Petroleo Brasileiro SA (Petrobras) v Petromec Inc

The riches of Brazil’s offshore oil discoveries have brought litigation in their train. In late 1996 Brazil discovered the immense oil fields in the Roncador area of the Campos Basin. As a result Petroleo Brasileiro SA or “Petrobras” as it is known, the then state-owned Brazilian petroleum company, decided, in approximately January 1997, to switch to the Roncador field, from its previously planned use in the South Marlim field, a production platform or rig, at one time known as Spirit of Columbus but renamed, less romantically, P-36, which it was at that time negotiating to acquire, effectively on hire-purchase terms, from businesses associated with Mr German Efromovich. It will not be necessary to distinguish between the several different companies which represented the Petrobras interests on the one side and Mr Efromovich’s interests on the other. It will be convenient to refer to the latter under the name of “Petromec”, taken from the name of Petromec Inc which was incorporated for the purpose of entering into a contract with Petrobras for the upgrading of the platform and the managing of the work. The litigation which has ensued following the breakdown of relations between the parties, litigation which might at one time have been settled, has unhappily created an industry of its own. There is also tragedy, because the platform, which after its upgrading commenced production in the Roncador field in May 2000, sank with the loss of eleven lives on 20 March 2001.

A fuller account of the background to the upgrade of the platform and of the dispute between the parties can be found in the judgment of Moore-Bick J in Petromec Inc v. Petroleo Brasileiro SA & Ors [2004] EWHC 127 (Comm), and in the judgment of Field J [2011] EWHC 2997 (Comm) from which this appeal arises. It will be sufficient for present purposes to outline the following matters.

The platform had been originally built in Italy for an Italian owner to whom it was delivered in August 1995. However, its owner could not find employment for it. Negotiations therefore ensued for its sale to Petromec and for its use by Petrobras. The platform needed to be refitted to make it suitable for use in the South Marlim field. A general specification was developed for that purpose. The concept was for the platform to be bare-boat chartered by Petromec to Petrobras for twelve years. The hire payments under the charter were designed to compensate Petromec (and the interests of the previous owner) for the platform and its rebuild. In effect, therefore, it was to be a fixed price contract, where the fixed price was represented by the hire payments over the twelve years. At the end of the bareboat charter, Petrobras had an option to purchase the platform.

On 6 November 1996 Petrobras and Petromec entered into a Memorandum of Agreement. It subsequently became a matter of issue between the parties whether or not that document constituted a binding contract. That issue was resolved by a previous judgment in this litigation in favour of the memorandum not being binding.

It was more or less just as Petrobras was considering whether to deploy the platform away from the South Marlim field and into use in the Roncador field that the negotiations for both the specification for the South Marlim rebuild and for the bareboat charter came to fruition in December 1996. A general specification (the South Marlim General Technical Specification) had been developed and submitted by Petrobras to Petromec in June 1996. However, changes to that specification were being developed as matters progressed. Such changes were being collected in a document called “Annex X Deviations of Basic Design” (“Annex X”). Naturally, the price of the bareboat charter, that is to say its daily hire rate, would, at least in some sense, reflect the value of the platform as rebuilt. Nevertheless, the negotiations on the South Marlim specification and the negotiations on the bareboat charter were dealt with by separate teams, one composed of engineers and the other of business managers with Mr Efromovich himself playing an important role.

It is not at all clear, on the judge’s findings, whether there was any full understanding on the part of either team as to the cross-ramifications of such negotiations. Thus the judge found:

“36.

The negotiation of the charter hire rate was a commercial negotiation. There was no in-put from the Petrobras and [Petromec] technical people as to the new compression specification and no-one on the Petrobras Negotiating Committee was consulted about this change.”

The reference to the “new compression specification” will become clearer below. It plays a significant role in this appeal. However, this was only one technical item (albeit an important item) among others which were being negotiated by reference to Annex X.

Despite the judge’s finding cited above, he also finds that there came a time in the financial negotiations when Petrobras, who was pressing for a reduction in the daily hire rate being quoted by Petromec, asked Petromec to requote “on the basis that Petrobras would provide the mooring lines and that the compression system should comprise the existing compression system on P36 and one new system with a capacity of 2 million m3/d” (para 34). This proposal led to a drop of some $19 million in the amended estimate for the upgrade. The judge does not, however, give dates for this proposal or for the amended estimate at this point of his judgment. Mr Efromovich could not remember the timetable of events (para 42).

On 2 December 1996 the charter rate being asked by Petromec was reduced from $166,400 to $154,500 per day. The judge accepted (paras 42-44) hearsay evidence from a witness (Mr Hawksley, who represented the previous owner’s interests) that that reduction was premised on the proposal referred to above, and that this occurred between the end of November and 2 December 1996. The $154,500 figure is referred to in a Petrobras board minute of 2 December. Ultimately, agreement on the charter hire was reached on 19 December 1996 when Mr Efromovich made a further offer of $149,800 per day which was accepted by the Petrobras board. The judge records an explanation of that reduction as being due to a recalculation of the interest rate applicable over the charter period (para 35). No binding contract was entered into at that time, however.

On 28 January 1997 Petrobras informed Petromec that it intended to employ the platform on the Roncador field instead of the South Marlim field. It had been considering that change internally, with the assistance of consultants, for at least a month. Conditions in the Roncador field were quite different in many respects from those in South Marlim and necessitated a new design for the upgrade. Work began on developing such a specification in March 1997. Despite the lack of contract (or possibly because of uncertainty on that score, or possibly because of the close working relationship that had existed over the years between Petrobras and Mr Efromovich, I know not), there was an unwritten understanding that the parties would go ahead on the basis of the charter agreement which had been arrived at, with extra compensation to be paid to Petromec for the additional cost of the Roncador upgrade over and above what the South Marlim upgrade (as amended by Annex X) would have cost. It was on the basis of this understanding that the bareboat charter was executed on or as of 20 June 2007.

On 14 July 1997 P-36 left Palermo for Quebec, where it was to be upgraded by Davie Shipyard. It arrived there on 29 August 1997 and work commenced on 26 September 1997.

It was not, however, until August 1998 that the contract expressive of the unwritten understanding between the parties came to be made. That contract was called a Supervision Agreement and was backdated to 20 June 1997, the date of the bareboat charter. The Supervision Agreement distinguished between the South Marlim specification as amended by Annex X (the “Original Specification”) and the new specification developed for the Roncador upgrade (the “Amended Specification”). Clause 12.1 provided:

“12.1

In consideration of Petromec’s agreement to upgrade the Vessel in accordance with the Amended Specification Brasoil agrees to pay to Petromec an amount equal to the reasonable extra cost (if any) to Petromec of Upgrading the Vessel in accordance with the Amended Specification over and above the cost that Petromec might reasonably have incurred in Upgrading the Vessel in accordance with the Original Specification.”

That agreement has led to a series of disputes being worked out in this litigation as to how much is involved on either side of that equation. In this appeal we are concerned with two issues of substantial importance concerning what would have been involved in upgrading the platform for South Marlim in accordance with the Original Specification. One issue concerns the compressors and the other issue concerns the risers.

The compressors were critical to the productive capacity of the platform. I should emphasise that the platform was a production, not a drilling, rig. Its purpose was the extraction of the oil from under the seabed. The purpose of the compressors was to inject gas under compression into the oil wells in order to keep the reservoir pressures up. The productive capacity demanded of the platform by Petrobras was 150,000 barrels of oil per day: and in order to achieve that the South Marlim general specification required a gas compression capacity of 6 million m3 per day, of which 4 million was the required operational capacity and 2 million m3 was a redundancy allowance. The concept was that there would be three compressors each of 2 million m3 capacity, so that there would always be one spare compressor if any of the three was out of operation at any time.

The general specification therefore provided as follows:

“P3.1.2 – Design Parameters

Gas Compression Capacity: 6000000 m3/d (20oC and 101.3 kpa abs)

…

GAS COMPRESSION

Gas from Safety Gas K.O. Drum is sent to 3 x 50% three stage compressors (3 x 2000000 m3/d). Each compressor stage shall be provided with a cooler, scrubber and compressor. A cooler shall be installed at third stage discharge.

