alternative possibilities, Dr New went on to consider ground movements.

The experts agreed that

“A three-stage approach to investigating a potential joint rotation induced failure mechanism is a valid methodology.

Stage I assessment of the most likely movements

Stage 2 assessment of the strains in the pipe line by relating movements to laboratory tests

Stage 3 evaluating the effects of such strains on the pipeline.

The main at St. Thomas Street failed because the applied stress levels must have exceeded the strength of the material. The difference between the applied design stresses and those necessary to cause the failure, must have come from extraneous loading.

Absent the excluded possible causes, there is the credible possibility that the additional extraneous loading came from the ground movements.

Mr Lance in his original report drew attention to the presence of a brick sewer near to the failed pipe. The possibility earlier had been canvassed by LUL that the movement of soil fines into the sewer could have resulted in the loss of support for the pipe. This explanation fell by the way but at para. 11.3 of his report Mr Lance expressed the view that if a localised settlement feature did exist, it had caused no recorded damage to the sewer which CCTV inspections undertaken in 1999 and 2003 assessed to be in good condition. In his opinion a local settlement feature capable of causing joint rotation failure in a run of 12 foot long cast iron pipes would also cause severe and visible damage to the nearby brick/concrete sewer. This was especially the case as the sewer had been fully re-lined internally with concrete prior to commencement of the Jubilee Line works to reverse the flow making it stiffer in the longitudinal direction and less able to accommodate significant ground movements. That no damage had apparently been caused to the sewer, in his opinion, is very strong evidence that no local settlement has taken place.

If the comparison was of like with like this evidence would be very relevant as to whether there was the potent localised settlement relied upon by TW. But in my judgment there are vital differences.

In the context of localised settlement the sewer is founded at a different level and in a different place to the failed iron pipe. Alluvial deposits are known to be present as shown in the TRL report by Bird at the foundation level of the pipe but, perhaps as a result of the complexity of the site, are not shown at that location in the section shown at Appendix F of Mr Lance’s report. The movements might therefore be expected to be different and this is reflected in the difference in measured settlement behaviour between the pin above the pipe (1071) and the pin above the sewer (1072).

Secondly, the sewer comprises the original brick structure which would be expected to be reasonably flexible, and a segmental concrete liner. Such a concrete liner would also be expected to be relatively flexible in transverse section due to rotation at the segment radial interfaces. This would be very similar to the behaviour of an unbolted segmental lining for a tunnel. The sewer is not particularly stiff longitudinally because there is no reinforcement to bear tension either in hogging or sagging. If movements had occurred beneath the sewer there might be some cracking at the segment interfaces.

Dr New and Mr Lance agree that there is in fact no evidence of serious cracking shown in the various sewer surveys using CCTV.

Thirdly I accept the evidence of Dr New that the sewer is a relatively unstressed structure compared to the pipe. Both must bear ground and traffic loadings but, in addition, the pipe must bear substantial tensile hoop strain due to the pressure of the water within, whilst the sewer does not have to bear the same bursting forces as are present in the socket of the pipe.

A comparison of the sewer and the pipe would be a difficult problem and that Mr Lance understandably has not sought to analyse their relative vulnerability.

There was some evidence derived from an interpretation of the recent CCTV survey film that at a location closest to the pipe burst a suspected build up of detritus could be an indication of some discontinuity in the concrete lining segmentation and thus point to settlement.

The feature depicted was indistinct and the movement of the camera trolley could well have other causes. I attach no significance to that evidence.

I am satisfied that the condition of the sewer and its lining affords no assistance in relation to the issue of settlement relative to the failed main.

Stage 1 of the investigation clearly establishes that the ground movements were significant and had the potential to damage.

The best available evidence of differential settlement at the location of failure was obtained from the closest measurements. The data for road pin 10,71 and the building pin 1743 at the nearest measured point on Mary Sheridan House are relevant, together with road pin 10,72 which is the second closest to the failure location and corroborates the anomalous settlement behaviour in the vicinity.

This is my judgment provides the best estimate of pipe joint movement induced by the ground movement in the immediate vicinity of the failure.

