Public Service Company of Indiana, Inc., and Riley Stoker Corporation, Cross-Appellees v. Bath Iron Works Corporation, Cross-Appellant

CUDAHY, Circuit Judge. This case concerns the August 1980 failure of a weld on a piece of equipment known as a “conical head,” which was designed by Riley Stoker Corporation (“Riley”), fabricated by Bath Iron Works Corporation (“Bath”) and purchased and installed by Public Service Company of Indiana, Inc. (“PSI”). Defendant Bath appeals a jury verdict against it and in favor of plaintiffs PSI and Riley, primarily on the basis that the trial court erroneously excluded evidence material to its defense. The parties also raise various claims of error relating to damages and prejudgment interest. We agree that Bath’s proffered evidence was improperly excluded, and reverse. I. The conical head that is the subject of this suit is a component part of a coal pulverizing mill located at PSI’s Gallagher Station in New Albany, Indiana. Each mill is a 13 foot cylinder, 9 feet in diameter, which is rotated by a shaft fitted into a hub at one end (the head) of the barrel. That head can be either conical or flat. As the cylinder rotates, steel balls crush the coal, which is then blown into the duct work attached to a boiler. PSI has twelve such mills at Gallagher Station. Six of those mills, connected with generating units 3 and 4, have been operating successfully since the late 1950’s with conical heads, which were designed by Riley and fabricated by Riley’s sister corporation, Riley Corn-wells. The other six mills were originally built with flat heads. Because the mills built with flat heads had proven inferior to those with conical heads, PSI decided in 1978 to replace the flat heads with conical heads, and requested a price quotation from Riley. Riley located the plans that had been used in the 1950’s to fabricate the conical heads (the “1950’s plans”) and sent these to Bath, which prepared a bid for the work. Riley then sent a quote to PSI, which ordered two conical heads from Riley, which Riley in turn ordered from Bath. Bath subsequently said it could not fabricate the heads on the basis of the 1950’s plans because those plans lacked sufficient detail. Riley then drafted new plans (the “1979 plans”) based on the 1950’s plans, and sent them to Bath. The 1979 plans included dimensional detail and instructional notes for Bath’s use. Later, PSI ordered *786 two more conical heads from Riley, which Riley in turn ordered from Bath. After a delay of about six months, Bath completed manufacture of the first two conical heads in October 1979, and after inspection by Riley and PSI representatives, one of the heads (“head no. 1”) was installed by PSI in its coal pulverizing mill IB. That head failed in service in August 1980, as the result of a 360° circular crack through the plate of the cone at the toe of the weld connecting the cone to the hub of the head. That failure is the subject of PSI’s claim against Bath. After the failure, both PSI and Riley hired experts to determine its cause, and in conjunction with that investigation the 1950’s heads were examined. The experts apparently concluded that the cause of the failure was insufficient weld metal and improper finish of the weld at the cone-to-hub connection. At trial, Bath conceded that the size of the weld was smaller than the specifications on the 1979 plans required. The question whether the weld was improperly finished was disputed. Over plaintiffs’ objection, the trial court admitted certain evidence on the issue of causation. In this connection, the investigation of the failure of the conical head installed in mill IB included an analysis of the three other conical heads fabricated by Bath which had not yet been installed. Riley’s expert, Teledyne, concluded that it would be possible to repair the three remaining heads by building up and smoothly blending the weld in accordance with Riley’s design. Riley prepared a written rebuild procedure, approved by PSI, which specifically required that the weld should be finished by grinding. Bath repaired one of the heads (“head no. 2”), and PSI and Riley retained an outside firm to repair another one (“head no. 3”). Riley’s claim against Bath is for the expenses it incurred in having head no. 3 repaired. Riley and PSI inspected and approved both rebuilt heads, and these were installed. Both rebuilt heads failed in service, although these failures apparently were not as severe as the first one had been, and the heads were still in operation at the time of trial. The fourth head was returned to Bath. PSI brought claims against Bath in negligence and strict liability and sought various items of damages resulting from the failure of the first conical head. Riley’s ease against Bath was founded solely on breach of contract, and Bath counterclaimed against Riley to recover monies due for the other heads. Bath’s affirmative defenses included incurred risk, contributory negligence, failure to mitigate damages, breach of contract, estoppel, waiver and inspection and acceptance. The action was bifurcated by the trial court and tried before a jury, which found in favor of PSI and Riley. After the damages phase of the trial, the jury awarded PSI $2,438,-080.20 