The gas production facilities and utilities shall be design to attend total gas compression capacity (6000000 m3/d)

The compressor performance shall be guaranteed to the flow of 2000000 m³/day at package inlet (at 20oC and 101.3 kPa abs) with gas molecular weight of 18.5…”

The operational capacity of 4 million m³ per day was inherent in the language of “3 x 50%...compressors (3 x 2000000 m3/d)”, where 50% was necessarily half of 4,000,000 m/3 per day.

The judge explained:

“20.

As gas is compressed its temperature rises and it is necessary to cool it at certain stages in the compression process. During the cooling process, liquid droplets are formed which are removed from the gas stream by separation equipment known as a “scrubber” or “knock out drum” prior to entry into the next compression stage. The whole package of equipment that compresses the gas, including a gas turbine that generates pressure through a drive shaft, and the scrubbers, is known as a “compression train”.”

The platform as built in Italy had a single (Delaval Stork) compressor on it. It was not suitable for Petrobras’s requirements.

The South Marlim specification was a generic specification applicable to all Petrobras production platforms to be used in South Marlim. As discussions on the specification developed, an alternative arrangement of having 2 compression trains each of 4 million m³ per day was considered. That would give both the necessary operational capacity of 4 million m3 per day and redundancy in an equivalent amount. Thus in July 1996 a possible deviation was minuted by Petrobras as follows:

“PETROBRAS re-confirmed the compression requirements as stated in the specification. [Petromec] prefers to use compressors based on LM 1600 gas turbines. PETROBRAS stated that their studies indicated:

● 3 x 50% units driven by LM2500’s or

or 2 x 100% units driven by LM 5000’s

[Petromec] will study the options further and submit to PETROBRAS approval.”

In October 1996 Petromec’s professional consultants recommended the adoption of 2 x 100% compression trains using RB211 gas turbines. This was approved by Petrobras as being commercially viable, so that on 6 December 1996 the following deviation to the specification was agreed:

“BRASOIL re-confirmed the compression requirements as stated in the specification. BRASOIL stated that their preliminary studies indicated:

-

3 x 50% units driven by LM 2500’s or

-

2 x 100% units driven by LM 5000’s

CONTRACTOR will study the options further and submit to BRASOIL approval. Both alternatives may be accepted.”

It will be recalled that 6 December 1996 was after Mr Efromovich had offered a reduced daily charter rate of $154,500, apparently on the basis of inter alia a compression system comprising the existing compressor train and one new train with a capacity of 2 million m³ per day (see at [8] above).

On 10 December 1996 Petrobras reverted to the original concept of a 3 x 50% compression system and abandoned the possibility of a 2 x 100% system and this was confirmed between the parties on 11 December 1996. Even so, an equipment list sent to Davie Shipyard on 3 January 1997 specified 2 x 100% compressors. I should also mention that the Roncador upgrade ultimately required 3 x 2.4 million m³ per day compressors.

As of 19 December 1996 Annex X contained a reference to Petrobras’s acceptance of responsibility for the supply of mooring lines (“All mooring system lines, anchors/suction piles and installation shall be supplied by BRASOIL”), but made no reference to the “deviation” with respect to the compressors. As stated above, the drafting of the compressor deviation had been going through various stages; however it was finalised via a fax dated 3 January 2007 which stated:

“P.3.1.2: BRASOIL re-confirmed the compression requirements as stated in the specification.

The gas compression system will consist of the following equipment: 1 off new gas compression train with a capacity of 2,000,000 Nm³/d, at 20ºC/101, 3 kPa abs.; and 1 off existing HP and Export Gas Compressor, as supplied by Delaval Stork driven by an EGT RLM 1600 gas turbine.”

The first issue which arises on this appeal is as to the proper construction of that deviation. There was an obvious mismatch between the specification’s demand for a compressor capacity of at least 6 million m³ per day and the compressor deviation’s statement that the compression system would consist of only one (new) 2 million m³ capacity compression train and the existing train already on board. It was common ground that the latter could not be used at all in South Marlim without great and uneconomic expenditure. Even if rebuilt, it would at best produce a capacity of materially under 2 million m³/d, namely 1.75 million m³/d. As developed below, the essential question raised by the compressor issue is to what extent the original specification’s demand for a compressor capacity of at least 6 million m³ per day and its consequences for the design of the platform remained the responsibility of Petromec as contractor as distinct from falling back on to Petrobras as employer.

The second issue which arises on this appeal is as to the meaning and effect of a further deviation in Annex X, albeit this one had been agreed already by July 1996. It concerned the means by which the various risers or conduits that move the fluids and utility services between the platform and the sea surface and seabed are to be accommodated and attached to the platform. The deviation concerning these risers reads as follows:

“The central caisson may be used for attachment of the risers and using the [d]raw works for pull in. BRASOIL accepts this concept. It was agreed that CONTRACTOR would contact flexible riser manufacturers and develop a study to maximise the use of the central caisson. The design would be discussed and finalised with BRASOIL.”

The judge explained that the functions performed by risers included movements of gas, oil and water from production wells to the platform, of gas from the platform into the well reservoirs, and of electric and hydraulic power to valves and other equipment. Such risers needed to be attached to the platform in an efficient manner. The deviation cited raised the question of using the central caisson of the platform to contain the risers. The central caisson is essentially a large metal tube or tower, up to 20 metres in diameter, which drops from the platform, through its decks, into the sea. In July 1996 Petromec had instructed consultants to carry out a feasibility study into the concept of using the caisson to contain the risers. In September 1996 those consultants (Noble Denton Europe) concluded that the 98 risers involved could be accommodated in the caisson, but in October 1996 doubts were raised by other consultants (Coflexip Stena Offshore). The South Marlim general specification had required dry attachment of the risers to the pontoons. Dry attachment means attachment above the surface of the sea. It is much harder to attach and maintain the risers beneath the sea level. The riser issue raised the question of how, and it therefore followed at what cost, the risers could and would have been accommodated and attached pursuant to the Original Specification, had the platform been upgraded for the South Marlim field. Ultimately, as part of the Roncador upgrade a different concept of attaching the risers to a so-called “spider deck” constructed between the lowest of the decks and the pontoons was developed and executed.

For convenience I will restate the critical language of the relevant deviation:

“P.3.1.2: BRASOIL re-confirmed the compression requirements as stated in the specification.

The gas compression system will consist of the following equipment: 1 off new gas compression train with a capacity of 2,000,000 Nm³/d, at 20ºC, 3 kPa abs.; and 1 off existing HP and Export Gas Compressor, as supplied by Delaval Stork driven by an EGT RLM 1600 gas turbine.”

It will be observed that this deviation contains two sentences somewhat in tension with one another. They have been referred to below and on this appeal as the “first sentence” and the “second sentence”. The first sentence appears to reconfirm the full compression requirements of the South Marlim general specification, viz 6 million m³ per day, with an operational requirement of 4 million m³ and a redundancy requirement beyond that. The second sentence however appears to say that the compression system will consist of only one new compressor train and the existing compressor which, as the judge accepted, was not suitable for its requirements. The question is how that tension is to be resolved.

It is easy to state in outline but extremely difficult to state in detail the question of construction raised by this issue. It was common ground below, and even to some extent in this court, that the deviation was an exercise in cost allocation. That is to say that the second sentence ameliorated in some way the cost implications placed on Petromec by the specification referred to in the first sentence. That at any rate remained Petrobras’s preferred way of putting the point, and, as will be seen, it was the judge’s solution. However, on behalf of Petromec, Mr Nicholas Vineall QC veered before us from upholding the judge’s reasoning to expressing his submission in a form which amounted to a more fundamental reshaping of the South Marlim specification as a whole.

At any rate this issue, together with other preliminary issues, had been stated by David steel J in an order dated 22 October 2010. As so stated in Cooke J’s order, the issue asked what the meaning and effect of the compression deviation was, and continued:

“In particular:

1.2

Does it mean, as Petrobras contends, that the cost that Petromec might reasonably have incurred in upgrading the Vessel is to be assessed by reference to the cost that Petromec might reasonably have incurred to purchase one new compressor with a capacity of 2 million cubic metres per day together with the costs associated with the retention of the existing compressor; but otherwise those costs are to be assessed by reference to a total compression requirement of 6m cubic metres per day with inter alia three new compressors and three trains of compression; or

1.3

Does it mean, as Petromec contends, that the cost that Petromec might reasonably have incurred in upgrading the Vessel is to be assessed by reference to a gas compression system consisting of one new compression train with a capacity of 2mn³/d (at 20ºC/101, 3KpA abs) and the existing HP and export Gas Compressor, as supplied by Delaval Stork driven by an EGT RLM 1600 gas turbine?”