It is accepted that this must be a rotation movement for the pipe in the spigot. In order to ascertain the slope of settlement, reference must be made to the settlement of the building pin 1743 and road pin 1071. The differential settlement is 16.7 mm. This separation of the point to the pin was 3.27 metres, which yields an angular rotation of 1:196. In my judgment it is a proper inference that the settlement was localised which produces the result of an angular rotation 1:111, there being a localised depression between pin 1743 and 1072 at pin 1071, the nearest to the point of burst. This is a conservative estimate of rotation because the chance location of the settlement points available will not necessarily represent the worst case of settlement in the area of failure. Furthermore, other measurements in the area indicate that only about half of the settlement contemporary with the failure had occurred at this time. For example, at building pin 1743, the 10.8mm settlement had become 21mm two months prior to the failure in August of 1999 as shown on the settlement contours as at 31^st December 1999.

The rotation calculated from vertical settlement only represents one component of actual rotation, as there is likely to be an associated horizontal component to settlement as well. As I have already observed, this component was not measured. It is referred to in the defendant’s statement of case on causation ‘the joint at the undamaged end of the failed pipe revealed a gap of a maximum width of 2mm at the side of the pipe rather than on the top or bottom of it’. This would seem to assume longitudinal rotation albeit in the horizontal plane. Mr Moxon-Browne QC describes the relation of one pipe to the other giving rise to rotation in the socket as a ‘kinking’. It is a helpful description.

Between the end of October 1995 and the end of January 1996, road pin 1072 had settled by up to 14 mm whilst the adjacent building with the building pins 1743, 1745, 1748 and 1749, settled by just a few millimetres. Such settlements should they have continued up to the time of failure, would give rise to relatively large rotation.

The laboratory tests were carried out to determine the strain caused in the pipe by relative rotation at the joint only. These were carried out by the Transport Research Laboratory (TRL). A pipe of a similar age and material characteristics complete with undisturbed joints was used. The tests could not include the strains of the substantial operational loads such as the water pressure, the traffic and any locked in strains over the history of the pipe, because it would be impossible in laboratory conditions to replicate these. The loadings put upon the joint were significant. The joint did not fail under laboratory conditions. The data yielded however was relevant to enable the contribution to the failure strain caused by the factor of ground movement induced rotation to be determined.

This regime of testing was set up by Dr New. LUL were kept apprised of the testing and Mr Lance was given the opportunity to witness and comment on the tests both before and after they were carried out.

At paragraph 9.7 of his report he said:

“The TRL test also enabled the relationship between joint rotation and induced socket strains to be quantified. The Stage II monitoring recorded the horizontal and vertical joint movements at different increments of applied load. This relationship enables the joint rotations to be linked directly to the recorded strains….”.

In paragraphs 9.1, 9.2, 9.3 and 9.4 he described the testing regime uncritically and then at 9.5 onwards through to 9.10, he analysed the results and expressed his conclusions upon them. Mr Lance served a further report in January and another report in February. At the trial he expressed a caveat

“The caveat is the speed in which the test was undertaken”.

He went on to say:-

“…the fact that the load was applied rapidly, in the sense that creep properties of the lead in the joint, had to be acknowledged, that was an issue that had to be understood in interpreting the test”.

He was further asked in cross-examination

“Q. Mr Lance…my question is: it did not even occur to you, as a scientist, that such a qualification was a legitimate qualification at the time that you wrote your report?

A. There are many factors that have to be taken into account in undertaking an engineering study. You cannot list all of them in your conclusions.

Judge Wilcox: What about the vital ones?

A. The vital ones are also attributable to the test. The test was not loaded in the way the pipe was in the ground so that had to be taken into account.

Mr Taverner: Mr Lance, you still have not answered my question. My question is this: I was suggesting to you that when you wrote this report, disqualification of this caveat was not even in your mind, it had not occurred to you. It is either a Yes or No.

A.. It had occurred to me, Yes.

Q. Had it occurred to you at the time that you wrote your supplemental report?

A., Of course

Q. But no mention here?

A .No.