in damages and awarded damages to Riley in the amount of $70,225.09. The trial court awarded PSI $83,205.46 in prejudgment interest, but denied Riley prejudgment interest. Facts relating to the parties’ contentions on damages and interest will be set forth later in this opinion where relevant. II. Stated simply, the plaintiffs’ theory at trial was that Bath’s failure to comply with the plans provided by Riley resulted in defects in the welds in the conical heads. Although PSI’s claims sounded in tort, and Riley’s in contract, both parties’ essential contention was that Bath’s deviation from the Riley plans was the sole cause of the failure of the heads. In particular, plaintiffs presented a great deal of evidence tending to show that Riley’s 1979 plans required leg lengths and a type of blended finish which, if followed, would have produced a satisfactory product, but that Bath failed to follow those plans. Evidence was presented indicating that for various reasons Bath was under pressure to complete the project quickly and, as the result of that pressure, did not take necessary steps to complete the head correctly. As noted, Bath acknowledged that it had made the cone-to-hub weld too small, thus deviating from the 1979 plans. It claimed that the *787 1979 plans had only specified an “as-welded finish,” rather than a blended finish, however, and that the as-welded finish was the actual cause of the failure. Bath urged this alternative theory of causation as a complete defense to both PSI’s tort and Riley’s contract claims. In order to prove this theory, Bath intended to demonstrate the following propositions. First, Bath wanted to show that PSI’s and Riley’s objective in 1979 was to design and have manufactured conical heads that would be identical to the conical heads that Riley had designed for PSI in the 1950’s, which had operated successfully since that time. Bath then wanted to present evidence about the 1950’s design and the 1950’s heads, to show that Riley’s 1979 plans did not in fact duplicate the 1950’s heads. Next, Bath intended to show that the heads thát were rebuilt (one by Bath, and one by another manufacturer) to conform to the 1979 plans both failed at the same critical weld, despite the fact that once rebuilt they complied with the specifications in the 1979 plans, as well as the specifications of the rebuild procedure. Once again, Bath wanted to produce evidence about the 1950’s heads in order to demonstrate the “design deficiencies” in the subsequent 1979 plans and rebuild pro cedure — i.e., that the later plans did not provide the correct specifications necessary to duplicate the 1950’s heads. Bath was allowed to demonstrate several of these propositions, but was foreclosed at various points from providing evidence about certain features of the 1950’s heads, primarily because the trial court viewed this evidence as irrelevant. Without such evidence, Bath contends, the jury was left with the erroneous impression that the 1979 plans did (at least more or less) duplicate the 1950’s plans, and so was bound to reject Bath’s theory that the differences between the plans, rather than Bath’s deviation from the 1979 plans, actually caused the heads to fail. Moreover, Bath argues that it was prejudiced by the trial court’s failure to make it clear that evidence about the 1950’s heads would not be admissible until midway through trial (even though it had allowed Bath to present its theories about this evidence during opening argument) thus unfairly foiling Bath’s trial strategy. Finally, Bath contends that the trial court made many inconsistent rulings regarding the 1950’s heads evidence, effectively allowing plaintiffs to present their version on the point but preventing Bath from contradicting it. The next type of evidence that Bath contends was erroneously excluded concerned events that took place after the second and third heads had been rebuilt, and had failed in service. Following the second and third failures, Riley again conducted a field inspection of the 1950’s heads. Based on measurements made during the field inspection, PSI and Riley prepared new plans for the conical heads (the “1983 plans”). Bath’s position was that the 1983 plans finally achieved the plaintiffs’ initial intention to duplicate the 1950’s heads. Bath wanted to use this evidence to argue that PSI and Riley recognized the real cause of the 1979 failure and, in 1983, finally corrected the design deficiencies that had been present in the 1979 plans. Bath believed that such evidence would support its theory that, had Riley prepared adequate plans in 1979, Bath would have produced heads that were identical to the 1950’s heads and the failure at issue would never have occurred. The trial court granted plaintiffs’ pre-trial motion in limine to exclude this evidence about the 1983 plans. The trial court excluded both the evidence about the 1950’s heads and the evidence about the 1983 plans primarily on the basis that this evidence was irrelevant, and also because in its view such evidence would have led to confusing the issues and misleading the jury. Because decisions regarding the admission and exclusion of evidence are peculiarly within the competence of the district court, we will