That formulation of the issue is not, it seems to me, tremendously illuminating: but I think it is essentially asking whether the second sentence is subject to the first sentence, or whether the second sentence exhausts Petromec’s obligations with respect to the whole of the specification’s requirements relating to compression, so as in effect to make the first sentence subservient to the second sentence.

In his judgment below Field J described the parties’ contentions in the following way, expanding on the order’s terseness, but I think with essentially the same contrast which I have just indicated:

“52.

In a nutshell, Petrobras’ case is that the second sentence governs the cost that Petromec might reasonably have incurred to procure the gas compression system described in the second sentence but otherwise (ie apart from the cost of procuring the gas compression system described in the second sentence), the cost that Petromec might reasonably have incurred in upgrading the platform in accordance with the [Original Specification] is to be assessed by reference to the reconfirmed gas compression requirements under the first sentence…

54.

Petromec submitted that the first sentence of the deviation makes no sense when the deviation is construed as a whole in the light of its factual matrix and should be ignored. On that basis, the second sentence should be given its ordinary meaning, namely, that Petromec’s sole obligation in respect of gas compression equipment was to supply a gas compression system consisting of one new 2 million m³/d gas compression train and the existing gas compressor.”

The difference between the effect and consequences of these opposing submissions was explained by the judge in this way:

“53…Thus on Petrobras’ case, whilst it would be responsible for meeting the cost of the two additional 2 million m³ per day trains it says are necessary to achieve the total compression requirement of 6 million m³ per day, Petromec would be responsible for meeting the whole of the rest of the cost incurred in upgrading the platform in accordance with the [Original Specification], including the cost of (i) extending the deck space; (ii) gas processing equipment (eg TEG Contactor used for dehydration of natural gas); (iii) the cooling system capacity and other associated items.

54.

Petromec submitted…If Petrobras wanted any additional gas compression equipment, it would have to instruct a variation to the specification and pay for it separately, including the cost of any additional deck space.”

The judge resolved this dispute in the following passage of his judgment:

“59.

In my judgment, the compression deviation was intended to operate as a cost allocation exercise. One new 2 million m³/d compression system together with the existing system were never going to be sufficient to answer Petrobras’ compression requirement of 6 million m³/d and it is common ground that Petrobras was going to have to pay the additional acquisition cost of such compression systems as might be specified to achieve Petrobras’ overall compression requirement…

60.

The question is whether any further costs that would arise out of Petrobras’ compression systems specification, such as additional deck space and gas processing equipment, are allocated by the compression deviation to Petrobras’ account rather than to Petromec’s account. In my judgment, they are. In my opinion, the meaning and effect of the compression deviation when construed against the relevant background is that, apart from Petromec having to supply the existing compressor and one new 2,000,000 m³/d gas compression train, the whole of the cost of acquiring, installing and accommodating such further compression systems, including additional deck space and processing equipment, as might be specified by Petrobras to meet their compression requirement of 6 million m³/d, was for Petrobras’ account.”

Having stated his solution by reference to the outcome of the issue, the judge then analysed the problem of construction in these critical paragraphs:

“61.

In my opinion, the first sentence of the compression deviation is not meaningless. Its function is to specify the area in which the cost allocation provided for in the second sentence is to operate and to require Petromec to supply and install the compression systems specified by Petrobras to satisfy its compression requirements, albeit at Petrobras’ cost and after the execution of an agreed variation. In my view, the construction which I propose to give to the variation is a commercial one. It recognises that it is for Petrobras to specify what compression configuration it wanted to achieve the overall daily compression requirement and it contemplates Petromec supplying and installing what is specified, with the cost of the supply, installation and accommodation and all associated costs being borne in accordance with the allocation agreed by the parties. I do not think that the first sentence imposes an overarching obligation on Petromec to deliver a rig capable, with modification if necessary, of accommodating the necessary compression systems to answer the requirement of 6 million m³/d. But even if it does, this does not narrow the width of the cost allocation provision so as to make Petromec liable for the cost of additional deck space to accommodate 3 x 50% systems if this be the configuration Petrobras specifies.

62.

In my opinion, once it is recognised that the compression deviation is a cost allocation provision, the fact that Petrobras reverted to a 3 x 50% configuration on 10 December 1996 is nothing to the point. The compression deviation was negotiated without reference to what the engineers were saying to each other. And on any view, some system or systems additional to those identified in the compression deviation were going to be required. It was for Petrobras to specify what its requirements were subsequent to the compression deviation being agreed. The original gas compression specification (3 x 50%) in the South Marlim GTS did not spring back into life. The effect of the deviation is that the acquisition cost and all associated costs of such additional systems as might be subsequently specified are to be for Petrobras’ account and not Petromec’s.”

On this appeal, Petrobras, by Mr Christopher Hancock QC, submits that the judge was correct to regard the deviation as a cost allocation exercise, but wrong to reject its preferred construction.

Petromec, by Mr Vineall QC, submits that the judge came to the right conclusion, albeit he seeks to support it by further or different reasoning (while at the same time suggesting that it is Petrobras’s formulations of the effects of its preferred meaning which have fluctuated). Thus he implicitly recognises that there is a difficulty in the judge’s subjection of the first sentence of the deviation to almost meaningless status and thus now prefers to describe the deviation not so much as a cost allocation exercise upon an otherwise agreed specification but as a fundamental change of specification: to the effect that the specification indicated by the first sentence is altered to a new specification indicated by the second sentence. He glosses this as the deviation saying that the two sentences are to be interpreted as joined by the words “save that”. This began as his secondary construction, for his primary construction, although rejected by the judge, was that the first sentence should be rejected in total as meaningless: but as his submissions developed, this became his primary submission. He also recognised that the judge’s assertion in his para [61] that his construction is “a commercial one” needed support, and he sought to do that by suggesting that there is evidence that, in at any rate the first three years of oil production at South Marlim, the existing compressor could be brought into action so as to cover, together with the new 2 million m³ per day compressor, a planned production which would only require 3.664 million m³ per day. Thus he suggested that there was practical support for an altered specification which substituted the limited compression equipment mentioned in the first sentence in place of the 6 (or even 8) million m³ per day of capacity otherwise referred to in the Original Specification. On that basis, there was to be a fundamental specification change in this regard which would suffice for a few years, and then the future could look after itself. In this connection he stressed that Petrobras was already contemplating a shift of the platform to the Roncador field (with all the changes which that would entail), so that a compressor specification which could not achieve the platform’s stipulated production capacity in South Marlim did not matter, or might be so to speak finessed. He also stressed that the compressor deviation, as part of the final Annex X deal, was commercially, not technically, driven: what was important from Petrobras’s point of view was that a deal was achieved, especially important if the vessel was to go to Roncador in circumstances where there was a shortage of suitable platforms, and it did not matter whether or not the deal quite made technical sense. Again, the future could look after itself.

The judge set out detailed findings in his judgment concerning the course of the negotiations which led to the agreement of the bareboat charter rate on 19 December 1996 and to the negotiations that were going on more or less contemporaneously concerning Annex X and in particular the compressor deviation, albeit that did not reach its final wording until 3 January 1997. (Even so, his findings nowhere reached, or even addressed, the considerations which Mr Vineall came to put at the forefront of his case.) As for the judge’s narrative of the negotiations, however, it is not at all clear whether or how these findings affected his analysis and judgment on the current issue, but it would seem that in his own mind they did. Thus the only citation of authority in his judgment was of passages from Investors Compensation Scheme Ltd v. West Bromwich Building Society [1998] 1 WLR 896 and BCCI v. Ali [2001] 1 AC 251 which deal with factual matrix. Moreover, he then went on to state that “The relevant factual background for purposes of interpretation consists of those matters set out in paragraphs 19-33 and 35-39 that were known to both parties down to 19 December 1996”. It may be observed that he deliberately omitted para [34] from that list, which is the paragraph which actually deals with the proposal that the upgrading costs be recalculated on the basis of the limited compression equipment which Petromec was to supply. Subsequently, in setting out his preferred solution of the problem he said that the compression deviation “when construed against the relevant background” had the meaning for which Petromec contended, but he is not explicit about what within that background may have influenced his judgment. On the appeal counsel speculated that the judge’s para [34] had been omitted from his list of paragraphs dealing with factual matrix because it concerned matters of negotiations which continue to be inadmissible on issues of construction: Prenn v. Simmonds [1971] 1 WLR 1381, ICSL v. West Bromwich at 913 at principle (3). In my judgment, however, it is not clear that any of the “factual background” material discussed by the judge stands clear from the Prenn v. Simmonds exception of inadmissibility. Nevertheless, the parties were not concerned to debate about such matters, but were content to make submissions as to the proper construction of the compression deviation on the basis that it was part of an effort to reach agreement on a price (the bareboat charter daily hire) and involved a concession to Petromec, the extent of which was in issue, as to its financial responsibilities under the Original Specification.