Q .No mention of it being an issue between the parties or a concern between the parties in the joint statement?

A. It certainly was not discussed in the joint statement.

If the two matters subject to the caveat, namely the short time scale during which the test was conducted and the properties of lead insofar as they affected the testing, it seems extraordinary that an expert mindful of his duty to the court should not have expressed this caveat in his written report and to his expert colleagues in like discipline if they were in fact matters of significance.

I have come to the conclusion, having heard Mr Lance giving his evidence, that he was not wholly objective in approaching his task as an expert. I was confirmed in this when in his supplemental report he reproduced the conclusions in section C5 of the PDL Report

Mr Lance was the principal prosecution expert witness at a criminal trial brought by the Health and Safety executive. He drew attention to this by referring to it in his curriculum vitae supporting his expertise. He permitted a member of the investigation team to draft substantial passages in 81 pages out of his 122 page report and in the appendices. When this became known in the course of the trial the Prosecution elected not to rely upon his evidence. He was asked about this by Mr Taverner. He made no mention that his expert report had not been relied upon because its independence was compromised in the view of the Court having been substantially contributed to by the prosecution investigators. He said somewhat economically that it was because he was ‘too close’ to the Heathrow Express Investigation Team and that they had been passing too much information between each other. One can readily understand his reticence about this experience but the fact that he was an expert witness at that trial was something that he thought proper to include in his C.V. and rely upon.

Mr Lance is clearly an honest man of considerable experience and technical knowledge and his evidence is clearly worthy of weight, subject to the caveat that I have expressed as to his objectivity as an expert witness in this case. Furthermore his evidence was of assistance when the evidence of Dr. New came to be tested.

Given the relatively limited scope of the sophisticated TRL test, I am satisfied that Mr Lance was right when he unqualifiedly asserted that

“The TRL test enabled the relationship between joint rotation and induced socket strain to be quantified”.

Mr Lance’s evidence quantification of loading I also accept as being broadly accurate. He considered the TRL tests in the same way that Dr New did and factored in the data about the stresses and strains in the lead and how they translated to the intrados of the socket. He assessed these as 40 microstrains. He expressed this as the upper bound figure and commented that Dr New’s quantification of 325 miscrostrains was an upper bound.

The essential difference between Dr New and Mr Lance is explained because Mr Lance discounted the reading from pin 1071 as anomalous, whereas Dr New in my judgment properly took account of it

When the TRL tests were carried out, the complete pin data had not then been disclosed. No one was therefore able to take a view as to the actual field rotation that may have occurred. It is to be noted further that Professor Burdekin on day 5 accepted that on the introdos of the spigot joint of the failed pipe there were no witness marks. He said he would not expect to find them and they would not be there until the lead in the jointing had been completely squeezed out as it was in the final stages of the TRL test which was a test carried way beyond any condition that anyone postulates in relation to the failed pipe. Professor Burdekin also said in evidence that whilst the lead was still present he wouldn’t expect to see a witness mark, because the lead is softer than the iron and does tend to limit the loads, that being the whole point of the flexible joints giving them some degree of flexibility. However, it does not mean they could transfer no load. There is the transfer of significant loads. This gives significant stresses at the spigot joint caused by a combination of the total loads, the previous history loads, the water pressure, whatever earth pressure there is above and some sheering and rotation at the joint.

The fragmentation pattern shown by the failed pipe is relied upon by the Claimants as indicative of the causal mechanism. Professor Burdekin and Dr New were of the opinion that the fracture pattern displayed by the burst pipe was indicative of prising/shearing. Dr New demonstrated to the court using three lengths of drainage pipe how such shearing could occur and in the course of examination in chief, used a homely demonstration with dried pasta tubes to explain the relationship between the “downwards kinking” and the “top popping off” as yielding a particular fracture pattern.

The metallurgy experts have agreed:

“…from the pattern of fractures as shown for example in the sketch produced by Dr T.J. Baker immediately following the failure .. that the prime point of initiation of the fractures must be associated with the fractures around the large single piece ejected from the pipe marked X”.

