In 2006 the Claimant, MT Højgaard a/s (“MTH”), as contractor, entered into an agreement with the Defendants, E.ON Climate and Renewables UK Robin Rigg East Ltd and E.ON Climate and Renewables UK Robin Rigg West Ltd (“E.ON”), as the employer, for the design, fabrication and installation of the foundations for 60 wind turbine generators (“the turbines”) for the Robin Rigg offshore wind farm in the Solway Firth.

Each turbine foundation consisted of a monopile and a transition piece. The monopile is a cylindrical steel pile driven into the seabed. It has a diameter of just over 4 m. The transition piece is also a steel cylinder. It has a slightly larger diameter and fits over the top of the monopile. It is about 8 m long and weighs about 120 tonnes. The tower which supports the generator fits onto the transition piece.

The transition piece is held in place by a grouted connection. The connection is 6.45 m long. During construction the transition piece is lowered over the monopile and is then held in place by temporary supports whilst grout is pumped into the annular space between the monopile and the transition piece. The grouted connection works primarily by friction and not, as one might think, by adhesion between the steel and the grout. It is a well established method of connecting tubular supports for offshore structures. It is possible to improve the vertical (or axial) load bearing capacity of the grouted connection by the introduction of shear keys. A shear key is simply a horizontal (or near horizontal) protrusion on the interior face of the steel which increases sliding resistance. They were not used at Robin Rigg.

In 2004 an organisation called Det Norske Veritas (“DNV”), an independent classification and certification agency, produced an international standard - DNV-OS-J101 (“J101”) - for the design of offshore wind turbines and grouted connections in particular. Unfortunately, the value attributed to a variable used in one of the equations in the standard was underestimated by a factor of about 10. MTH’s designer, Rambøll Danmark A/S (“Rambøll”), like everyone else at the time, was unaware of this error when it carried out the design.

The design was carried out between mid-2006 and mid-2007, after which fabrication and installation followed during the course of the next two years. In 2009 it was discovered that movement was taking place in grouted connections that had been designed using the standard. It was then that the error in the equation in the standard came to light.

The foundations of the turbines constructed by MTH at Robin Rigg have also shown signs of movement and they all require remedial work. The cost of this work has been agreed in the sum of €26.25 million, liability for which depends on the court’s decision in this action.

E.ON holds MTH liable for the failure of the connections. It puts its case in two ways. First, it says that by the terms of the agreement MTH warranted that the wind turbine foundations would have a service life of 20 years. Second, it says that if MTH’s only obligation was to use reasonable care and skill in applying J101, then it failed to do so in at least three respects.

In effect, therefore, the real issue in this case is which of the parties must bear the responsibility for the error in J101.

The case was very well prepared and it was conspicuously well argued on both sides. Mr. David Streatfeild-James QC and Mr. Mark Chennells, instructed by Fenwick Elliott, appeared for MTH, and Mr. John Marrin QC and Mr. Paul Buckingham, instructed by Wragge & Co, appeared for E.ON.

The use of grouted connections to join tubular steel structural members is not new. The technique has been used in offshore structures since the 1970s. However, offshore wind turbines are a relatively recent development and the first offshore wind farm using grouted connections was built in about 2002. It was known as Horns Rev 1 and was off the coast of Denmark.

Prior to its construction tests were carried out by the University of Aalborg on model grouted connections at a scale of about 1:8. These were done with and without shear keys. The tests were intended to demonstrate the static capacity of the connection under extreme load conditions and its fatigue performance. The static extreme load tests showed in all cases that the strength of the connection was sufficient for the anticipated loads. Some problems were encountered during the fatigue tests as a result of cracking in the connecting flanges of the steel test pieces, but there was no sign of fatigue damage to the grout and so it was concluded that the grout had sufficient fatigue capacity also. The grouted connection used for the test pieces followed the design guidelines issued by DNV at the time. In addition to the physical tests, two independent finite element analyses of the grouted connections were carried out. These showed that the stress level in the grout was very moderate: in effect, they confirmed the results of the tests.

It is important to note that the Aalborg tests were concerned to simulate the bending load imposed, principally, by wind forces. They did not apply a load along the axis of the pile and so did not test the capacity of the connection to carry the weight of the structure above it. It was, I think, taken for granted at the time, as a result of experience with grouted connections in other offshore structures where the principal load was axial, that a suitably designed grouted connection had the capacity to carry the required axial load: the concern was whether or not the grouted connection could withstand the extreme bending load to which a wind turbine was subject and the fatigue imposed by repeated bending loads.

On 29 March 2004, at the Global Windpower conference, a paper was presented by Petersen and Andersen, both of DNV, on the structural design of grouted connections in offshore steel monopile foundations. The work giving rise to the paper consisted of finite element analyses and was carried out in connection with the development of J101. The paper stated that:

“The governing load from the wind turbine is the overturning moment and it can fundamentally be transferred as a force-couple in the top and the bottom of the grouted region. Hence, unlike traditional grouted pile/sleeve connections used offshore (mainly axial loaded) the present application does not need shear keys to transfer loads.”

The authors noted that the typical grouted connection for monolithic foundations for a state-of-the-art offshore wind turbine had a diameter of 4-5 m, a steel thickness of the tubular sections of 50-75 mm and the thickness of the grout annulus of the order of 100 mm. The importance of this paper was that it was carrying out the analyses on larger diameter structures than had been used in the past and, of course, much larger than the test pieces used in the Horns Rev research.

So long as the bending load did not cause any damage to the grout, it was assumed that the force-couple referred to in the Andersen/Petersen paper would, if anything, enhance the axial strength of the grouted connection. Effectively, the horizontal load applied to opposite sides and ends of the transition piece as the structure bent under wind load increased the friction between the grout and the steel in the grouted connection. However, many designers in the industry regarded it as conservative to approach the design of grouted connections by ignoring any enhancement produced by the force-couple from the bending loads.

J101 was published in June 2004. In the general section of its introduction it said this:

“A100 General

101

This offshore standard provides principles, technical requirements and guidance for design, construction and in-service inspection of offshore wind turbine structures.

…

103

The standard shall be used for design of support structures and foundations for offshore wind turbines. The standard shall also be used for design of support structures and foundations or other structures in an offshore wind farm, such as transformer stations and meteorological masts. The standard can be used as a stand-alone document.”

The standard was not concerned with the components of the turbine itself, such as the nacelle, rotor, generator and gearbox.

MTH was invited to tender for the foundation works for Robin Rigg on 24 May 2006. It was a condition of the Instructions to Tenderers that the tender was to be completed in accordance with the Employer’s Requirements or it would not be considered (although there was scope for minor departures and qualifications). Tenders were to be submitted by 19 July 2006. There was then to be a period of further negotiation.

MTH submitted its tender on 19 July 2006. The tender included a document known as “the Rambøll memo”, which was dated 6 July 2006. In relation to the grouted connections, the Rambøll memo said this:

“M.10.3: Grouting

… No shear keys are included as the connection easily can transfer the axial force through friction between steel and grout.

...

M.10b.5: Grouting

1.

The grouted joint at the substation foundation is similar to the one at the wind turbine foundation. Shear keys are not included as the axial capacity of the connection can be sustained by friction between the pile and transition piece.”

Tender Questions were issued by E.ON on 18 August 2006 prior to reducing the shortlist of tenderers down to two. The Tender Questions did not refer to MTH’s proposal to omit shear keys, although there was a question about the omission of grout seal testing (Question No 15). On 20 September 2006 MTH was notified that it was one of the two shortlisted tenderers. There was a final meeting on 26 September 2006, following which MTH was selected as the contractor. The design work began in October 2006 under the authority of a letter dated 13 October 2006, which authorised MTH to carry out two weeks of design work. By a further letter dated 31 October 2006, a further three weeks of design work was authorised. Finally, by a letter dated 20 November 2006 a further two weeks of design work was authorised.

The contract was concluded on 20 December 2006. The Form of Agreement was a two-page document signed by the parties which recited E.ON’s wish to build 60 wind turbines and MTH’s agreement to perform the works in accordance with the Agreement for the Contract Price. That was €101,454,052. The main body of the Agreement was in nine parts (Parts C-L - there was no Part F), together with the Annexes to the Employer’s Requirements and Volumes 2A, 2B and 3 of MTH’s Tender Return. Part D contained the Conditions of Contract (“the Conditions”).

Clause 5.3 of the Conditions set out the order of precedence of the contractual documents, which was as follows:

By Part C of the Contract (List of Definitions):

“‘Fit for Purpose’ means fitness for the purpose in accordance with, and as may properly be inferred from, the Employer’s Requirements.

‘Good Industry Practice’ means in relation to any particular undertaking or task … those standards, practices, methods and procedures … to be performed with the exercise of skill, diligence, prudence and foresight that can ordinarily be expected from a fully skilled contractor who is engaged in a similar type of undertaking or task in similar circumstances consistent with recognised international standards.”

The Conditions included the following clauses:

“8.1

GENERAL OBLIGATIONS

The Contractor shall, in accordance with this Agreement, design, manufacture, test, deliver and install and complete the Works:

(i)

with due care and diligence expected of appropriately qualified and experienced designers, engineers and constructors (as the case may be).

…

(iv)

in a professional manner … in accordance with … Good Industry Practice

…

(viii)

so that the Works, when completed, comply with the requirements of the Agreement.