However, it was the extent of that concession that remained in issue, and in that connection it seemed to me that there were no findings in the judge’s judgment, arguably admissible or inadmissible, which were at all helpful. In the course of his submissions Mr Vineall came to recognise this. He was at one point asked by the court what matters of common knowledge he relied on as factual matrix for the purposes of his submissions on construction, and his answer took matters no further forward. He referred to the fact that the negotiation was a commercial one and that the technical team did not participate in it; that the first sentence of the deviation had emerged from previous formulations of Annex X; and that the genesis of the compression arose in the proposal discussed in the judge’s para [34]. I am doubtful about the admissibility of any of those matters, but it is not necessary to decide the points. However, what was clearly relevant was that the underlying terms of the general specification from which Annex X set out the agreed deviations contained provision for 3 x 50% compressor units amounting to a total capacity of 6 million m3 per day. It may also have been relevant that the technical team which agreed Annex X knew that there was continuing uncertainty as to whether the requirement for compressor capacity would be changed from 6 to 8 million m³ per day. However, ultimately it was the general specification and Annex X which were the agreed documents and which became annexed (in theory if not in practice) to the Supervision Agreement (as appendices A and B, together amounting to the Original Specification).

It is in truth in any event not easy to analyse the relevant date for the purpose of considering matters of factual matrix. There was no contract until 20 June 1997 at earliest. The Supervision Agreement was not agreed until August 1998. Nevertheless, the parties were content to take 19 December 1996 as the relevant date and so was the judge (at para [49]).

In this connection it is worth mentioning that Mr Vineall’s reliance on matters such as (i) Petrobras’s consideration of a switch to Roncador, or (ii) the possibility that it was concerned to secure agreement on the bareboat charter hire in order to secure the platform for Roncador where it is said that it was the only suitable platform in the market, or (iii) the possibility that the existing compressor could have been brought, albeit with difficulty, into service for South Marlim, seem to me to run into difficulty. The first matter (i) was not, on the facts found by the judge, a matter of common knowledge until Petrobras told Petromec on 28 January 1997. There is no finding as to the second matter (ii). In any event, these two matters were at the relevant time matters of subjective consideration only. It is not possible to construe contracts on the basis of one party’s private considerations. In any event, it is not possible to say whether there was more pressure on Petrobras or on Petromec to secure a deal. After all, the platform had been built and was without employment.

Mr Vineall also relied on certain textual points. Indeed, he submitted ultimately that a textual analysis could suffice by itself. He stressed the words “will” (a mandatory term) and “consist of” (not, he pointed out, “include”) in the second sentence, as well as the phrase “compression system” which picks up the language of the specification. He submitted that the language of the second sentence could not be clearer, whereas the first sentence was not as clearly mandatory. He also relied on maxims of construction such as that more specific language should supersede more general language, and that, because Annex X was a document which originated with Petrobras, it should be construed against it (contra proferentem).

The issue is a short point of construction, of some difficulty. On any view, the matter could have been better expressed. The difficulty is increased by the fact that there has been no real attempt in this court to explain how the compression system affects the specification or design for the construction or upgrading of the platform by reference to the siting of the compression trains. We have just been left with the general understanding that the acquisition of the compression train equipment is one matter, and their siting and integration is, naturally enough, another matter. Nor have we been taken to other passages in the general specification (or in Annex X) which might throw light on the problem of construction. However, in my judgment, the submission of Petrobras is to be preferred. I say so for the following reasons.

First, the first sentence is plain in its meaning, which is informed by the general specification, to which it expressly refers. The compression requirements of the specification are quite plain: apart from paragraph P3.1.2 of that specification, there are also many other passages in the specification which are explicit in requiring an oil production of 150,000 barrels of oil per day, a compression of 6,000,000 m³ per day, and a “gas lift” (ie individual gas train) capacity of 2,000,000 m³ per day: see not only para P3.1.2, but also G1.1.2 and P1.3. G1.1.2 is headed “Main Characteristics of the Converted Unit” and specifies inter alia the figures for daily oil production figure, gas compression and individual gas train capacity. Indeed, P3.1.2, in reflecting these figures, is concerned specifically not with the overall design of the conversion but rather with the “Design Parameters” of the “Gas Systems”, ie the compression trains themselves as pieces of equipment.

Secondly, in the light of the first sentence, but also as a matter of common and commercial sense, it is impossible to read Annex X as intended to provide for a merely nominal or fantasy design, or, as Mr Vineall has at times suggested, a design whose only real function was to finesse the agreement of a daily rate for the bareboat charter, whatever the consequences for the platform and its productive capacity, and whether its future lay in South Marlim or Roncador. Whatever the two sentences of P3.1.2 of Annex X mean they have to make sense of a real world, in which the platform was to be upgraded for use in South Marlim. That was the common, objectively founded, intention of Annex X at the time of its agreement, whatever may have been Petrobras’s possible thoughts at that time of switching the dedication of P-36 to Roncador (which would and in due course did require a fundamentally new approach to its upgrading). Moreover, even after the platform was switched to use in Roncador, and the Supervision Agreement was made with its provisions for adding the extra costs of the Roncador upgrade to the “cost that Petromec might reasonably have incurred in Upgrading the Vessel in accordance with the Original Specification”, the function of that Original Specification (ie the general specification for the South Marlim upgrade as amended by Annex X) was to be a true, commercial, bench-mark for that agreement, and not a mere fantasy proposal.

Thirdly, however, it seems to me that Petromec’s construction involves the adoption of a fantasy specification for the South Marlim upgrade. Its premise is that although it is agreed that the South Marlim upgrade has to be capable of producing 150,000 barrels of oil a day, with a compression system in accord with that capacity, nevertheless the contractor’s responsibility is to produce a platform which will totally fail to meet those requirements and instead the employer will have the responsibility, under the specification amended by Annex X, to design, engineer and create a platform which can meet its overall specification. The judge may have seen this difficulty, although he does not expressly acknowledge it, and may have thought that it was solved by speaking of cost allocation. However, a true cost allocation, such as Petrobras’s argument contemplates, is simply a matter of switching responsibility for the cost of purchasing the equipment required by the underlying specification. What the judge appears to have contemplated was that, subject to a future variation in the specification, the platform upgrading to be designed, engineered, and constructed would be a fundamentally different platform from that contemplated by the same specification. That is in truth not a matter of cost allocation under an agreed specification, which is Petrobras’s contention, but a fundamental change in specification contemplating it is said a further fundamental change in specification down the line. Subject to that further change, Petromec’s construction involves Petromec engineering an upgrade which will not be fit for the specification’s purpose. Of course, in theory anything might be agreed, but the question is whether this is a sensible construction to put on the two sentences in question.

Fourthly, it seems to me that, consistently with the matters discussed above, the whole weight of the general specification is against Petromec’s argument. I will merely refer to the most general provisions of that specification. I have already referred to G1.1.2 headed “Main Characteristics of the Converted Unit”. Equally important, however, is the opening paragraph, G1.1.1 headed “Intent”. It reads in part as follows:

“The intent of this Specification and the accompanying documents of the BASIC DESIGN DOCUMENTS LIST…is to provide all the applicable technical requirements for the delivery of the…platform, ready in all respects for the intended service…Within the contracted scope of work, the CONTRACTOR shall develop the Basic Design and Detailed Engineering Design documents, studies, analysis, workshop drawings and specifications, provide all labor, services, facilities and equipment, handling, storage, fabrication, erection, procurement, installation, commissioning, testing, transportation and pre-operation (on its final location) of all systems, materials, equipment, machinery, fittings etc., or as may be required to complete the Unit, for its new purpose.”