Thames Water’s case is that the bursting was caused by forces acting outward from within the socket by prising and/or shearing loads. LUL submit that the explanation is that some form of rapidly applied localised load was applied from the outside.

The fragmentation patterns of failed pipes constructed of materials such as cast iron and clay are usefully described in the Young & Trott work Buried Rigid Pipes (1984) and the experts adopt the characterisation of fracture pattern identified in that textbook in their analyses. First (a) the longitudinal cracking caused by overload. Second (b) the circumferential crack due to beam failure. Thirdly (c) socket failure due to excessive vertical shear. Fourthly (d) the burst socket due to excessive radial forces, and finally fifthly (e) the bearing fracture due to concentration of reaction (or sometimes of load).

Dr New referring to Young & Trott concluded that

“… the failure takes the form indicated by (c) or (d) in viii or a combination of both. That is, bursting by excessive radial forces and/or fracture by excessive shear across the joint. This is entirely consistent with failure due to differential ground movements as deduced by a process of elimination described in paragraph 2.05. I formed the opinion that the prying/levering forces caused by the longitudinal rotation of the spigot within the socket (due to differential ground movement), possibly with some shearing, may have given rise to substantial strain within the socket. This mechanism is illustrated in figure 10 of my report. These strains were over and above those caused by the normal operation and use of the main and at this stage of the investigation, I consider that they could have been the cause of failure.”

Again, on the pattern of the fracture, Professor Burdekin’s conclusion in his Expert report states:

“45.

The pattern of fractures in the failed cast iron water main at St. Thomas Street is typical of the one in which there has been a high contact force is between spigot and socket with the spigot tending to ‘lever’ a piece out of the socket.

It is concluded that the most likely explanation for the failure is a combination of loadings, causing concentrated forces at the spigot to socket joint. Since normal pressure and self weight loadings give only very low stresses, and over pressure and vehicle weights have been excluded, the most likely cause to the additional stress is to cause the failure is from ground movement”.

The pattern of fracture fragmentation in terms of Young & Trott is not consistent with the fragmentation caused by a localised external load, either at the top or at the bottom of the pipe.

Mr Nicholas Glover in his evidence said that a major prising force would produce an isosceles triangle and relied upon patterns depicted in un-named textbooks rather than upon his own limited experience of cast iron water pipes. I rejected his evidence in this respect. His research and thoroughness was shown to be wanting and in contrast to the approach followed by Professor Burdekin and Dr New. I am mindful that Mr Glover, for some months, has not been well. I have taken account of this. The defendants have known of this and chosen to rely upon him rather than any other expert evidence. I observe that none of the reports presented by him conform to the requirements of the CPR or protocols. No reason has been given as to why they did not. He is a Consultant who is employed by Sandburgs who have extensive forensic experience. I have made all due allowances to take account of Mr Glover’s apparent condition, perhaps at the risk of being less critical of his approach, but his tested evidence was not impressive.

Mr Guy Lance at 5.2 of his expert report said:

“The shape and location of the failed top section of the pipe indicated to me that a sheared type failure of the pipe had taken place. This is because the shape of the failure surface, which extended between the socket access and the barrel crown at an upward angle. This type of failure can develop either from the application of short-term external vertical load to the barrel of the pipe, or from in-plane bending forces in the socket, such as those arising from joint rotation (prising effect) or from an external load acting on a hard spot at the support (hammer and anvil effect).”

Whilst he conceded that a fracture pattern may be consistent with joint rotation and the prising effect, he was clearly inclined to the view that one would expect to see evidence of local damage to the inside of the socket at the failed end of the pipe as he would expect to find arising from joint rotation. It is clear that Mr Lance gave weight to the absence of a witness mark on the intrados of the spigot joint of the failed pipe and the evidence of Mr Glover. I am satisfied on the evidence of Professor Burdekin, into whose discipline this matter appropriately falls that the joint rotational prising effect would not produce such a mark.