…

(x)

so that each item of Plant and the Works as a whole shall be fit for its purpose as determined in accordance with the Specification using Good Industry Practice.

…

(xv)

so that the design of the Works and the Works when completed by the Contractor shall be wholly in accordance with this Agreement and shall satisfy any performance specifications or requirements of the Employer as set out in this Agreement, and

….

30.2

MAKING GOOD DEFECTS

The Contractor shall be responsible for making good any defect in or damage to any part of the Works including spares held in the Employers inventory which may appear or occur before or during the Defects Liability Period and which arises from, any of the following:

a)

any defective materials, workmanship or design, or

…

…

d)

Works not being Fit For Purpose providing that the Employer has substantially complied with the requirements of the Operations and Maintenance manuals and ‘Defect’ shall be construed accordingly.

The Contractor shall forthwith make good the defect or damage including replacing all defective inventory and at his own cost in the minimum time practicable.”

Part I of the Employer’s Requirements, in Section 1, contained the General Description of Works and Scope of Supply. This contained the following provisions:

“1.6

Key Functional Requirements

The Wind Farms are to be designed, constructed and operated to provide the lowest lifetime cost option capable of meeting the full requirements of this Specification. … All main works items shall be of a design that has demonstrated successful service elsewhere, preferably in an offshore environment, and the Contractor shall complete Schedule 1.3, which provides details of such service.

…

The Works elements shall be designed for a minimum site specific ‘design life’ of twenty (20) years without major retrofits or refurbishments; all elements shall be designed to operate safely and reliably in the environmental conditions that exist on the site for at least this lifetime.

…

The Contractor shall identify in Schedule 1.4 any components which the contractor considers cannot be designed for a life of twenty (20) years and any components or systems which are not designed for this minimum operating life.”

In relation to this last provision, MTH did respond by identifying the coating system as a component whose life could not be designed for 20 years: it said that the expected lifetime of the coating system before major repair works was 10-15 years.

Part I of the Employer’s Requirements, in Section 3, contained the Technical Requirements for the Design Basis (Wind Turbine Foundations). Section 3.1, Introduction, contained the following provisions:

“It is stressed that the requirements contained in this section and the environmental conditions given are the MINIMUM requirements of the Employer to be taken into account in the design. It shall be the responsibility of the Contractor to identify any areas where the Works need to be designed to any additional or more rigorous requirements or parameters.”

and

“The Contractor shall undertake the detailed engineering design of all structures, fixtures and fittings required to comply with the requirements of this Agreement and the Conditions of Contract. The Contractor shall assume full responsibility for design and installation of the structures.”

Section 3.2.2, General Design Conditions, contained the following provisions:

“3.2.2.2 Detailed Design Stage

The detailed design of the foundation structures shall be according to the method of design by direct simulation of the combined load effect of simultaneous load processes (ref: DNV-OS-J101). Such a method is referred to throughout this document as an ‘integrated analysis’

The design of the foundations shall ensure a lifetime of 20 years in every aspect without planned replacement. …

3.2.3.2 Document Hierarchy

Subject to current legislation, the Contractors design shall be in accordance with international and national rules, circulars, EU directives executive orders and standards applying to the Site. Unless otherwise specified in the Contract, the following hierarchy of standards shall apply (1 highest and 8 lowest):..

1.

IEC 61400-3 (if formally published). See note below regarding draft standards.

2.

DNV-OS-J101:2004.

…

7.

Other standards….

Where conflict arises between standards, the standard with the highest priority as indicated above shall take precedence.

…

Where aspects of the foundation design are not covered explicitly by standards, or the Contractor’s design departs from standard (i.e. the implementation of recent research papers to effect economies) such aspects and departures shall be specifically stated as part of the tender documentation or the detailed design documentation, as appropriate ….”

In fact, IEC 61400-3 remained in draft form at all relevant times, so DNV-OS-J101:2004 (“J101”) was the primary standard.

Section 3.2.6, Life, contained the following provision:

“All parts of the Works, except wear parts and consumables shall be designed for a minimum service life of 20 years.”

Section 3b, Design Basis for Offshore Substations & Meteorological Mast, contained similar provisions about the life of the structures. These were as follows:

“3b.5.1 Scope

The design of the structures addressed by this Design Basis shall ensure a lifetime of 20 years in every aspect without planned replacement. The choice of structure, materials, corrosion protection system operation and inspection programme shall be made accordingly.

…

3b.5.6 Service Life

All parts of the Works, except wear parts and consumables shall be designed for a minimum service life of 20 years.”

Section 10 of Part I of the Contract, Technical Requirements, provided:

“10.2.5

Structural Design

… The design shall also address, but not be limited to such specific aspects not covered by standards as:

…

The use or otherwise of shear keys in grouted connections ….

All primary structures shall be designed to require no scheduled inspection or maintenance in the operational life of the wind farms …

10.5.1

Grouted Connection

… The connection shall be designed to require no scheduled inspection or maintenance.

…

The Contractor shall determine whether to employ shear keys within the grouted connection. If shear keys are used, the design and detailing shall take due account of their presence for both strength and fatigue design to the satisfaction of the Certifying Authority and the Engineer. If shear keys are to be omitted then the Contractor shall demonstrate with test data, that the grouted connection is capable of transmitting axial loads at the grout/steel interface without dependence upon flexural (normal) contact pressures, which may not always be present, to the satisfaction of the Certifying Authority and the Engineer. Such demonstration shall also account for joint performance under different temperature conditions ….”

As one might expect, J101 is not a simple document. It runs to about 140 pages and is entitled “Design of Offshore Wind Turbine Structures”. DNV is described as:

“… an autonomous and independent foundation with the objectives of safeguarding life, property and the environment, at sea and onshore. DNV undertakes classification, certification, and other verification and consultancy services relating to quality of ships, offshore units and installations, and onshore industries worldwide, and carries out research in relation to these functions.”

Part A of Section 1 contained the following by way of introduction:

“A100 General

101

This offshore standard provides principles, technical requirements and guidance for design, construction and in-service inspection of offshore wind turbine structures.

…

103

The standard shall be used for design of support structures and foundations for offshore wind turbines.

…

106

The standard has been written for general world-wide application. National and governmental regulations may include requirements in excess of the provisions given by this standard depending on the size, type, location and intended service of the wind turbine structure.

A200 Objectives

201

The standard specifies general principles and guidelines for the structural design of offshore wind turbine structures.

202

The objectives of this standard are to:

- provide an internationally acceptable level of safety by defining minimum requirements for structures and structural components (in combination with referenced standards, recommended practices, guidelines, etc.)

- Serve as a contractual reference document between suppliers and purchasers related to design, construction, installation and in-service inspection

- serve as a guideline for designers, suppliers, purchasers and regulators

- specify procedures and requirements for offshore structures subject to DNV certification

- serve as a basis for verification of offshore wind turbine structures for which it DNV is contracted to perform the verification.”

In Part D, Definitions, the standard defines the verbs “shall”, “should” and “may” in the following terms:

“101

Shall: Indicates a mandatory requirement to be followed for fulfilment or compliance with the present standard. Deviations are not permitted unless formally and rigorously justified, and accepted by all relevant contracting parties.

102

Should: Indicates a recommendation that a certain course of action is preferred or is particularly suitable. Alternative courses of action are allowable under the standard where agreed between contracting parties, but shall be justified and documented.

103

May: Indicates a permission, or an option, which is permitted as part of conformance with the standard.”

Section 2, Design Principles, included the following provisions:

“A100 General

101

This section describes design principles and design methods for structural design, including:

- design by partial safety factor method with linear combination of loads or load effects

- design by partial safety factor method with direct simulation of combined load effect of simultaneous load processes

- design assisted by testing

- probability-based design.

…

104

As an alternative or as a supplement to analytical methods, determination of load effects or resistance may in some cases be based either on testing or on observation of structural performance of models or full-scale structures.

…

C200 Target safety

201

The target safety level for structural designs to the normal and low safety classes according to this standard is a nominal annual probability of failure in the range of 10^-5-10^-4 with 10^-5 being the aim for designs to the normal safety class and 10^-4 being the aim for designs to the low safety class. This range of target safety levels is the range aimed at for structures, whose failures are ductile with no reserve capacity.”

Part H, which was headed Probability-based Design, contained the following provision:

“H100 Definition

101

The structural reliability, or the structural safety, is defined as the probability that failure will not occur or that a specified failure criterion will not be met within a specified period of time.”

Section 9 of the standard was headed “Design and Construction of Grouted Connections”. This contained the following provisions:

“A200 Design principles

201

Design rules for grouted connections are given for axial loading combined with torque and for bending moment combined with shear loading, respectively.

Guidance note:

Long experience with connections subjected to axial load in combination with torque exists, and parametric formulas have been established for design of connections subjected to this type of loading. For connections subjected to bending moment and shear force, no parametric design formulas have yet been established. Therefore, detailed investigations must be carried out for such connections.

202

For design of grouted connections, it may be conservative to assume that axial load and bending moment do not interact. When it can be demonstrated for a grouted connection that it will be conservative to assume that axial load and bending moment do not interact, the grouted connection shall satisfy two separate requirements. The first requirement to satisfy is the capacity requirement specified for the combined action of axial load and torque under the assumption of no concurrently acting bending moment and shear force. The second requirement to satisfy is the capacity requirement specified for the combined action of bending moment and shear force under the assumption of no concurrently acting axial force and torque.