In my judgment, these responsibilities remain, as they must, on the contractor, Petromec, save to the extent that it has been relieved of them by any deviation in Annex X which has the result of a fundamental alteration of the specification. The judge’s solution, however, has been to put back on to the employer, Petrobras, the essential functions of a contractor in achieving the designed upgrade so far as compression and thus productive capacity are concerned. I am unable to find this fundamental shift in Annex X, particularly in the light of the firm language of the first sentence (“re-confirmed the compression requirements as stated in the specification”).

Fifthly, I am unable to agree with Mr Vineall’s suggestion that the first sentence must be discarded as meaningless, or with the judge’s suggestion that its only function is “to specify the area in which the cost allocation provided for in the second sentence is to operate”. The reference to “P3.1.2” is sufficient for that identification, and in any event the judge has discarded the essential force of the first sentence’s language. He has in effect promoted the second sentence to pole position, giving it a meaning which is inconsistent with the first sentence having its natural force, and has concluded that the first sentence therefore has to be denuded of meaning. However, prima facie the first sentence, especially when it consists of such broad and insistent language, ought to be regarded, if it possibly can be, as the dominant sentence.

Sixthly, it has to be remembered that Annex X contains what are described as “Deviations”. In other words, the general specification remains the primary document, and the content of Annex X contains exceptions from it. As a matter of principle, an exception should not easily be construed as having a wider purport than its language requires, a fortiori when it follows language such as is found in our first sentence.

Seventhly, the second sentence is concerned with “equipment” (“the following equipment”). That is the nature of P3.1.2, to give the “Design Parameters” of the gas compression equipment. That is why the underlying P3.1.2 of the general specification emphasises that each compression stage “shall be provided with a cooler, scrubber and compressor”. Other technical performance parameters are listed in the specification. The two sentences of the deviation have to be read consistently together, if they possibly can be. In my judgment, they can be so read, if the second sentence is understood as a limited qualification on what would otherwise be the absolute reconfirmation of the requirements of the general specification. Thus Petromec does not have to supply “3 x 2000000 m³/d” compression trains of the stated design parameters, but only one, expressly described as a “new gas compression train” (plus the “existing Gas Compressor”). It follows that the responsibility for supplying the other two compression trains, to make up the “requirements as stated in the specification”, falls back on to Petrobras. That is the cost allocation exercise submitted by Mr Hancock.

Eighthly, it is to be noted that no design parameters are given for the existing compressor, it is simply described as “existing” and its basic characteristics are described (“as supplied by Delaval Stork driven by an EGT RLM 1600 gas turbine”). No design or capacity specifications are given for it, either of course in the underlying general specification or in the Annex X deviation with which we are concerned. Plainly, it would make no sense for the P3.1.2 specification design characteristics to apply to it. It is as it is. The only comment on it by the judge was that MSE (Consultants) Ltd (Petromec’s own consultants) had advised Petromec that it “was not suitable for Petrobras’ requirements” (para [24]). That was common ground at the appeal. Mr Vineall pointed to agreed expert evidence at trial to the effect that, although the existing compressor could in theory have been modified to meet the specification requirements for each of the three new compressors, such a modification “would have been extensive, impractical and uneconomic” (and even then would only have produced a capacity of no more than 1.75 million m³/d). Plainly, the deviation did not contemplate any modification being done to the existing compressor. Nor is any reference made in the deviation to an existing compressor train. In my judgment, the existing compressor is mentioned only because it is there. No other sense can be given to the reference to it.

Ninthly, the P3.1.2 deviation does not track the whole of P3.1.2 in the general specification, which is almost one and a half pages long. Plainly, the majority of P.3.1.2 is intended to be unaffected by the deviation. There is no exclusion, for instance, of the sub-paragraph which states: “The gas production facilities and utilities shall be design to attend total gas compression capacity (6000000 m³/d)”.

Tenthly, the judge’s reasoning does not compel or persuade a different conclusion. I am, respectfully, unable to agree with the assertion at the end of his critical para [61] that the first sentence and the underlying specification do not impose an overarching obligation on Petromec to deliver a rig fit for its purpose. Nor can I agree with the assertion that even if that is the function of the first sentence, the second sentence nevertheless trumps that overarching obligation in every respect beyond the provision of one new compression train and the existing compressor.

For these reasons I would conclude that Petrobras’s construction is to be preferred, with the effect and consequences, at any rate generally speaking, for which Petrobras contends. I say “generally speaking”, for I am not sure how that construction would affect any costs in dealing with the existing compressor. (I do not refer to its modification, which, as stated above, was clearly not contemplated.) If it would have simply remained on board in a redundant state, I see no reason at all for thinking that Petrobras should have incurred any cost at all in respect of it. If, however, it needed to be removed to make way for the new compressors, I can see arguments both ways, which I am unable to weigh, for saying that the costs of its removal might fall on either party. As it is, the parties did not descend to such details.

In sum, on this issue I would allow Petrobras’s appeal and adopt its construction.

Some further facts need to be stated about the risers and the possibility of using the central caisson to contain them (see paras [23]-[24] above).

The general specification for the riser system for the South Marlim upgrade stated (in H.13.2.2(i)) that “…All risers shall be connected to the outside perimeter of the Unit, according to drawing [there followed a number].” The drawing showed that the risers were to be attached in a specific order around the periphery of the pontoons. H.13.3 of the specification provided:

“Risers and control umbilicals arriving at port and starboard sides shall be connected to a riser hanger structure, arranged on top of pontoons, which shall provide the following operational features:

(i)

Considering the Unit in pull-in draught, the connecting point shall rise to a minimum air gap of 3.5m, measured from the connecting point to the surface of calm water;

(ii)

In addition, the riser hanger structure and accessories shall have a minimum immersion of 9 meters, taking into account the unit at operational draft, in order to avoid supply boats collision…”

The judge explained that “pull-in” draught was when the platform’s pontoons were floating sufficiently high for it to be pulled to destination, and that “operational” draught was when the platform was stationed at its operational location. He continued (at his [74]):

“There was therefore to be dry attachment of the risers at locations on the pontoons that would be 3.5 metres above the waterline at pull-in draught and that would be 9 metres below the water when the platform was in operational mode in the South Marlim sea.”

The general specification also provided as follows:

“G1.1.3 para 15: The CONTRACTOR shall follow the Basic Design requirements, but changes may be proposed, if they improve operation, decrease weight or simplify construction. Such changes, as any deviation from the technical requirements intended to be performed by the CONTRACTOR, have to be previously submitted to BRASOIL’s approval.”

“H.13.2: Although pontoons are the connecting positions for all risers on this document, CONTRACTOR may submit to BRASOIL’s approval an alternative positioning for risers support.”

The submission of such a proposal is what occurred in the context of preparing the Annex X deviations, for H.13 of Annex X provided (as already set out at [23] above):

“The central caisson may be used for attachment of the risers and using the craw [sc draw] works for pull in. BRASOIL accepts this concept. It was agreed that CONTRACTOR would contact flexible riser manufacturers and develop a study to maximise the use of the central caisson. The design would be discussed and finalised with BRASOIL.”

Against this background, the formulation of issue 3 was as follows:

“In relation to risers:

3.1

What number of risers would have been required for South Marlim and how were they required to be bundled or grouped? How would that have differed from the number and bundle grouping required for Roncador?

3.2

What is the meaning and effect of [the deviation from Annex X set out in the previous paragraph]…Does it mean, as Petrobras contends, that the result would have been that Petromec could reasonably have had to install a spider deck, the same as or similar to [that] installed for Roncador? Or does it mean, as Petromec contends, that Petromec could reasonably have used the central caisson for the attachment of the risers and in any event would not have had to install a spider deck?”

The judge answered the first part of issue 3.1, uncontroversially, in accordance with the agreement of experts on both sides, that if P-36 had been upgraded for South Marlim the number of individual risers would have been 98 (in Roncador there were 89). As for the second question in issue 3.1 (“and how were they required to be bundled or grouped?”), the judge found that only 52 of the 98 risers could have been accommodated in the central caisson and the remaining 46 would have had to have been attached to the pontoons. That is also common ground on this appeal, on the basis that attachment of the other 52 would have been to the central caisson. What remains controversial, however, is whether the central caisson idea would have been proceeded with at all in such circumstances, or whether the parties would have adopted the spider deck solution.