In addition to the fracture pattern, the experts considered the morphology of the fractures, that is to say the detailed technical consideration of the broken surfaces of the pipe which may give indications as to the sites and likely causes of fracture initiation. These were considered by both Professor Burdekin and Mr Nicholas Glover.

They both agreed in the Experts Points of Agreement:

“It is agreed that the fracture surfaces do not show any classical chevron pattern, typical of brittle fractures in some materials. It is therefore not possible to identify fracture initiation positions unambiguously and with certainty from fracture surfaces.”

Mr Glover in the penultimate (unnumbered) paragraph on page 2 of his report of 24^th June of 2003 says:

“The chevron marking pattern on the fracture face at the suggested initiation site was indicative of a very localised, point loading close to the top of the pipe…”

Professor Burdekin confirmed in his report that there were no clearly distinguishing features on the fracture surface that could be identified unambiguously as origins, other than where the pipe had been deliberately broken for removal after the failure by a use of impact hammers. There was evidence of some ‘features’ on the fracture surface, where particular patterns are present at a number of positions but it was not evident to him that these were fracture origins. Professor Burdekin’s particular speciality includes this field. He accepted in cross-examination that because of the amount of graphite in grey cast iron, the metal does not show how the cracks are propagated as they move through the material because it does not reveal chevron markings.

Mr Glover, in his oral evidence, produced photographs of the fracture surfaces which he said depicted signs of the fracture initiation site. The photographs had not been put to either Professor Burdekin or to Dr New. The photographic evidence was not impressive. Mr Glover was constrained in cross-examination to accept that iron does not show any classic chevron pattern.

When the main was removed and broken up, a jackhammer was used. Mr Glover in his second expert report illustrated the fracture morphology of the jackhammer indent. Referring to the photograph produced on the final day of his oral examination, there was no suggestion that any similar feature was depicted in that photograph. I reject Mr Glover’s evidence that there was evidence of localised point loading close to the top of the pipe. I reject his conclusion that the fracture morphology could only imply that the additional loading had caused the pipe to fail which was abnormally large and applied rapidly and locally. I accept the evidence of Professor Burdekin and Dr New that there was cogent evidence showing that the pipe failed as a result of loading from within and around the top of the pipe.

It is not for LUL to make out a positive case. It is for Thames Water to prove its case. LUL were of course placed in a measure of difficulty by Mr Glover who referred to point-loading in his written statements and when asked about it by Mr Taverner in cross-examination explained that what Mr Taverner considered ‘a point’ was not what he considered ‘a point’ and that the expression ‘point loading’ was ‘additional interpretation’ and a ‘loose use of language’. It was not evident what event or phenomenon Mr Glover was referring to.

In the evidence of Mr Harris, the effects of extraordinary impact loads were canvassed and also referred to by Mr Moxon-Brown QC in his opening. No evidence was adduced about these matters. They were properly canvassed in cross-examination as possibilities when the evidence of Professor Burdekin and Dr New was tested. I am satisfied that the possibilities thus canvassed have been demonstrated to be without substance.

A number of samples were taken from the socket and barrel areas of the failed pipe to establish and compare the tensile properties of the grey iron. These were compared with samples taken from the pipe used in the TRL testing. The failed pipe had marginally thicker walls than the TRL test pipe. Generally, the failed pipe was shown to be of good quality and stronger than the TRL pipe, although the testing followed by Sandberg’s did not fully comply with the size requirements of BE EN 10002-1 as represented in Mr Glover’s report.

One sample from the failed pipe MK064T fractured in the shank where a casting defect was discernible. This was a wormhole of a serpentine elongated shape with a rounded profile. The presence of the wormhole had not been evident when the fracture surfaces were the subject of careful scrutiny by a number of people, including Dr Baker. This, in my judgment, is because in all probability it was not on the surface.

The sample did not fail before any load was applied. A failure stress of 133.6MPa was certified by Sandbergs. However, because this failure was in the waisted section, held by the grip an adjustment was made to reflect this. Because of the effect of the proximity to the grip and the stresses arising from this, an adjusted actual strength of 85MPa is indicated. The sample represented a very small part of the socket as a whole, and may I am satisfied have been as much as 290mm away from the fracture site.