…

204

A grouted connection can be established with or without shear keys as shown in Fig. 1.

Guidance note:

Shear keys can reduce the fatigue strength of the tubular members and of the grout due to the stress concentrations around the keys. If shear keys are used in a grouted connection subjected to bending, they should be placed at the mid level of the connection in order to minimise the influence on the fatigue damage, because the maximum grout stresses from bending will develop at the top and bottom of the grout member.”

Part B of Section 9 was headed “Ultimate Limit States”, and provided as follows:

“B100 General

101

The ultimate capacity of axially loaded grouted tubular connections may be calculated according to the method given in DNV Rules for Fixed Offshore Installations, January 1998. The method is reproduced in the guidance note with torque included.”

The guidance note to this part then set out a number of parametric equations, one of which gave the interface shear strength due to friction. The equation for this showed that the interface shear strength was directly proportional to a variable, δ, where δ was the height of surface irregularities. However, the guidance note stated that, for rolled steel surfaces, this was to be taken 0.00037 R_P, where R_P was the pile outer radius. The value for δ turned out to be too low, by a factor of about 10: unfortunately, this did not emerge until about September 2009 - after the discovery of the first failures of turbine foundations that had been constructed in accordance with J101.

The note went on to state that the equations had been proven valid within the following limits, including:

Part C, “Fatigue Limit States”, included the following provision:

“101

… The design lifetime shall be based on the specified service life of the structure. If a service life is not specified, 20 years should be used.”

Part D, “Requirements to Verification and Material Factors”, provided that:

“D100 Experimental verification

101

If no sufficient documentation of the behaviour of a grouted connection is available, experimental verification of the behaviour must be carried out.”

This section of this judgment is taken largely from MTH’s opening because I did not understand the material to be controversial.

On 28 September 2009 DNV provided MTH with a letter which identified for the first time fundamental problems with J101. The same (or a similar) letter was sent to other operators or stakeholders in the offshore wind industry, including E.ON. The letter made the following points:

The DNV standard clearly needed to be revised to address this. As a result, it was decided that a Joint Industry Project (“JIP”) would be established.

MTH immediately informed E.ON of this, making clear that until discussions had taken place with DNV it was not possible to predict the consequences for Robin Rigg.

A document was issued dated 11 October 2009, “Capacity of Grouted Connections in Wind Turbine Structures”, which was subtitled “Proposal for Joint Industry Project on Grouted Connections”. This stated that its objective was to improve the basis for a reliable design methodology for grouted connections in large diameter structures, before setting out in a little detail the Scope of Work proposed for the JIP. This was to involve the following:

Members of the industry were invited to participate in this exercise on payment of a fee which was proposed to be €100,000. The first meeting was held on 28 October 2009, following which the initial proposal was revised.

What became known as “JIP1” started in November 2009 in accordance with the objectives that had already been identified. The intention was that it would conclude its work by May 2011.

On 10 April 2010 the JIP produced a Project Status Report covering the period from November 2009 to 21 March 2010. The summary stated that, based on the tests performed so far, “… traditional cylindrical grouted connections can not be recommended due to low long term capacity”. As the report made clear, certain parts of the scope of the project were being altered or extended in response to work which had already been performed.

A report summarising the JIP Background for Recommendations on Design of Grouted Connections (the “Background Report”) was originally produced on 21 May 2010, but was the subject of a revision dated 31 January 2011.

The JIP Summary Report on the “Capacity of Grouted Connections in Offshore Wind Turbine Structures” (No 2010-1053) (the “Summary Report”) was originally produced on 11 July 2010. This went through five revisions, the last being dated 12 May 2011.

As part of this suite of reports, the JIP also prepared Report No 2010-0651, “Testing of Grouted Connections Subjected to Static Axial Load and Dynamic Sliding”. This described the experiments that had been conducted as part of the project.

Mr. Streatfeild-James’s submissions described the findings of the Summary Report in these terms:

“(1)

As a result of the JIP work, the understanding of how grouted connections without shear keys behaved had improved. As a result, it was understood that axial capacity could be explained in terms of surface irregularity. This meant both surface roughness (as a result of steel production and corrosion) and tolerances (both circumferentially and longitudinally).

(2)

Initially, capacity depended on surface roughness and tolerances, giving (in effect) a bond (as shown in Fig 3 stage I at C16/26/13). As the bond capacity was exceeded, a slip occurred.

(3)

After a slip had occurred, the capacity depended on tolerances and radial stiffness. It was likely that immediately after installation there was significant axial capacity because of such tolerances, the likelihood being that the circumferential welds provided significant irregularities.

(4)

However, there was no specification of minimum tolerances, and so no basis upon which one could document a reliable capacity.

(5)

As a bending moment was applied, it was likely that the grouted connection experienced ovalisation. This had the effect of breaking the initial bond around the circumference.

(6)

The deformations of the joint led to relative sliding between the steel and the grout. While these individual movements may be small, the accumulated effect may be large.

(7)

The accumulated effect of many small movements is that wear occurred to the grout surface and capacity was reduced over time.

(8)

The laboratory tests showed that the long term resistance to sliding was so much reduced that a ‘reliable lower bound on resistance to be used in design can hardly be provided. This means that design solutions should be sought that can be documented to show sufficient capacity with a resistance between steel and grout surface equal to zero for cylindrical shaped grouted connections.’

(9)

‘Due to the low long-term axial capacity of cylindrical shaped grouted connections, it is believed that the design of such connections in monopiles can no longer be recommended’. The 3 reasons given were the problems caused to interface shear capacity by increasing diameter, the low long term effective resistance and the lack of requirements for minimum tolerances.”

As Mr. Streatfeild-James put it, these conclusions revealed what was self-evidently a dramatic failure in a code promulgated for use internationally in offshore work.

In the Background Report the JIP explained what had happened in the following terms (again, I am indebted to Mr. Streatfeild-James and Mr. Chennells for the following summary of what was said in this report):

“(a)

In producing the original version of the equation in J101 which described interface shear capacity, Sele and Kjeoy had in a paper in 1989 expressed /R_p as a constant (0.25*10^-3). Data from the Department of Energy gave rise to a different constant (0.34*10^-3), but with the difference being explained by the difference in surface finish of the tubes used in testing. Although the JIP could not precisely trace the evolution of these figures to that in J101 (0.37*10^-3), it was speculated that this may have been due to further tests, or to consideration of changed grout strengths. However, this difference was not regarded as being significant in the light of what the JIP had determined about  more generally.

(b)

The JIP’s work led it to decide that  ought in fact to be a constant, and so not vary with R_p, and the tests carried out during the JIP suggested it should have a value of between about 0.066 and 0.077. A plot of the observed settlements reported in 2010 and the test data is shown at Figure 10, the red function representing a constant parameter ( = 0.070mm) (C16/11/25). The black line represents the J101 variable parameter.”

Section 4 of the Background Report addressed the consequences of the relative sliding between steel and grout, and said that the wear of the contact surfaces was “a failure mode that [had] not been properly accounted for in design standards”. The rest of the Section addressed the JIP work done to assess this, but it was recorded that after 20,000 cycles the resistance against sliding was reduced to less than 20% of the initial bond capacity.

So, as MTH put it, J101 was fundamentally flawed. Following it would inevitably result in a design which significantly overstated the axial capacity of the grouted connection.

However, the work of the JIP did not stop there because it went on to carry out further work in a second Joint Industry Report (“JIP2”). The proposal for JIP2 set out the basis on which this work was to take place in rather more detail, and included the following:

“1 Summary and Motivation

Recent experience and new test data of plain grouted connections shows that the static axial capacity for large diameter connections is significantly lower than that expressed through design practice until 2009 …

It is realised that this new learning will have an impact on the design practice for grouted connections with shear keys that are subjected to alternating dynamic loading. (By alternating dynamic loading is understood a similar load amplitude in positive axial direction as in the negative direction …)

By a significant reduction in the capacity of large diameter connections without shear keys, this will also thus lead to significant reduction in capacity of grouted connections with shear keys subjected to alternating dynamic loading

…

It is realised that there are insufficient test data for grouted connections with shear keys subjected to alternating dynamic loading. This makes design of such connections difficult as the safety level is unknown and recommendations can hardly be given in design standards by any responsible organisation.

...

3

Objective

The objective of the proposed project is to provide a sound data basis and to establish a reliable design methodology for grouted connections with shear keys in:

● Large diameter monopile structures subjected to a static axial load and alternative dynamic bending moment where grouted connections are used to transfer the forces from the transition piece to the monopile.”

So far as the mode of failure was concerned there was no change to the JIP’s conclusion that the application of the bending moment had the effect of distorting the grout and thereby breaking the initial bond between the grout and the steel. This led to relative sliding between the steel and the grout with consequent wear of the grout. As this wear increased over time so the transition piece slowly began to move downwards relative to the monopile.

Following DNV’s letter of 28 September 2009 information began to circulate amongst operators that settlements were being discovered in other offshore wind farms. By the spring of 2010 it became clear to E.ON that there was likely to be a settlement problem at Robin Rigg. E.ON attended a two-day workshop on the problem on 21/22 May 2010. Meanwhile E.ON and MTH were in dispute about the responsibility for the remedial work. E.ON’s engineer alleged that MTH was in breach of “overriding fitness for purpose obligations” and MTH responded by saying that any fitness for purpose obligation was qualified by its duty to comply with J101.