Relevant to those questions is whether the 46 risers which could not have gone into the central caisson could have been dry attached to the pontoons as the specification required. The answer to that further question was, No. The judge found that “Realistically, therefore, attachment to the pontoons of those risers that could not be accommodated in the central caisson would have had to be done under water through the use of divers” (at para [94]).

Petrobras therefore submitted that it would not have approved the central caisson solution to the attachment of the risers, and attention would have turned to the device of a spider deck, as occurred in relation to the Roncador upgrade.

Petromec, on the other hand, submitted: (a) that the central caisson solution had already been agreed by Petrobras in the H.13.2 deviation which formed part of Annex X (“BRASOIL accepts this concept”); (b) even though the central caisson solution could not accommodate more than 52 risers, nevertheless Petrobras was bound to go with that solution unless it had reasonable grounds for not doing so (see G1.1.3 of the general specification); (c) Petrobras did not have reasonable grounds for rejecting the central caisson solution because the remaining 46 risers could have been successfully attached to the pontoons even though by a process of wet rather than dry attachment; and (d) consideration of a spider deck solution did not come into it at all because, although it was always open to propose an alternative design solution, this had not been done prior to the relevant date of agreement, which was 19 December 1996.

It is not clear why it suits Petrobras to argue that a spider deck solution would have been adopted at South Marlim (as the cheapest and most efficient solution): since in theory if the platform could have been upgraded for South Marlim more cheaply, then prima facie Petrobras would have to pay more to Petromec for the Roncador upgrade. Be that as it may, however, Petrobras wishes to argue that the comparison between the cost of the actual Roncador upgrade and the hypothetical South Marlim upgrade would have involved a spider deck for South Marlim, just as it did for Roncador.

The judge’s answer to these questions was to adopt Petromec’s submissions. Thus, (i) Petrobras was bound to accept a central caisson solution, because it had agreed to it in the Annex X deviation; (ii) even though the central caisson could not accommodate all 98 risers but only 52 of them, the wet attachment of the remaining 46 risers to the pontoons could not reasonably be refused; and (iii), as for the spider deck solution, it was out of court because clause 12.1 of the Supervision Agreement asked what costs Petromec might reasonably have incurred in upgrading the platform for South Marlim “in accordance with the Original Specification”, and the spider deck formed no part of the Original Specification and could not be speculatively brought within it just because it might have, but had not, been proposed at the time of the agreement of Annex X. He said: “In my judgment, attachment of the risers to a spider deck was wholly outside the scope of the Original Specification” (at para [83]).

The judge reasoned points (i) and (ii) as follows:

“[85] The riser deviation derives its contractual effect from the Supervision Agreement. Like the compression deviation, it was part of the basis on which [Petromec] negotiated what was in effect a fixed price for the South Marlim upgrade. In my judgment, construed against this background, Petrobras could only refuse to accept a design for making maximum use of the central caisson for the attachment of the necessary risers if it had reasonable grounds for doing so. As a consequence, in practice, Petrobras were obliged to accept such a design if it were feasible and did not adversely affect to a material extent the safety and operational efficiency of the platform.”

The judge then considered the expert evidence as to various methods which could have been used in conjunction with the central caisson for the attachment of the risers, and concluded that the only solution that could be reasonably imposed on Petrobras was the wet attachment of the 46 (out of the total of 98) risers which could not be accommodated in the caisson itself. The judge then continued:

“[95] It is clear that Petrobras prefers dry-attachment of risers to wet attachment. However, in the case of P18 (designed in 1996), 71 of the 74 risers were wet-attached with diver assistance when it was discovered that [that] platform’s pull-in draft was worse than expected. In addition, 43 out of a total of 65 risers were wet-attached in the case of P52 in 2007 because steel catenary risers were used on account of the depth of the water and such risers need[ed] to be attached low down on the platform. It is also the case that 9 risers out of 70 were wet-attached on P26 to avoid having them run beneath the accommodation sections and on P36 itself, 6 risers were wet-attached to pontoons for similar safety reasons.

[96] Mr Grealish accepted that dry attachment is a simpler option than wet attachment but neither he nor Mr Taggart thought that wet attachment was not feasible.

[97] Mr Pinto stated in his first witness statement that he did not think that Petrobras would have agreed to wet-attachment of risers on P36 at South Marlim unless it was necessary for safety or specific design reasons: Petrobras’ preference would have been to install risers on a spider deck. Leaving aside the consideration that he was called as a factual and not an expert witness (this statement is essentially an expression of opinion), it is to be noted that he accepts, as he was bound to given the examples in paragraph 9[5] above, that Petrobras permitted wet attachment if it were necessary for design reasons.

[98] In my judgment, had Petromec proposed a design whereby 52 risers were to be attached to the central caisson and 46 were to be wet-attached to the pontoons, Petrobras could not reasonably have withheld their agreement. Such wet-attachment was feasible, had been done on a considerable scale in the case of P18 and the cost would have been for Petromec’s rather than Petrobras’ account.”

In this appeal, Mr Hancock has submitted that the judge’s reasoning is in error. He submits that (i) the central caisson deviation was not a firm agreement to use that solution irrespective of its merits but rather an acceptance that it could be considered further: such an agreement neither committed Petrobras to adoption of that solution, nor excluded any preferable solution such as a spider deck; (ii) a spider deck solution was not “outside the scope” of the Original Specification, since it was always open to the parties to adopt “changes…if they improve operation, decrease weight or simplify construction” (general specification G1.1.3); (iii) wet attachment of 46 risers was not subjected by the judge to that consideration and in any event did not follow from the use of a central caisson design for the attachment of the remainder of the risers; (iv) the judge was wrong to infer from the use of wet attachment on only 4 out of 25 platforms where it was necessary for safety or design reasons that it could reasonably have been imposed on Petrobras or would have been accepted by Petrobras in the current case, especially where there was a simpler and more cost effective means of dry attachment in the solution of a spider deck; (v) the attachment and commissioning of each riser was not part of the upgrade and was for Petrobras’s account, not Petromec’s.

Petrobras therefore seeks an anwer to issue 3.2 in these terms:

“The meaning and effect of [the central caisson riser deviation in Annex X] is that the result would have been that Petromec could reasonably have had to install a spider deck the same as or similar to that installed for Roncador.”

Mr Vineall, on the other hand, has submitted that the judge is correct for the reasons which he gave in his judgment. He says that so far as the judge’s answer depends on construction of the deviation, his reasoning is impeccable; and that so far as his answer depends on findings of fact as to what would have happened had the South Marlim upgrade been performed, his findings, such as whether Petrobras could have reasonably objected to the wet attachment of 46 riders, are unassailable. In this connection he also submits that (a) dry attachment was unlikely to be feasible in any event, and (b) the Roncador upgrade was not comparable and that it cannot be supposed that the spider deck solution adopted for it would have been adopted for the South Marlim upgrade.

In my judgment, the critical question for this court to decide on this appeal is whether the deviation in question did or did not commit the parties, both parties, to use of the central caisson for the attachment of the risers. If it did, then it seems to me that the rest of the judge’s reasoning may be said to follow: and I have heard or read nothing in Petrobras’s submissions to suggest that this court should interfere with the judge’s conclusions of fact on issues which were debated before him at length with the assistance of much technical and expert evidence.

If, however, the judge was wrong to conclude that the parties were committed to the use of the central caisson for the attachment of the risers, the question would arise whether any of the judge’s findings of fact could survive, and whether the judge had tackled and answered other questions of fact which might arise. Thus, would Petrobras be committed, on some objective test of what it was reasonable to do, to the use of the central caisson however many or few risers could be accommodated in the caisson itself, and irrespective of whether the remaining risers could be dry attached or would be wet attached? If the central caisson solution was ultimately not acceptable, what were the parties obliged to do, or, perhaps, what would the parties have done, about attachment of the risers, in accordance with the Original Specification? Would they have persevered with dry attachment, on the basis that, although difficult and expensive to perform, it was the solution required by the specification and much preferred by Petrobras? Or was dry attachment indeed not practically or economically feasible, as Petromec submits and the judge appears to have found, albeit in a somewhat different context? Or would they have come up with the allegedly much preferable solution of a spider deck, which was permitted by G1.1.3 of the specification, and resolved the contractual conundrum in that way? It seems to me that the judge has answered none of these questions, as on his way of looking at the matter he was not obliged to do. I do not see how this court can begin to answer such questions for itself. And the parties, although indicating in a somewhat pro forma way what their own answers would be or might have been to some of these questions, have understandably not really striven to obtain such findings from this court.