In my judgment there is no evidence that the presence of the wormhole in fact contributed to the failure of the pipe. The overall behaviour of the wormhole depends on a combination of factors: its shape and size and the stress field and material in which it lay.

Mr Glover commented in his third report that it was apparent that the defect was large in size and yet was not associated with the fracture in any way. Professor Burdekin did not accept that the wormhole was particularly large, its diameter at its head being about 10mm. He observed that the rounded wormhole defect would have had a negligible effect on the full cross sectional area of the socket, but would have significant effect upon the cross sectional area of a small scale laboratory test specimen.

Despite the non-compliance with the testing standard, the Sandberg tests were of some value. In the tensile test results, the failed pipe test gave results in the range 111 to 143 MPa for the socket whilst the TRL pipe gave results in the range 97 to 118 MPa for the socket and 108 to 110 MPa for the spigot. These are results that tend to confirm the strength values for the socket expressed by Professor Burdekin. I accept the evidence of Professor Burdekin in relation to the strength of the failed pipe, based on both flexural and tensile testing, namely that it was in the range 1000 to 1500 micro strain.

Mr Lance clearly adopted the view of Mr Glover, which I have rejected, namely that the fracture morphology could only imply that the additional loading which could cause the pipe to fail was abnormally large and applied rapidly and locally.

However, he went on to refer to the other factors, which in his view showed that the pipe failure resulted from the application of a dynamic force unrelated to joint rotation. He expressed his belief that the Claimant in proposing that the failure resulted from excessive joint rotations had relied on unsafe settlement data and inappropriate use of material design factors.

“The claimant’s own test clearly demonstrated the ability of another section of the same pipeline to support the deduced joint rotations. The test piece also demonstrated an ability to support strain at levels above the supposed maximum design levels. In addition the test showed a distinctive different failure mode from the actual pipe. My analysis shows minor joint rotations arising from the recorded settlements and moderate levels of applied strain in the joint at failure. Taking into account the excellent condition of the pipe material, I can only conclude that failure resulted from the application of a dynamic forced unrelated to joint rotation”.

In evaluating the significance of the TRL test, Mr Lance failed to give due weight to the locked in stresses of the pipe and neglected to take into account the road pin data.

Mr Lance and Dr New closely agreed the strains due to the operational loading by virtue of the dynamic traffic loading soil water pressure weight and the like. Dr New’s figure is 292 micro strain and Mr Lance 353 micro strain. To the operational strains should be added the relative rotation from the TRL test, giving a slope of 1:111. This gives 325 micro strain. It is inevitable that given the long history of the pipe in the ground that other residual strains caused by the preceding events will have become locked into the pipe. This is an important factor, which I accept is difficult to quantify. Indeed the finite element analysis commissioned by TW had to be abandoned because of the inherent complexity of the task. Since the pipe was laid there have been many factors that could have applied strain to the pipe and the residual strain in a pipe prior to the commencement of the JLE works is unknown.

There is a further unknown giving rise to potential additional strain namely the effects of “stress raisers”. The gauge measurements for the TRL tests were taken at accessible and convenient sites, but where there is any form of geometrical discontinuity in a solid under stress, such as a notch, re-entrant corner, shoulder, screw thread, hole, scratch, sharp curvature in a beam and the like, this may raise the stress of the solid. The measuring gauges could not be placed at these sites.

The shape of the socket is geometrically complex. The shape of the casting exhibits features of a curved beam, and also has a major entrant corner or shoulder at the base of the spigot entry. The strains caused by the spigot rotating in the socket will cause a complex transfer of loads in and from the inside of the socket into the barrel of the pipe. Stress concentrations are therefore to be expected.