Subsequently a more constructive approach developed and the parties, together with input from Rambøll, embarked upon the resolution of a remedial scheme. However, this process was very drawn out and it was not until 25 April 2013 that the parties entered into a compromise agreement by which E.ON agreed to carry out an agreed programme of remedial work designed by Atkins - subject to liability for the payment for the work being determined by these proceedings. I therefore turn to the first issue in the case.

The point of disagreement between the parties is whether the clauses that I have set out above, in particular clause 8.1 of the Conditions when read with clauses 3.2.2.2 and 3b.5.1 of the Technical Requirements, impose a strict obligation to achieve a service life of 20 years or merely an obligation to design the foundations on the basis of a 20 year design life in accordance with J101.

MTH’s submission was that its essential obligation was to exercise Good Industry Practice to produce a design compliant with J101: it did not take the risk that J101 might contain a fundamental error. In his elegant submissions, Mr. Streatfeild-James submitted that the central approach of J101 was to design with safety in mind. The objective was to keep the probability of failure to an acceptably low level: in this case, an annual probability of failure of 10^-5.

Mr. Streatfeild-James submitted that this was reinforced by the stochastic approach taken in the requirements relating to the characteristic load effect to be adopted. For example, the choice of a load or load effect with a return period of 50 years. (Footnote: ¹)

There are references in J101 to the adoption of a service life of 20 years if a service life is not specified by the employer. So it is clear that J101 can accommodate a service life greater than 20 years if that is what is required by the employer. It is therefore not surprising that the Technical Requirements refer to a design life.

Mr. Streatfeild-James submitted that the approach underlying J101 is inconsistent with the imposition of a warranty as to the minimum life of the structure. As he put it in his note of closing submissions:

“The effect would be that the 2 isolated references to ‘ensure’ in general sections of the Specification would supersede all the more detailed and specific references to ‘design life’. That cannot be right. There is, it is submitted, no way of reading the two provisions together - one must trump the other.”

Mr. Marrin submitted that Part I made it clear that MTH was to assume “full responsibility for design” (see Part I, Section 3.1, quoted at paragraph 25 above) and that the “purpose” of the works referred to in clause 8.1(x) is to be determined from the Technical Requirements (it is common ground that the reference to the “Specification” in the clause is a reference to the Employer’s Requirements containing the Technical Requirements). He submitted also that the reference to Good Industry Practice in clause 8.1(x) relates to the determination of the purpose and does not qualify the obligation as to fitness for purpose. Mr. Streatfeild-James disagreed: he submitted that it was the “fitness for purpose” that was to be determined using Good Industry Practice.

On this point, as a matter of construction, I prefer the submissions of Mr. Marrin. However, I cannot see that it makes much difference because if the relevant purpose is determined in accordance with the Technical Requirements as a 20 year service life, there is no room for argument in this case as to whether or not it has been achieved. Plainly, if the grouted connections of the foundations had to have a service life of 20 years, they have failed to achieve it.

Mr. Marrin referred me to a decision of the Supreme Court of Canada, The Steel Company of Canada Ltd v Willand Management Ltd [1966] SCR 746. The employer engaged the contractor to carry out repairs to three roofs which were damaged in a storm. The roofs in question had originally been constructed by the contractor, who sued for the cost of carrying out the repairs. The tenders for the original work were made and accepted on the basis that the roofing, roof insulation and sheet metal work was to be done in accordance with descriptions and specifications which the employer had forwarded to the contractor when issuing the invitation to tender. These descriptions and specifications contained complete details as to the materials and methods of construction to be employed, which included the requirement that the insulating boards were to be attached to the steel sheeting on the roofs by the use of “Curadex or approved equal”. The damage occurred because the Curadex adhesive failed to perform the function for which it was intended.

The Supreme Court of Canada held that when a contractor expressly undertakes to carry out work which will perform a certain function in conformity with plans and specifications and it turns out that the work so constructed will not perform the function, the express obligation to construct work capable of carrying out the duty in question overrides the obligation to comply with plans and specifications. Accordingly, the contractor’s claim failed in spite of the fact that it had carried out the work in accordance with plans and specifications provided by the employer.

The court approved the following statement of English law in Hudson’s Building and Engineering Contract, 8^th edition, at page 147:

“Sometimes, again, a contractor will expressly undertake to carry out work which will perform a certain duty or function in conformity with plans and specifications, and it turns out that the work constructed in accordance with the plans and specifications will not perform that duty or function. It would appear that generally the express obligation to construct a work capable of carrying out the duty in question overrides the obligation to comply with the plans and specifications, and the contractor will be liable for the failure of the work notwithstanding that it is carried out in accordance with the plans and specification. Nor will he be entitled to extra payment for amending the work so that it will perform the stipulated duty.”

This passage still appears in the current edition, the 12th, at paragraph 3-103. Mr. Marrin referred me also to a fairly recent decision of the Court of Appeal for British Columbia, Greater Vancouver Water District v North American Pipe and Steel Ltd [2012] BCCA 337, in which the earlier decision of the Canadian Supreme Court in The Steel Company case was said to be a complete answer to the appeal. The Claimant agreed to supply water pipes to the water authority. The water authority specified the type of pipe and how it was to be a protectively coated. The pipes proved to be defective. The Supply Agreement contained the following relevant provisions, in clauses 4.4.3 and 4.4.4:

“The Supply Contractor warrants … that the Goods … will conform to all applicable Specifications … and, unless otherwise specified, will be fit for the purpose for which they are to be used ….

The Supply Contractor warrants and guarantees that the Goods are free from all defects arising at any time from faulty design in any part of the Goods.”

The Court of Appeal approved a statement of the law by the judge which was in the following terms:

“The general rule is that defects caused by an owner’s specification are not the responsibility of the contractor, unless the contractor expressly guarantees that the construction would be fit for a specific purpose, or a warranty can be implied by the owner’s actual reliance on the contractor’s skill and judgement.”

In Greater Vancouver Water District, Chiasson J, giving a judgment in which the other two members of the court concurred, said, at [33]:

“Clause 4.4.4 is clear and unambiguous. Reference to authorities that deal with difficulties construing contractual provisions that may contain an implied warranty are of no assistance in this case. North American guaranteed that the pipes would not have defects arising from faulty design. The trial judge held that the pipes did have defects arising from faulty design. In my view, on the plain language of the contract, North American is liable for any damages that resulted from those defects. It does not matter whose design gave rise to the defects. There is no such qualification in clause 4.4.4.”

If, for the purpose of this case, one treats J101 as “an owner’s specification”, then these decisions are authority for the proposition that the existence of an express warranty of fitness for purpose by the contractor can trump the obligation to comply with the specification even though that specification may contain an error.

It is settled law in England and Wales that when construing a contract such as this one the court must have regard to all the terms and must read the document as a whole and in the light of the background knowledge which would have been reasonably available to the parties in the situation in which they were at the time when the contract was made. As Lord Clarke said, in Rainy Sky SA v Kookmin Bank [2011] 1 WLR 2900 at paragraph 14:

“The principles have been discussed in many cases, notably of course, as Lord Neuberger MR. said in Pink Floyd Music Ltd v EMI Records Ltd [2010] EWCA Civ 1429; [2011] 1 WLR 770 at para 17, by Lord Hoffmann in Mannai Investment Co Ltd v Eagle Star Life Assurance Co Ltd [1997] AC 749, passim, in Investors Compensation Scheme Ltd v West Bromwich Building Society [1998] 1 WLR 896, 912F-913G and in Chartbrook Ltd v Persimmon Homes Ltd [2009] 1 AC 1101, paras 21-26. I agree with Lord Neuberger (also at para 17) that those cases show that the ultimate aim of interpreting a provision in a contract, especially a commercial contract, is to determine what the parties meant by the language used, which involves ascertaining what a reasonable person would have understood the parties to have meant. As Lord Hoffmann made clear in the first of the principles he summarised in the Investors Compensation Scheme case at page 912H, the relevant reasonable person is one who has all the background knowledge which would reasonably have been available to the parties in the situation in which they were at the time of the contract.”

And, at paragraph 23:

“Where the parties have used unambiguous language, the court must apply it. This can be seen from the decision of the Court of Appeal in Co-operative Wholesale Society Ltd v. National Westminster Bank plc [1995] 1 EGLR 97. The court was considering the true construction of rent review clauses in a number of different cases. The underlying result which the landlords sought in each case was the same. The court regarded it as a most improbable commercial result. Where the result, though improbable, flowed from the unambiguous language of the clause, the landlords succeeded, whereas where it did not, they failed. The court held that ordinary principles of construction applied to rent review clauses and applied the principles in The Antaios (Antaios Compania Naviera SA v Salen Rederierna AB) [1985] AC 191. After quoting the passage from the speech of Lord Diplock cited above, Hoffmann LJ said, at p 98:

“This robust declaration does not, however, mean that one can rewrite the language which the parties have used in order to make the contract conform to business common sense. But language is a very flexible instrument and, if it is capable of more than one construction, one chooses that which seems most likely to give effect to the commercial purpose of the agreement.’”