not reverse such a determination on appeal unless it constituted a clear abuse of discretion. See United States v. Micklus, 581 F.2d 612, 617 (7th Cir.1978). Moreover, Bath has the burden on this appeal to show that its substantial rights have been prejudiced by *788 the exclusion of the proffered evidence. See Ellis v. City of Chicago, 667 F.2d 606, 611 (7th Cir.1981). Under these standards, we evaluate each of Bath’s arguments in turn. III. A. Evidence about the 1950’s design and finished heads Prior to trial, plaintiffs made a motion in limine to exclude evidence pertaining to three issues: the rebuild procedure, the failure of the rebuilt heads and the 1983 plans. The motion did not request the exclusion of evidence about the 1950’s design or the actual features of the 1950’s heads. In opening arguments, both PSI and Riley referred briefly to the existence of the 1950’s conical heads. These were also mentioned in a statement of undisputed facts which the trial judge read to the jury. Bath’s opening argument described in detail what Bath intended to prove with respect to the 1950’s design, stressing that plaintiffs’ objective in preparing the 1979 plans was to duplicate the 1950’s heads, but that neither plaintiff had actually examined the weld on the 1950’s heads prior to preparing the 1979 plans. Also, Bath described the testimony Bath intended to introduce regarding the size, contour and finish of the 1950’s weld, and how Bath would tie that testimony into its defense that the 1979 failure resulted from improper design. Plaintiffs did not object to this argument. On the second day of trial, the district court ruled on plaintiffs’ motion in limine. It decided that Bath could present evidence of events up to and including the rebuild procedure and the failure of the rebuilt heads, but could not go beyond that point to show the 1983 plans. Thereafter, as part of Riley’s case-in-chief, a metallurgical engineer employed by Riley, Robert Hallst-rom, testified. Mr. Hallstrom had taken measurements of the 1950’s heads for the purpose of preparing the 1983 plans, but this was not disclosed on direct. On cross-examination, Bath attempted to elicit testimony from Mr. Hallstrom about whether anyone from Riley had examined the dimensions of the 1950’s heads, but Riley objected to this line of questioning, on the basis that the measurements had been made in conjunction with preparation of the 1983 plans and therefore were excluded by the ruling on the motion in limine. Mr. Hallstrom was allowed to testify that Riley had not made any inspection of the 1950’s heads prior to the failure of the rebuilt heads. An exhibit offered by Bath containing the measurements taken by Hallstrom was not admitted. Riley then offered the testimony of William Dobson, an expert witness, who testified about the 1950’s heads on both Riley’s direct and PSI’s cross-examination. Bath also cross-examined Mr. Dobson about the measurements of the 1950’s heads, even though he stated that he had obtained these measurements from Mr. Hallstrom in 1983. Bath contends that Mr. Dobson’s testimony about the radius and leg length of the 1950’s weld was erroneous, and that it therefore attempted once again to introduce Mr. Hallstrom’s measurements to impeach Mr. Dobson’s testimony. The trial court sustained Riley’s objection that testimony about the measurements were outside the scope of direct examination. Next came PSI’s direct case. Several of PSPs witnesses testified about the 1950’s design and described the 1950’s heads. For example, Gerald Piskorowski, a PSI service manager, testified about the heads on direct examination. The court also allowed Bath to pursue this subject during cross-examination, during which Mr. Pisko-rowski described the leg length and radius of the 1950’s weld. Bath contends again that this testimony gave the erroneous impression that the 1950’s weld was very similar to the weld depicted on the 1979 plans provided to Bath. Next, PSI put on two expert witnesses, who both testified about the 1950’s weld. Dr. Caulfield testified that the fatigue life of the 1950’s weld was much longer than the fatigue life of the weld called for in the 1979 plans, due to the radii and profiling of the 1950’s weld. Dr. Lawrence also testified about the *789 1950’s head, and stated that it had a longer expected life than the head that failed in 1980. On cross-examination of these expert witnesses, Bath asked each witness about how the 1950’s heads had been ground, drawing an objection each time on the basis of relevance. Surprisingly, the court sustained the objection during cross-examination of Dr. Caulfield, but overruled what appears to be precisely the same objection during cross-examination of Dr. Lawrence. The court also sustained a relevance objection with regard to Bath’s question on cross-examination of Dr. Lawrence about whether the life of the 1950’s weld was greater than the life of the weld depicted on the 1979 plans. Although Dr. Caulfield had testified directly about this issue, when it came up again on cross-examination of Dr. Lawrence, the trial court stated it saw no relevance to any comparison between the 1950’s head and the 1979 drawing. Then, during the presentation of Bath’s defense