If the question of which party was to pay for attachment was relevant to any such factual issues, I do not think that that issue was adequately addressed to permit of any answer other than the judge’s answer, which was that attachment was for Petromec. After all, H.13.2 of the specification said that “CONTRACTOR must submit to BRASOIL’s approval the installation procedures for each riser” and H.13.2.3 said that “Risers and control umbilicals shall be connected to a riser hanger structure…”. Nothing explicit was brought to our attention which made it plain that actual attachment was for Petrobras. Nevertheless, although the point was disputed in Petromec’s skeleton argument, in oral submissions Mr Vineall conceded that the judge had erred at the end of his para [98] in failing to observe that actual hook-up costs at the site of the platform on location are for Petrobras. Be that as it may, I do not think that the factual issues raised above could be answered any better whichever answer is given to that question. On the contrary, this lightly delivered concession would raise other questions which the parties did not address: for instance, would there be sufficient performance of Petromec’s design obligations for structures which would support dry attachment if dry attachment would not be feasible? Moreover, Mr Vineall submitted or accepted that dry attachment could have been made feasible, but only at the cost of producing a rig which would not have met the specification’s requirements of stability and buoyancy. This emphasises that there are ramifications involved in the issue which are difficult to grasp.

There is also this qualification: Mr Vineall conceded in the course of oral submissions (contrary to his written skeleton) that it was likely that there would have been a spider deck on the South Marlim upgrade, but as a matter of variation (and therefore at Petrobras’s cost), and not as a matter of proposal pursuant to G1.1.3.

So I address the critical question of construction: Did the deviation commit the parties to use of the central caisson for the attachment of the risers? The strength of Mr Vineall’s case is in the July 1996 forerunner to the August 1996 version of Annex X which was the first to contain the deviation which came down to the final contractual form of Annex X. In the July 1996 version the deviation is formulated as follows:

“[PETROMEC] discussed the possibility of using the central caisson for attachment of the risers, and using the [draw] works for pull in.

PETROBRAS expressed possible acceptance of this concept. It was agreed that [PETROMEC] would contact flexible risers manufacturers and develop a study to maximise the use of the central caisson. The design would be discussed and finalized with PETROBRAS.”

By August 1996 this had become: “The central caisson may be used for attachment of the risers…BRASOIL accepts this concept”. The final sentence however remained the same (other than that “PETROBRAS” had become “BRASOIL”). Mr Vineall emphasises “may be used” and “accepts this concept”. He described this as agreement “in principle”, but later thought it prudent to abandon the expression “in principle” and talk in terms of commitment.

In my judgment, however, the final form of the deviation remains no more than an agreement in principle. It was subject to the same “study to maximise”, without which the design could not be further “discussed and finalized with BRASOIL”. The expression “may be used” is permissive, not mandatory. It is not “will be used”. The concept is accepted, but it remains a “concept”. What is to happen if the “study” which was yet to be developed demonstrates that the concept will not work, or that it will accommodate so small a percentage of the risers that its adoption will be unhelpful? And what is to happen if, as seems to have occurred, the adoption of this concept would not permit dry attachment (but perhaps dry attachment was impractical anyway)? It is one thing to say, as the judge reasoned, that, once a deviation has been agreed, that is to say committed to, then its working out may be subject to an objective standard, which the judge finds in the G1.1.3 language of “improve operation, decrease weight, or simplify construction”. It is quite another thing, however, to posit that an objective standard is to be applied at a stage short of commitment. If an employer specifies a certain item of specification, why should a contractor, short of contract, be entitled to demand a different specification on the premise that objective assessment will support him? He can seek to persuade his prospective employer, but he has no basis for demanding the right to prove his case on objective grounds. If, moreover, one posits that the specification has gone beyond proposal and has reached contractual status, so that the item of specification is agreed, why should the contractor, having signed up to that specification, be entitled to require an amendment of it because he can prove that such an amendment would be useful and efficient? He can persuade, but, if he cannot do so, why should he be entitled to insist, on proof of his view of the facts?

Therefore, I can accept that, if the use of the central caisson for the attachment of riders was an agreed amendment of the specification, then, if there was any outstanding detail of that amendment which needed to be worked out, it should be capable of being worked out on objective principles of implied reasonableness: for business efficacy requires such a mechanism. If, however, the amendment has not yet been committed to, as a matter of contract, I cannot see why the employer cannot say: “It does not suit me. I must be the judge of that. You cannot force a change of contract on me. In one way or another, you need to persuade me.”

It seems to me that this is how G1.1.3 is intended to work. The contractor “shall” follow the agreed specification. But changes “may be proposed”, but only “if they improve operation” etc, However, such proposals are not sufficient for agreement, “Such changes…have to be previously submitted to BRASOIL’s approval”. Proposals “may” be made, but changes “have” to be agreed.

In the circumstances there was no commitment to use of the central caisson come what may. The concept was given a green light, but agreement on any particular new specification or design had not yet been reached. It must be remembered that there was no detailed alteration to the general specification which remained committed to the dry attachment of the risers to the pontoons. That is consistent with the language of the deviation to which I have drawn attention. So even if there could have been agreement on use of the central caisson for attachment of some of the risers, there was still basic agreement that the pontoons needed to be capable of taking the dry attachment of the remaining risers.

That is my construction, and that is reflected in what happened (but I do not say that such subsequent events assists the exercise of construction). Thus Petrobras gives the following account of what happened (there are no findings by the judge). The central caisson concept was further considered by the parties’ consultants, but by October 1996 problems had been encountered. On 26 February and again on 13 March 1997 Petrobras reminded Petromec that a detailed study for the adoption of the concept of central caisson attachment had not been submitted to Petrobras for approval. By this time, however, it was becoming clear that the possibility of attaching all the risers to the central caisson had been abandoned. Thereafter no viable study or arrangement was ever commissioned or presented to Petrobras. In the meantime Petrobras was continuing to present its need for dry attachment. A weekly progress meeting report of 2 June 1997 stated, under the heading of “Riser Support Structures”, that a spider deck was being considered. In the end that was adopted even though 6 of the 89 risers concerned would need to be wet attached to the pontoons rather than the spider deck, but that was because there was a requirement in the specification that “gas and oil lines shall not pass under accommodation areas and also under areas where the emergency and fire pumps are located”.

In sum, I would answer issue 3.2 by saying that there was no contractual commitment to the use of the central caisson for riser attachment, and that as a result Petrobras could not be required, on the basis of such a commitment, to agree to wet attachment of the 42 risers.

I make no finding or holding as to how the incidence of the cost of actual attachment of the risers would fall: but I note that Petrobras submits that it falls on Petrobras, and Petromec concedes that it does. On that basis, it seems to me that it becomes even harder to support the judge’s conclusion that Petrobras would have been contractually bound to accept wet attachment.

In these circumstances, although it seems probable that Petrobras would never have in fact agreed to wet attachment of nearly half the risers, and that the parties would instead have found the solution of a spider deck even if the platform had been upgraded for South Marlim, I do not formally so find. But as to the latter part I have at para [75] above recorded Mr Vineall’s concession that that was the probability of events.

As for the judge’s holding that the concept of a spider deck remains wholly outside the Original Specification which is referred to in clause 12.1 of the Supervision Agreement, I am prepared to agree: on the basis that the definitive time for deciding is 19 December 1996. That is how the matter was argued before the judge and before us. It is therefore difficult to see how consideration of a spider deck comes into account – save as a matter of mitigation. In that respect, it follows from my construction of the central caisson riser deviation that, since there was no commitment in terms of that deviation or concept, Petromec’s obligation remained that of accommodating dry attachment of the risers on the pontoons. On that basis, Petrobras would be entitled prima facie to calculate the cost to Petromec of compliance with that obligation as part of the hypothetical South Marlim upgrade. Nevertheless, Mr Hancock told the court that he does not seek to do that, but rather to concede the lesser cost to Petromec of building a spider deck. However, we have not been asked to make any such relevant calculations of comparative cost, and we cannot and do not do so. That must therefore remain for future consideration. But I record Mr Hancock’s proffered concession.