It is not uncommon for construction and engineering contracts to contain obligations both to exercise reasonable care, or to do the work in a workmanlike manner, and to achieve a particular result. Indeed, where the contractor has a design obligation, terms as to fitness for purpose of the completed work are sometimes implied: such contracts are likely to include also the lesser obligation to carry out the design with reasonable care and skill. The two obligations are not mutually incompatible. For example, a building can be designed and constructed so that it is robust enough to last for its design life, notwithstanding that the workmanship is untidy or its appearance an eyesore - perhaps because the walls consist of a mixture of different types of bricks (albeit each type of brick suitable in itself). An obligation to carry out the work in a workmanlike manner should prevent such blemishes from arising. At a general level, therefore, I would not accept a submission that the two sets of obligations cannot exist side by side.

But Mr. Streatfeild-James’s submission was more focused than that. He submitted that an obligation to ensure that the foundations will have a service life of 20 years leaves no room for an obligation to carry out the design with a particular design life in mind. In this context it is important to note that clauses 3.2.6 and 3b.5.6 refer to a “minimum” service life of 20 years. If, in order to ensure that the foundation will have a service life of 20 years, it is necessary to take a longer period as the design life for the purpose of J101, this in my view does not create any inconsistency with clauses 3.2.6 and 3b.5.6. If the word “minimum” was not included in these two clauses, then I could see that Mr. Streatfeild-James’s submission would have more force.

From a purely practical point of view I can understand that any employer wishing to have an offshore wind turbine built would want the designer to comply with the latest applicable standard relating to the design or construction of such turbines. Such an employer would thereby hope and expect that the turbine so designed and constructed would not only operate satisfactorily but also would last for its design life. However, that employer could, if he wished, stipulate also for a term that the turbine should achieve a particular service life. But, as Mr. Streatfeild-James pointed out, such a term would usually come at a price: the contractor would have to consider what additional measures he should take in order to reduce the possibility of failure within the service life to an acceptable level and the cost of those measures would be factored into the price. But I consider that if the employer is prepared to pay for the additional comfort of having the service life guaranteed, then that is his choice. However, whether or not that was the subject of a conscious decision in this case we do not know.

On this issue, and in the light of the authorities that I have discussed, I have reached the conclusion that the words of clause 3.2.2.2 are clear and that they are not inconsistent with the other terms of the contract or with the intentions of the parties as reflected by the Agreement as a whole. This is not one of those cases where the court is being asked to choose between two available meanings of the words used. In this case the two competing provisions are either mutually inconsistent or they are not. I consider that they are not and that MTH did assume full responsibility for design as stated in section 3.1. Subject only to the qualification about the life of the coating, in my judgment MTH warranted that the foundations would have a service life of 20 years and E.ON is entitled to rely on that warranty notwithstanding that MTH was required to design the grouted connections in accordance with J101. Since those connections failed within 2-3 years, MTH was in breach of that obligation.

However, if I am wrong about this I must go on and consider the claim in relation to the design. But before doing so I need to say something about the problem with the value of δ that was given in J101 and the evidence that was called by the parties.

DNV first published rules of the design of grouted connections, with and without shear keys, in 1989. These involved the adoption of a value for the coefficient of friction between the grout and the steel, µ, and for the surface roughness or irregularity, δ.

The value that was initially derived for δ, through the expression δ/r (where r is the radius of the pile), was 0.00025. This was based on a best fit with the test data: it was not a measurement of actual surface roughness. The value was later changed to 0.00037, which was the value taken in J101.

As Dr. Billington pointed out in his report, at first sight there is no practical explanation as to why surface roughness should vary in direct proportion to the pile radius. The answer to this conundrum finally emerged during the evidence of Professor Schaumann.

The short explanation is that surface roughness or surface irregularity is not simply a reflection of the smoothness of the surface of the rolled steel plate - in other words, what one would feel when running a finger across the surface. It takes into account also irregularities on a much larger scale, such as lack of circularity of the pile (or transition piece) or deformities in a longitudinal direction. In fact, it seems that these departures from a perfect profile may be more effective at providing a mechanical interlock than irregularities at a micro level on the surface of the steel.

With the benefit of this insight it is possible to see that taking a value for δ that is directly proportional to the radius of the pile is not as counterintuitive as might at first appear. At least, in my judgment, it is not a matter that ought to have put a competent designer on notice that there might be a flaw in a parametric equation that incorporated δ as a variable.

Mr. Carstens is a civil and structural engineer employed by Rambøll who specialises in the design and re-analysis of offshore structures. For the first ten years of his professional career he worked in the oil and gas industry and for the last 10-12 years he has specialised in the offshore wind industry. Before he began work on the Robin Rigg design he had been involved in the design of three or four other offshore wind farms in accordance with J101, either in a draft or the version issued in 2004. One of these projects was known as Kentish Flats, for which Rambøll put in a lot of work in the development of a finite element analysis. This was Mr. Carstens’ first experience of working with J101, which was then in draft.

Mr. Carstens said that having been through the design process more than once, Rambøll knew what DNV would and would not accept in terms of the design. Thus, he said, Rambøll knew how to use the various parameters in the model and how to satisfy the requirements of the code. He said that there was nothing in the design of the foundations for Robin Rigg that was out of the ordinary or which might have called for extended check procedures. He said that the Horns Rev wind farm had been constructed in 2002 just outside the windows of Rambøll’s offices in Denmark and he said that they knew very well what had been done.

When he was asked about the use of shear keys he said this (at Day 3/19):

“... we had calculations that showed that the axial capacity was only utilised around one third of what was necessary, so it would be strange engineering to put something in that could damage your fatigue life of the grout when you only utilised one third of what was actually necessary.”

And, at page 20:

“We had a lot of focus on the bending moment of the grouted connection, because that is what was different from the oil and gas. So we analysed it thoroughly and found out that we needed approximately this 1.5 times the diameter to transfer the bending moment. And we then subsequently checked what the axial load capacity of that was inherent in the design that was necessary for the bending moment, and we came up with utilisation ratios of about one third.”

In relation to shear keys he said also that even in 2009, when the problems surfaced, DNV would not allow the use of shear keys without extensive testing, as they then carried out during the JIP2 investigation. He said that to add shear keys at a later stage in the design process would be a modest fabrication exercise, although it could potentially have an influence on the required wall thickness of the grouted connection.

I thought that Mr. Carstens was an honest witness and I accept his evidence unless otherwise stated.

Mr. Andersen is a structural engineer who is now the head of the offshore wind department at Rambøll’s Esbjerg office. In 2006 he was a member of Mr. Carstens’ team of designers. Mr. Andersen’s evidence was not particularly contentious and he was not cross-examined at any length.

Until March 2012 Mr. Pedersen was a senior project manager at MTH and in that capacity was involved in the Robin Rigg project. He said that the design was essentially left to Rambøll. He said that he did not really question the omission of shear keys, because it had been explained to him how they worked and the risks that were involved in the use of shear keys. He said that if Rambøll had suggested that it was necessary to carry out large scale tests taking some 4 to 5 months, that could have been a big issue and MTH would have looked into it very carefully. Again, Mr. Pedersen’s evidence was not particularly contentious.

Mr. Hoult is a structural engineer who was employed by KBR until about September 2006, when he was engaged by MTH to carry out some work in relation to the preparation of the specifications and technical documents for Robin Rigg. Whilst he was employed by KBR he had been involved in the design of an offshore wind farm at Barrow and before that he had also been involved in a previous tender for the Robin Rigg project in 2002 which eventually came to nothing.

He said that after the 2006 tender he had no further involvement in the project until he was asked to prepare a witness statement in these proceedings. However, it appears from the contemporaneous documents that Mr. Hoult attended a meeting on 13 July 2010 at which the inspection findings of the grouted connections at Robin Rigg were to be discussed. Mr. Hoult said that he had no recollection of attending this meeting, although he did recall having had some discussions with Mr. Swanwick about general considerations in relation to the performance of grouted connections.

In his design for the offshore wind farm at Barrow Mr. Hoult included shear keys. He did this because he had carried out a finite element analysis which he said caused him to question the axial capacity of the grouted connection without shear keys. From his model simulation he concluded that a mechanism was possible by which the transition piece could progressively move down the pile if the interface shear was overcome during each load cycle. Unfortunately, for reasons of confidentiality, Mr. Hoult has not been able to produce the finite element analysis that he carried out or to give any details of it. So, as MTH pointed out in its final submissions, the highest that E.ON’s case can be put on this is that a different designer on a different project about which details are not known adopted shear keys on the basis of an analysis which has not been provided. Whilst what Mr. Hoult did is obviously relevant, I agree with MTH that it is very difficult for the court to reach any conclusion about what Rambøll should have done on the basis of Mr. Hoult’s approach to the Barrow project in the absence of any details of the finite element analysis that led Mr. Hoult to do what he did.

Mr. Swanwick is a mechanical engineer, who was employed by E.ON as a senior project engineer at the start of the Robin Rigg project, although by the end he had become the engineering manager for the project. So far as the decision to omit shear keys was concerned, Mr. Swanwick said that E.ON was satisfied that both Rambøll and MTH had a pretty strong track record in this area. They had responded to E.ON’s specification saying that they intended to omit shear keys and he said that in those circumstances E.ON relied on the expertise of MTH and Rambøll. He agreed that E.ON did not ask for any clarification or explanation of Rambøll’s decision to omit shear keys.

Although Mr. Swanwick was cross-examined at some length, I have not found his evidence to be of much direct relevance to the issues that I have to decide. However, as with all the other witnesses, I thought that he gave his evidence honestly and fairly.