The maximum test pressure strain is 338 micro strains and without the unknown of the locked in strains and stress raiser effect, the operational and conservative joint rotation estimated strains give rise to 291 + 325 micro strains. I accept the evidence of Dr New that the strain due to joint rotation alone was likely to have been similar to that derived from the sum of all the normal operating loads on the pipe. The actual strains in the pipe were therefore at least doubled. Even if the effects of locked in strain and stress concentrations were to be discounted, the additional 325 + micro strain represents a very significant proportion of the strain in the pipe at failure, even taking a less optimistic view of the strength of the iron. There are no other significant strains supported by evidence and calculation proposed by either the claimant or the defendants.

Dr New convincingly demonstrates, based upon a range of authoritative sources as to the highest allowable design strain, that a value of 500 micro strain should not be exceeded. To reconcile this upper bound design value with likely strength of the iron, very conservative design strengths for all cast iron structures have been developed to take into account the material variability inherent in structural castings and the brittle nature of the material itself. If one does not rely on what may appear to be conservative design criteria, the iron will fail. This view is supported in the defendant’s statement of case on causation at paragraph 23.

I am satisfied that strains induced by relative rotation caused by ground movement led to substantial strains in the failed socket which were of such magnitude that when added to those from the normal operating and imposed loads resulted in the strain in the cast iron exceeding that which was permissible and led to the failure by way of a prising/shearing fragmentation.

I am satisfied that the cause of failure proposed in the evidence of Mr Lance, Mr Glover and Mr Harris, for the reasons that I have given earlier, cannot be supported on the evidence.

I am also persuaded by the evidence of Dr New contained in his supplemental statement and confirmed in oral evidence that the explanation of the high localised rapidly applied load causing the fracture cannot be supported.

Dr New produced graphic impressive evidence as to the effects brought about where a pipe fails due to an established localised point of loading such as proposed by LUL’s experts. One is a photograph of a pipe of a brittle material (not cast iron) which shows a failure pattern depicted in figure 7.3 in Young & Trott. The fragmentation pattern is wholly different to the failed main in this case. The second photograph is particularly relevant. It is a photograph of a recent failure of a 36-inch cast iron water main, which occurred at Wallace Road, London, NW1 on 24^th November 2003. The failure occurred during drilling, in a metal fenced off area in the road by a mini JCB. The fragmentation of the failed pipe at St. Thomas Street bears no resemblance to the fragmentation pattern caused by localised loading.

It is accepted by all experts that the road was not being excavated, so any possible direct hit on the pipe by dynamic localised load must have been transmitted through the bituminous layers of pavement at sub base, some 1.4 metres down to the crown of the pipe.

There is clear and undisputed evidence of the witness mark left by a jackhammer when the failed pipe was in fact broken out after the failure. A vivid description of the effect of large rapidly applied and localised force. No such mark or indentation of any kind was left on the outside of the fail pipe and in the vicinity of the fractured surface.

The possibility that the mode of failure was caused by an abnormally large, rapidly applied, localised loading in my judgment has been positively excluded.

I am satisfied that the deep tunnel excavations, shaft sinking and compensation grouting caused differential settlement in the shallow made-up ground surrounding the pipe. This was in part the result of the type of predicted settlement, which was in fact greater than anticipated. It was also in consequence of the induced consolidation effects of the tunnelling and the other works in particular, the east vent shaft which affected the ground water regime.

These works triggered the localised movements in the Alluvium proved to be present beneath the failed section of the pipe so as to affect the shallow foundations of the pipe. Because of the natural geology erosion processes and man made disturbance the localised movements both complex and variable had horizontal and vertical components.

The probability exists that the effects of the differential settlement caused by London Underground works led also to some of the shallow supporting ground becoming metastable and that this proximate event causing final failure was a relatively minor triggering event.

The state of affairs brought about by London Underground was the prime and effective cause for the differential settlement, which led to the imposition of loadings to the joint, causing shearing/prising and thus bursting.

I am satisfied on the evidence that had the settlement caused by LUL works not have occurred, then the state of the pipe was such that it would have continued in use for many years to come.

I reject the suggestion that the ground movement caused by LUL was so slight that it could be characterised as ‘the straw that broke the camel’s back’, by adding a nominal loading to a pipe grossly loaded as a result of historic locked in strains and stresses.

There will be judgment for Thames Water on the issue of liability and causation.