In the circumstances, it seems to me that issue 3.2 should be answered as follows:

“The contention of neither party is adopted. However, Petrobras is correct to submit that it was not committed contractually to the adoption of the central caisson for the attachment of the risers. In the circumstances, the contractual requirement of a design permitting dry attachment of the risers remained, but it is recorded as being common ground that the probable outcome would have been the solution of building a spider deck.”

If the parties wish to suggest that any different answer should be entered into an order in the light of this judgment, they are at liberty to make submissions accordingly.

Issue 2 asked: “Would the upgrade for South Marlim have required the same or similar deck space requirements to that for Roncador, including an aft deck extension and the relocation of the transverse piperack to a more forward location?”

Issue 2 is contingent on issues 1 and 3 because both have significant knock-on effects on the amount of deck space that was required. The judge determined issue 2 on the basis of his determination of issues 1 and 3 and he made no alternative findings in the event that he was wrong. However, it was common ground that if issues 1 and 3 were determined in Petrobras’s favour, the required aft deck extension would have been different and significantly larger than 600 m³. The parties’ experts however differed as between a figure of 809 m³ (Mr Fogg, Petrobras’s expert) and 712 m³ (Mr Uttley, Petromec’s expert).

It is also common ground that, if Petrobras were to succeed on issues 1 and 3, as by and large in my judgment it does, it would not be possible or sensible for this court to resolve that dispute. In the circumstances, the parties are agreed that the court should simply set aside the judge’s answer to issue 2 (para 1(2) of his order refers).

Issue 4 asked: “To what extent would the existing cooling arrangements have required upgrading for South Marlim? How did that differ from the cooling arrangements required for Roncador?”

It is common ground that issue 4 is contingent on issue 1. The parties have agreed that, If Petrobras’s appeal on issue 1 succeeded, para 1(4) of the judge’s orders should be set aside and substituted with:

“The upgrade for South Marlim in accordance with the Original Specification would have required two new semi-submersible seawater lift pumps (with a capacity of 1,550 m³/h) together with a total of nine new discharge coolers.”

I would so order.

In sum, I would allow Petrobras’s appeal in accordance with this judgment.

I am grateful to Rix L.J. for his comprehensive description of the circumstances which have given rise to the present appeal and the issues that arise for decision. I agree with his conclusions, but since we are differing from the learned judge I think it right to explain briefly the reasons for my decision.

Issue 1 concerns the construction of paragraph 3.1.2 of Annex X which deals with the provision of compressors. It provided as follows:

“P.3.1.2: BRASOIL re-confirmed the compression requirements as stated in the specification.

The gas compression system will consist of the following equipment: 1 off new gas compression train with a capacity of 2,000,000 Nm³/d, at 20ºC/101,3 kPa abs.; and 1 off existing HP and Export Gas Compressor, as supplied by Delaval Stork driven by an EGT RLM 1600 gas turbine.”

Annex X constituted a variation of the General Technical Specification which was intended to form part of the contract under which Petromec undertook to upgrade the rig for use as a production platform in the South Marlim oilfield. The specification therefore provides a vital part of the context in which Annex X is to be construed. It was common ground that paragraph 3.1.2 was intended to operate as a mechanism for allocating part of the cost of the work relating to the supply and installation of compressors, but, notwithstanding Mr. Vineall’s attractive arguments, I do not think that it can have been intended to vary the work that Petromec was actually to carry out or the specification to which the platform was to be built. Petrobras had made it clear throughout that for operational reasons the platform needed to have the capacity to compress 6 million Nm³ per day of gas (allowing 50% spare capacity for equipment failures and maintenance), to be provided by three compression trains each with a capacity of 2 million Nm³ a day. The specification was drawn up for the practical purpose of constructing a platform which met all the various operational requirements, including gas processing capacity. It was not, as Rix L.J. put it in his paragraph [44], a fantasy specification, but one that was intended to determine the physical properties of the platform and Annex X was to be read as part of it.

It was common ground that the compressor installed as part of the rig’s original equipment was not capable of providing any useful service in the South Marlim field. For practical purposes, therefore, it had to be discounted. It follows that if the specification was to be met, three new compressors with the necessary ancillary equipment had to be purchased and installed, even if that meant extending the area of the deck, and as matters then stood the parties envisaged that Petromec would undertake an obligation to design and produce a platform that met those particular aspects of the specification. All that must have been fully apparent to both parties.

That is the context in which paragraph 3.1.2 is to be construed. I find it striking that it begins with a re-confirmation of the compression requirements set out in the specification – i.e. a total capacity of 6 million Nm³ of gas per day – and the fact that it is expressed as a free-standing sentence lends further support to the view that it was intended to state an overriding requirement. It therefore provides the context in which the second sentence must be understood. In my view it cannot properly be subordinated to the second sentence, let alone disregarded as being inconsistent with it. In that context the second sentence can, and in my view should, be read not as prescribing what equipment is to be on board on completion of the upgrade, but as defining the scope of Petromec’s obligation in relation to the supply of equipment. As such it cuts down Petromec’s obligation under the specification to provide three compression trains each with a capacity of 2 million Nm³ per day. It may also impose on Petrobras the costs of removing the original compressor (subject to credit for any residual value it may have), although it is unnecessary to decide that point. However, since nothing is said about how the required compressors are to be accommodated on the platform, it cannot be read as cutting down the scope of Petromec’s obligation to carry out the design and construction work required to comply with the specification, including the provision of ancillary equipment and any necessary extension of the deck.

The judge held that the effect of the second sentence of deviation P3.1.2 was to allocate to Petrobras the whole of the cost that flowed from the need to purchase and install two additional compression trains, including the costs of providing additional deck space and gas processing equipment, but, with respect, I think he was wrong about that. To construe the deviation in that way is effectively to relieve Petromec of the burden of designing and constructing a platform with the gas compression capacity that Petrobras required. That could only be explained by treating Annex X as providing a conventional basis for the pricing of the upgrade, but for the reasons I have already given I do not think that it can properly be understood in that sense. In short, the language of paragraph 3.1.2 is not capable of bearing the weight that Petromec seeks to put on it. I accept that the background to this paragraph of Annex X supports the conclusion that its purpose was to allocate to Petrobras part of the cost of the upgrade that would otherwise fall on Petromec, but by itself that tells one very little. The debate is all about identifying the elements of the cost to which this paragraph relates and nothing in the background lends assistance on that question. If one starts from the position that the entire cost of purchasing and installing the compressors was intended to be borne by Petromec, clear words are required to relieve it of any particular part of those costs. I do not think that the language of the deviation is sufficiently clear to achieve that, otherwise than in relation to the purchase of two additional compression trains.

For these reasons, as well as those given by Rix L.J., I too would allow the appeal on this issue.

As to issue 3 (the attachment of the risers), there is little I can usefully add to what Rix L.J. has said. For the reasons he gives I think it is clear that although Petrobras agreed that Petromec could develop a proposal for using the central caisson for the attachment of some, if not all, of the risers, it did not commit itself to accepting any such proposal together with whatever consequential amendments to the specification it might reasonably entail. The relevant section of Annex X states that Petrobras accepted the concept of using the central caisson and that Petromec could develop a study for discussion with Petrobras. In my view the thrust of the agreement is clear: Petrobras did not reject the idea outright; rather, it gave Petromec the opportunity of putting forward a proposal for its consideration. In the event it proved impossible to accommodate all the risers within the central caisson, which meant that some would have to be attached to other parts of the rig. Petromec proposed that they should be wet-attached, but Petrobras did not favour wet-attachment for what were, in my view, sound practical and commercial reasons. However, whether its position was reasonable or not, it could not be required to depart from the specification against its will and the documents on which Petromec relies do not support the conclusion that it did so.

There is nothing I wish to add in relation to issues 2 and 4. I therefore agree that the appeal should be allowed on the basis proposed by Rix L.J.

I agree with both judgments.