Mr. Hillier is currently employed by E.ON as the project manager for the Robin Rigg project, but he was not involved in it until about May 2011. He was therefore concerned only with the investigations subsequent to the discovery of the failures in 2009. In the circumstances, and without meaning any disrespect to him, I did not derive much assistance from Mr. Hillier’s evidence.

Professor Schaumann is a professor at Leibniz University, where he is the Head of the Institute for Steel Construction and has held a chair since 1996. His PhD was on the numerical simulation of steel members and frames exposed to fire. He has been involved in the wind energy industry since 1987 and he has extensive experience in the design of offshore wind farms. His first piece of research on grouted connections for monopiles for offshore wind turbines was in 2003. This was for a research institute in which he has been involved called ForWind, which is a research centre for wind energy.

Over the past ten years Professor Schaumann has been involved in three research projects in relation to grouted connections for offshore wind farms. Although his mother tongue is German, he speaks and writes impeccable English. Overall I thought that he was an impressive witness.

Dr. Billington has been involved in the design and certification for grouted pile connections for offshore structures for 35 years. His doctorate was on the elastic behaviour of box girder bridges, for which he received an Armstrong Medal. However, his experience in the design of an offshore wind farms dates from 2008. That involved an experimental wind farm that was constructed in 2009/10 with prototype foundations. The first large scale operational wind farm with which Dr. Billington was involved was built in 2010. He said that he did not have any cause look at the JIP1 report before 2011.

Dr. Billington is undoubtedly an eminent engineer, with extensive experience of grouted connections, but he does not have a record of involvement in the design of the foundations for offshore wind farms that is comparable to that of Professor Schaumann. As a general comment, I formed the view that his approach as an expert was somewhat over strict, perhaps as a result of being a little tainted by hindsight. Nevertheless he is undoubtedly a very experienced and well qualified expert, whose views call for respect.

E.ON alleges that MTH, by its designers Rambøll, failed to comply with J101 in three major respects.

First, E.ON alleges that it was not demonstrated that for a grouted connection it was conservative to assume that axial load and bending moment did not interact. This was said to be a breach of paragraph A202 of Section 9 which, for convenience, I set out again below:

“202

For design of grouted connections, it may be conservative to assume that axial load and bending moment do not interact. When it can be demonstrated for a grouted connection that it will be conservative to assume that axial load and bending moment do not interact, the grouted connection shall satisfy two separate requirements. The first requirement to satisfy is the capacity requirement specified for the combined action of axial load and torque under the assumption of no concurrently acting bending moment and shear force. The second requirement to satisfy is the capacity requirement specified for the combined action of bending moment and shear force under the assumption of no concurrently acting axial force and torque.”

Second, E.ON alleges that MTH failed to evaluate the allowable upper limit for R_P/t_P, as required by the guidance note set out above and, in particular, the passage which read:

“The upper limit for the ratio R_P/t_P can be exceeded for low utilization of the axial capacity of the grouted connection. The allowable upper limit for R_P/t_P must be evaluated for the actual connection and the actual utilization.”

(My emphasis)

Third, E.ON alleges that MTH should have carried out experimental verification of the behaviour of the proposed grouted connection, which it contends was required by paragraph D100 of Section 9.

However, in addition E.ON contends that clause 10.5.1 of the contract required MTH to adopt shear keys unless it was demonstrated by test data that these were unnecessary. This allegation is closely connected with E.ON’s first allegation in relation to the design in relation to paragraph A202 of J101, so I shall take those two allegations together.

E.ON submits that the wording of clause 10.5.1 is quite clear and provided that, if shear keys were to be omitted, MTH was required to demonstrate with test data that the grouted connection was capable of transmitting the axial loads without dependence on flexural contact pressures.

Mr. Streatfeild-James submitted that the election not to use shear keys was to be made, and was made, at the time of tender. MTH’s tender proposal stated expressly that shear keys were not to be used. The contract was made on that basis and so MTH undertook to design and construct the grouted connections without using shear keys. E.ON had the choice, if it wished, to require MTH to provide the relevant test data before accepting MTH’s tender proposals and dispensing with the requirement to use shear keys, but it did not do so. Further, Mr. Streatfeild-James submitted that both the certifying authority and the engineer must be taken to have been satisfied that the omission of shear keys was appropriate.

Further, E.ON contends that the same approach is reflected in paragraph A202, which required, submits E.ON, MTH to demonstrate that it would be conservative to assume that the axial load and bending moment did not interact if that assumption formed the basis of the design. E.ON submits that MTH took no steps to meet either of these requirements.

So far as paragraph A202 is concerned, MTH submits that it should not be criticised for using a specified standard in the manner in which the authors of that standard intended or for failing to check the research which underpinned the standard. Mr. Streatfeild-James submitted that it is unrealistic to expect the designer to start from scratch, ignoring previous work and the knowledge and experience of the certifying authority.

In relation to clause 10.5.1, I accept the submissions of Mr. Streatfeild-James. In my view, the short answer to this allegation is that MTH was not under an obligation to provide test data that, on the basis of the design contracted for - namely, one without shear keys, was not necessary. MTH’s tender was submitted on the express basis that its design did not involve the use of shear keys. That was accepted by E.ON. MTH’s general obligation, therefore, absent any express warranties, was to exercise reasonable care and skill in designing the foundations in accordance with J101.

I therefore turn to paragraph A202 of J101. E.ON submits that the conservative nature of the assumption had to be demonstrated by MTH in some positive way, either by carrying out tests or by undertaking a finite element analysis.

It is important to bear in mind that DNV was not only the author of the standard but also the body nominated under the contract as the certifying authority. In this capacity DNV started reviewing the design documentation in about mid November 2006, although its appointment as certifying authority was not finalised formally until April 2007.

In addition, E.ON appointed Offshore Design Engineering (“ODE”) to monitor, supervise and coordinate a wide range of activities including the design and engineering of the structures. From about October 2006 onwards ODE was provided with Rambøll’s Design Briefs for review.

In its Design Brief for the grouted connection Rambøll stated that a finite element analysis model of the grouted connection - that is, the transition piece, grout and pile - would be established and would form the basis for the non-linear analysis of the interaction between the grout and the steel. At paragraph 5.2, the Design Brief went on to say this:

“Generally the ULS analysis of the grouted connection will be performed in accordance to section 9, subsection B of [J101]. Conservatively the grouted connection is analysed for axial loading and bending separately.”

It then went on to explain in some detail how the Ultimate Limit State (“ULS”) analysis of the grouted connection would be carried out. This document was sent to ODE by E.ON on 28 November 2006.

DNV provided comments on Rambøll’s Design Brief on 16 November 2006, but made no observation on Rambøll’s assumption that it was conservative to treat axial loads and bending moments separately. Rambøll’s Design Brief was provided to ODE for comment on 30 October 2006, and on 28 November 2006 ODE confirmed that it had no material comments.

The next stages in the design process were the submission of the Design Basis and, finally, the Design Report. The Design Report for the grouted connection, submitted to DNV on 19 February 2007, repeated the statement that it had been prepared in accordance with J101 and again said that:

“Conservatively the grouted connection is analysed for axial loading and bending separately.”

In June 2007 Rambøll issued a “Design Report – Summary” which included a finite element analysis of the grouted connection. This showed that the maximum stress in the grout (which was at the lower edge of the connection) was well within the allowable stress for the grout (Ducorit S5). All fatigue lives were above the target fatigue life.

In response to the Design Reports DNV issued a series of Design Review Letters. In relation to the grouted connection, by e-mail dated 30 May 2007, DNV confirmed that the grouted connection complied with J101 in respect of both ULS and FLS and, on 2 June 2007, DNV provided the results of its own finite element analysis of the grouted connection in a Technical Report which concluded that the design of both the grout and the steel complied with J101 (Footnote: ²). The DNV finite element analysis showed that there was an area of highly localized over-stress within the lower 70 mm of the grout, but when this area was ignored the maximum stress for the grout was very comparable to that found by Rambøll. In its turn, ODE saw and approved the Design Report for the grouted connection on 14 March 2007.

This review process led to DNV issuing a Design Verification Letter on 22 June 2007 in which it confirmed that, as an independent third party, it had verified the documents and drawings for the Detailed Design of the Primary Foundation Structures, including its own independent finite element analysis. The letter referred specifically to Rambøll’s Design Basis dated 28 November 2006.

Against this background, I turn again to the words of A202. It is to be noted that it is written in the third person and it does not say who it is that must demonstrate that the assumption is conservative or to whom that demonstration is to be made. In my view, on a fair reading of the paragraph it is for the designer to show that the assumption is conservative, but the paragraph is silent as to the means by which this might be done.

It was Professor Schaumann’s evidence that, in his experience, provisions such as this are directed to “the relation between the designer and the approval certification body” (Day 5/81). Accordingly, if the certifying body shares or makes the same assumption, then the designer would have satisfied the requirement of the paragraph.

In this context Professor Schaumann referred to the 2004 paper by Andersen and Petersen and to a paper written in 2007 by Nielsen (Footnote: ³), who concluded in the latter:

“... the possibility of cyclic depletion of the axial load carrying capacity is effectively eliminated by the compressive strength requirements of the grout for it to withstand the governing overturning moment of the ULS, a requirement which ensures that the mechanical frictions will be maintained through the lifetime of the structure.”

Professor Schaumann said that this view was consistent with his own work where he had looked at the effect of fatigue on the grout.

Dr. Billington’s evidence was that a reasonably competent engineer would ask himself the question whether it was conservative to make the assumption. He said that one way of passing through this “gateway” would have been to adopt a design using shear keys. However, it seems to me that that is not to validate the assumption but rather to render it unnecessary.

Dr. Billington was referred to a paper by Lamport et al which was presented to the Offshore Technology Conference in Houston, Texas, in April 1987. This gave the results of tests which had been carried out on 18 grouted pile-sleeve connections to examine the influence of combined axial load and bending moment. The pile and sleeve diameters were 18 inches and 21 inches, respectively - much, much smaller than the piles that were in use in the early and mid-2000s. Dr. Billington considered that these results were of limited relevance, not only because they were much stiffer connections (being much smaller) but also because they employed shear keys. However, Dr. Billington did go on to accept that there was a “handful of people” who thought that axial load and bending moment could be assessed separately and in that there was “a smaller handful of people” who thought otherwise (Day 7/48). These comments must be seen against the background of a fairly small specialist industry in which only a limited number of designers specialised in the design of foundations for offshore wind turbines.

Whilst I am prepared to accept in principle Dr. Billington’s reservations about over reliance on the Lamport paper, his evidence supports the view that there was at least a respectable school of thought in the offshore wind turbine industry which considered, on reasonable grounds, that the axial load and the bending moment could be assessed separately and that this represented a conservative approach.

In the context of A202 the assumption that the axial load and the bending moment could be assessed separately was a reasonable one if there was a respectable school of thought in the industry that considered this to be the case. It was not, therefore, negligent of Rambøll to adopt it. Further, if that school of thought includes the certifying body under the contract, as was the case here, it was not necessary for MTH to do any more in order to demonstrate that the assumption was conservative. In this context I accept the evidence of Professor Schaumann that the paragraph is really directed to the position as between the designer and the certifying body so that it requires the designer to demonstrate to the certifying body that the assumption is conservative. It follows that if that is a position which the certifying body has already adopted, then the designer does not have to do any more. That was the state of affairs here, as is reflected by the fact that both DNV and ODE accepted the assumption without comment or question.

For these reasons I consider that E.ON’s allegations in respect of the breach of the provisions of paragraph A202 are not made out.

There are two points here. The first is what is meant by “low utilization” of the actual capacity of the grouted connection. There was some debate about the actual utilisation ratio taken by Rambøll. As calculated by Rambøll it was 0.33. Dr. Billington calculated it as 0.38, or 0.57 if a coefficient of 0.4 was used (instead of 0.6).

Dr. Billington appeared to rely on DNV’s post-failure position that “low” in this context should be taken to be 0.093. In my view, the position taken by DNV following the discovery of the failures of grouted connections in 2009 provides no reliable guidance as to what is meant by “low utilization” on an ordinary reading of J101 in 2006-07. At that stage DNV was effectively seeking to compensate for the error in its parametric equations.

It must be recalled that in engineering calculations a utilisation ratio of 1 would usually have a significant factor of safety already built-in. That is the case here (as Mr. Swanwick confirmed at Day 4/78). It was Professor Schaumann’s evidence that conventional engineering wisdom is that a utilisation ratio of or below 0.8 is regarded as low (evidence which was supported in part by the evidence of Mr. Swanwick, at Day 4/105, in the context of the design of pressure vessels). I accept this evidence. So whether the actual utilisation ratio in Rambøll’s design was 0.3 or 0.6 does not matter: in either event I find that it was “low” within the meaning of J101.

This leads to the next step, which is an assessment of the allowable upper limit for R_P/t_P. J101 stated that its equations had been proved to be valid where R_P/t_P ≤ 30. In Rambøll’s design the value was 43. Since the utilisation ratio was low, the value of 30 could be exceeded but, as J101 stated, the “… allowable upper limit for R_P/t_P must be evaluated for the actual connection and the actual utilization”. It is far from clear what this means in practice and it was the subject of much evidence and submission at the trial.

It was Dr. Billington’s view that J101 required this evaluation to be by way of large scale ultimate load testing of the actual grouted connection. Professor Schaumann disagreed. He said in his second report that in 2006/2007 it was believed that the behaviour of grouted connections was well known in the light of the Aalborg tests. He said that large scale tests are very expensive and time-consuming and would not ordinarily be a cost associated with a design such as this.

In cross examination Professor Schaumann agreed that what had to be evaluated concerned the axial capacity of the connection being considered by the designer. He then said that when the R_P/t_P value of 30 was exceeded he would evaluate the allowable upper limit by considering whether the load transfer mechanism of the grouted connection would provide sufficient stability against buckling. In other words, to check whether an R_P/t_P value of 43 was allowed or not (Day 5/100-2). Professor Schaumann concluded that with a value of 43, the utilisation of the total buckling resistance was about one third, which he regarded as fairly low. He said that this analytical consideration was based on very safe assumptions and that he would expect a finite element analysis to yield a higher margin (Day 5/105). However, Professor Schaumann accepted that he had not seen any such analysis carried out by Rambøll.

Professor Schaumann did not agree that, in order to assess the actual capacity of the connection, it would be necessary to carry out physical load tests. He said also that he was not able to say whether or not a static load test, if carried out, would have shown that the connection might fail, although he recognised that Mr. Carstens thought otherwise (Day 5/110-112). He made the point in his second report that the lack of axial capacity of the grouted connections was caused by a combination of two factors: (i) the inaccuracy in the equation for axial load capacity within J101 (i.e. the incorrect value of δ) and (ii) the effect of cyclic bending moments on the axial load bearing capacity (which was not accounted for in J101).

In cross examination Professor Schaumann agreed that the error in the value of δ was of the order of a factor of 10. But he pointed out again (at Day 5/117) that the damage was caused by a combination of the error in that value and the consequences of the cyclic loading. He said that without the cyclic loading he would not expect the difference to be of the same order.

Although Rambøll carried out a finite element analysis to test the strength of the grout, it did not carry out any form of evaluation of the actual grouted connection. If there was any evaluation of the allowable upper limit of R_P/t_P, then there is no record of it. To this extent, therefore, I find that Rambøll did not comply with the requirements of the guidance note to paragraph B 101 of J101.

However, I do not accept Dr. Billington’s view that such an evaluation could only have been done - or even should have been done - by large scale static load tests. If J101 required the designer to carry out such an expensive and time-consuming verification process, I would have expected it to say so in clear terms. Since the only alternative method of evaluation put forward was Professor Schaumann’s buckling check (a check that Dr. Billington agreed should have been carried out), I consider that this would have been appropriate if Rambøll had carried it out. But if Rambøll had carried out such a check I am satisfied, on the basis of Professor Schaumann’s evidence, that it would have supported a value of 43 as being within an allowable upper limit for the value of R_P/t_P. Accordingly, Rambøll’s failure to carry out the required evaluation made no difference to the design of the grouted connections and so did not cause any loss to E.ON.

In addition, taking the evidence as a whole I am not persuaded that large scale static load tests would necessarily have revealed the problem. In his first report Dr. Billington explained that appropriate testing would have been of a half length connection at 75% scale. He said that this would have revealed “… the problems with the surface roughness parameter within the design” (page 39, paragraph 17), a view with which Mr. Carstens was also inclined to agree (see Day 3/53).

However, this opinion has to be considered in the light of the view that Dr. Billington held at the time that the value for δ in J101 bore no relation to reality. This was based on his understanding that δ was a measurement of “… the height of surface irregularities” (see paragraph 6.3.1.5 of his first report). But, as I have already explained, δ was not in fact a measurement of anything but was a value derived from a best fit with test data and, in any event, reflected also overall departures from tolerance in the pipe and sleeve as a whole.

Since both experts accept that cyclic movement of the structure played an important part in the failure of the grouted connections at Robin Rigg it is far from clear to me that a static axial load test would have caused a test sample to fail. So even if Rambøll had carried out a static axial load test of the type contended for by Dr. Billington, E.ON has not satisfied me that it would have revealed the potential weakness of the connection or the problem with the value of δ.

I should note that there was nothing to prevent E.ON from carrying out static load tests of the type contended for by Dr. Billington. They would have taken (on Dr. Billington’s evidence) less than six months, and would probably have cost less than 1% of the sum at stake in this litigation.

For convenience, I will set out again paragraph 101 of D100:

“101

If no sufficient documentation of the behaviour of a grouted connection is available, experimental verification of the behaviour must be carried out.”

MTH’s case was that, since verification under the standard was to be carried out by DNV, this paragraph is directed at DNV rather than at the designer. E.ON submitted that it was an obligation upon the designer.

When he was asked about D101 during cross-examination, Mr. Carstens said this (Date 3/19):

“I believe, my Lord, that this is not specific to axial capacity. This is a clause that says you can do experimental verification if you wish to do so, if you think that there is not sufficient documentation of the behaviour of the grouted connection that you have designed.

Now, when DNV writes standards, they cannot envisage what sort of grouted connections designers will come up with. You could do different things, you could use other grout materials. So far it had been a high performance grout that had been used on the various wind farms. One might suggest that you would do a wind farm with ordinary grout, and in that case it might be that DNV would say: well, we don’t have sufficient documentation for the behaviour.

It might also be that you could choose some other steel, or weld in some square brackets, or whatever you could think of in the connection. This is DNV’s way of making sure that they could require additional testing if they thought that it was not documented how the behaviour of the grouted connection is.”

and, at page 92:

“That is not, my Lord, what I’m saying. What I’m saying is that this is a provision for DNV that if they think that there is not sufficient documentation, then they can call for experimental verification.”

In his first report Professor Schaumann said that in 2006 it was believed that there was sufficient knowledge of the behaviour of grouted connections in monopile foundations as a result of the experience gained in previous projects and the experimental investigations that were conducted prior to the Horns Rev 1 project. However, in cross-examination he was constrained to accept that there was really no documentation available to MTH of the performance of other turbine foundations that had been constructed prior to Robin Rigg and that there were no experimental test data, apart from the results of the Aalborg and earlier tests. But if MTH and Mr. Carstens are correct in their belief that the experimental verification, if required, must be called for by DNV, Professor Schaumann’s evidence does not greatly affect the matter: if it is for DNV to decide whether or not to require experimental verification, then the extent to which documentation was available to MTH ceases to be relevant.

In his first report Dr. Billington simply assumed that the obligation under paragraph D101 rested on the designer. He did not say why he considered this to be the case. As I have already mentioned, Dr. Billington has enormous experience in the design of grouted connections for offshore structures (other than wind turbines) but he did not become involved in the offshore wind farm industry until 2008. He was therefore not involved in designing or considering the design of foundations for offshore wind farms in 2006-07.

In the light of this evidence, and considering J101 as a whole, I consider that paragraph D101 is directed to the body concerned with the verification of the design. It is clear that it is referring to the particular grouted connection that is the subject of the design, as opposed to grouted connections generally, and it seems to me that DNV was in a far better position than any particular designer would have been to decide whether or not the behaviour of a particular type of grouted connection was sufficiently documented so as not to require experimental verification. Appendix M to J101 concerns project certification of offshore wind farms and deals in particular with the DNV project certification scheme and provides how it is to be carried out. As Mr. Streatfeild-James submitted, to require the designer to decide whether or not experimental verification was required may involve each designer coming new to a particular type of grouted connection having, in effect, to reinvent the wheel. Why should this have to be done when the verification body may well know that the behaviour of a particular type of grouted connection is well documented?

Even if I am wrong about this, the fact that DNV did not require experimental verification of the behaviour of Rambøll’s grouted connection to be carried out suggests that it had formed the view that there was sufficient documentation of that behaviour. That in my view is sufficient.

Dr. Billington referred to some wear tests carried out on behalf of E.ON at Loughborough University by a PhD student, Mr. James Dallyn. Dr. Billington relied on this work as showing that wear of the grout was not a likely cause of the failure at Robin Rigg. He was cross-examined in detail about this work, some initial results of which were presented to a JIP2 meeting on 25 October 2012 with the aid of some PowerPoint slides. The first piece of this work failed part-way through and no results were produced because of problems in extracting the data.

The subsequent results demonstrated anomalies that Dr. Billington was unable to explain: in particular, at two points during the test the grout sample was measured as getting thicker as the load increased (a possible explanation for this that was put forward elsewhere in the documents was distortion of the test rig). The provisional conclusion from this piece of work was that wear of the grout over 20 years was estimated at 0.71 mm, a figure that Dr. Billington mentioned in his second report. Dr. Billington relied on this work as showing that wear could not have been a significant factor in the failure at Robin Rigg.

Although Dr. Billington was asked twice in cross-examination whether this work could be relied on confidently as indicating the likely amount of wear at Robin Rigg, he was not prepared to say yes. All he was prepared to say was that he thought that the work provided some help, in that it demonstrated that some wear does occur albeit that it was quite small (Day 6/145-7 and 154-5).

At the conclusion of Mr. Streatfeild-James’s cross-examination on this topic I asked Dr. Billington whether or not he would be prepared to advise a commercial client to spend substantial money in developing a proposal for an offshore wind farm on the basis of this work, to which he said that he did not think that he would.

In these circumstances I am not prepared to reach any conclusions on the basis of the work that had been carried out by Mr. Dallyn by the time of the trial. That is not a criticism of him, rather it is a conclusion that the results of the work that had been done by that time, such as they were, are quite insufficient to enable the court to form any reliable conclusions about the cause of the failure at Robin Rigg. However, I must emphasise that I am not expressing any concluded view about the potential reliability of this work if and when it is completed.

In his initial discussions with Inge Lotsburg, the author of JIP1, Dr. Billington understood that the main problem was seen as being the overestimate of the value of δ in J101. In January 2013 Atkins had produced a Technical Note which concluded that the cause of slippage was abrasive wear of the grout and consequent reduction in the mechanical interlock between the grout and the steel.

However, in the summer of 2013 Dr. Billington commissioned a Mr. James Lockley, of Atkins, to carry out a finite element analysis in order to see if it was possible to model a mode of failure by ratcheting down (or the “walkdown” mechanism).

In his report in reply, Dr. Billington summarised the differences between his walkdown mechanism and that favoured by Professor Schaumann in the following terms:

“In my Expert Report I have presented results of finite element analyses which demonstrate a walkdown failure caused by an interaction between the axial load and the bending moments and natural shear forces which I have described as being driven by the axial force and ratcheted down by the bending and shear. This occurs due to the low axial capacity and the distortion of the cross section due to the bending and shear which causes separation (gapping) between the grout and steel thereby further undermining the axial capacity.

Professor Schaumann, in his Expert Report, maintains that the interaction between axial load and bending moment is beneficial in increasing the effective axial capacity but that this is undermined by wear of the grout caused by abrasion of the steel against the grout in the regions of sliding and separation. I do not agree with this and in this report have included a reference to research work at E.ON’s Ratcliffe Research Technology Centre which demonstrates the overall wear during the design life of Robin Rigg to be less than 1mm at the ends of the connection.”

The last sentence of this passage was a reference to the work of Mr. Dallyn.

Mr. Lockley’s work was based on three different load cases. The first was an alternating north-south oscillation. The transition piece structure was assumed to bend in one direction then revert to the zero point and then bend equally far in the opposite direction. The second load case was bending to each point of the compass in succession, returning to the zero point in between. For example, the structure would bend first to the north, then back to the zero position, then bend to the east and then back to the zero position, and so on. The third load to case was a rotation around the central zero point without ever returning to the zero position.

During the course of cross-examination it became clear that Dr. Billington was unable to say how often any of these load cases might occur in practice. In his report Dr. Billington said that the results showed that there could be an annual slippage in excess of 10mm. However, it emerged in evidence that this was based on the assumption that load case 2 could occur sufficiently frequently as to produce 25% of the ULS loading about 40,000 times per annum, or about every 15 minutes (Day 7/13). In fact, this is not what the fatigue histogram (which was not produced at the trial) apparently showed. Dr. Billington accepted that the 40,000 events per annum at which loads could reach 25% of ULS could be made up of any combination of the three load cases (Day 7/7).

In the case of load case 1, the slippage per cycle at a load of 50% ULS was said to be .0067 mm if the coefficient of friction was taken at the lower bound of 0.4 (and nothing if the coefficient was taken at the upper bound of 0.6), but that was a load that would not be seen more than about 100 times per annum, so the worst case would be an annual slippage for load case 1 of about 0.7 mm per annum. No slippage figures were given for a load of 25% of ULS for load case 1. Dr. Billington made the valid point that this analysis was carried out using the value of δ used in J101, whereas it is now known to be about an order of magnitude lower so that the actual effects would be much more severe.

However, so far as this point is concerned, unless the designer was aware of the error in J101 in relation to the value of δ, I would expect him or her to take the results of Mr. Lockley’s finite element analysis (if carried out) at face value. If it were to be assumed, for example, that 25% of ULS for load case 1 would occur about 40,000 times per annum and that 25% of load case 2 might occur about 400 times per annum, the total slippage over 20 years would be of the order of 5mm, which might well be acceptable.

It was perhaps unsurprising, therefore, that in the end Dr. Billington accepted that all that this exercise really showed was that if there were wind loads of a certain type with a certain frequency, there could be an unacceptable degree of slippage and therefore, as a matter of prudence, he said, a reasonable designer should do something about it (Day 7/34-35).

In fact, the environmental conditions for which the foundations had to be designed were clearly set out in the design documents. Unfortunately, Dr. Billington’s report in which Mr. Lockley’s work first appeared was not disclosed until a few weeks before the trial (and then without the fatigue histogram on which it was based). MTH understandably objected to the very late production of this material and I consider that they were justified in doing so. In the event, there was no analysis before the court of the assumptions made in Mr. Lockley’s work - which were contentious - in the light of the assumptions made and agreed at the time of the design.

For these reasons I am not prepared to conclude that a similar exercise, if properly carried out by Rambøll in the course of its design using the agreed environmental conditions, would necessarily, or even probably, have demonstrated that the assumption that axial loads and bending moments could be treated separately was not a valid one. In addition, I am not satisfied that Mr. Lockley’s work comes anywhere near demonstrating that Dr. Billington’s “walkdown” theory is correct. It may be: but on the evidence before that court I do not find that to be the case.

In the light of my conclusion on the contractual issues, I consider that E.ON is entitled to a declaration that “the Problem” (as referred to in the Grouted Connection Agreement) has arisen as a consequence of a breach by MTH of clause 8.1 of the Conditions and section 3.2.2.2 of the Employer’s Requirements.

I will hear counsel on the appropriate form of relief, any questions of costs and other matters arising out of this judgment.