case, the deposition of Francis Ouellette, a Riley employee, was read to the jury. Mr. Ouellette’s testimony was that Riley’s objective in preparing the 1979 plans was to produce conical heads that would be exactly like the 1950’s heads; that in order to accomplish that objective, Riley would have to examine the 1950’s heads, but did not do so; that an important instructional note regarding grinding had appeared on the 1950’s plans but was absent from the 1979 plans; and that many instructional notes were absent from the 1950’s plans because Riley employees, who Mr. Ouellette believed had fabricated the 1950’s welds, knew how to make the welds. Bath next called its chief engineer, Donald Smith. He testified that he knew about the 1950’s heads, and that he had examined and measured the weld on the 1950’s heads.' However, when Bath attempted to elicit testimony about that examination, the trial court sustained an objection, stating that such testimony would be immaterial. Because of this and prior similar rulings, Bath believed it was not permitted to ask questions of its expert witness, Dr. Maddox, concerning the dimensions and finish of the 1950’s heads. It did attempt to elicit this testimony during re-direct examination of Dr. Maddox, but was met by the trial court’s ruling that such questions were outside the scope. We think it is unnecessary to discuss individually the propriety of each ruling attacked by Bath, but we agree with Bath that the trial court committed prejudicial error by refusing to allow Bath at various points to provide testimony about the 1950’s design and finished heads. Throughout the trial, the adequacy of the 1979 design was at issue. A great deal of evidence was presented that the objective in preparing the 1979 design was to duplicate the successful 1950’s heads. Thus, evidence about the 1950’s plans was clearly relevant to the issue whether the 1979 plans were adequate to accomplish their objective. PSI apparently concedes this point, but argues that testimony about the actual finished 1950’s heads (as distinct from the 1950’s plans) was irrelevant because there was no evidence indicating that the 1950’s heads were different from what was indicated on the 1950’s plans. We disagree. Bath presented undisputed testimony by Mr. Ouellette, a Riley employee, that the 1950’s plans lacked certain instructional notes and apparently for this reason the person who prepared the 1979 plans found it necessary to add such notes and further detail in order to provide plans that would actually produce heads that were the same as the 1950’s heads. This was quite adequate foundation to support Bath’s contentions that the features of the finished 1950’s heads were relevant to the adequacy of the 1979 design and that Riley should have examined the 1950’s heads in order to ensure that the updated 1979 plans would produce heads matching those made in the 1950’s. In addition, we note that it is somewhat disingenuous for plaintiffs to argue that evidence about the characteristics of the 1950’s heads was irrelevant, when two of plaintiffs’ witnesses testified about those characteristics. In this regard, PSI contends that evidence about the 1950’s heads was a relatively minor matter, and so *790 Bath should not have been permitted to present contrary evidence merely to impeach plaintiffs’ witnesses. However, as we have stated, the characteristics of the 1950’s heads were central to Bath’s defense theory that it was Riley’s failure to provide plans in 1979 sufficient to produce those successful heads, and not Bath’s deviation from the 1979 plans, that caused the head to fail in 1980. Put simply, the trial court’s rulings resulted in a situation in which plaintiffs were able to present testimony about the 1950’s design and heads, but Bath was prevented from presenting its version of that issue in any cohesive and complete fashion. Many of PSI’s arguments regarding the relevance of the 1950’s evidence to the causation issue consist of a rehash of plaintiffs’ evidence that the cause of the 1980 failure was Bath’s shoddy work, not Riley’s inadequate plans. Were the issue before us the sufficiency of plaintiffs’ evidence on causation, these arguments would be significant. But, as Bath points out, the weight of the evidence appears as it does only because the “evidentiary counterweights were excluded.” Bath’s Reply Br. at 6. Thus we conclude that the trial court abused its discretion in refusing to allow Bath to fully present its evidence on the 1950’s heads. We also conclude that the erroneous exclusion of this evidence prejudiced Bath’s substantial rights. While it is true that Bath was able to present through argument and some testimony its theories as to causation, at many points it was prevented from presenting the facts about the 1950’s heads that would have tended to support those theories. This distinction may or may not have been central to the jury’s verdict; we will not speculate on this issue since it is clear that, at the very least, the trial court’s inconsistent and to some extent belated rulings precluded Bath from presenting its defense in an organized and complete fashion. 1 B. Evidence about the 1983 plans As previously noted, the trial court granted plaintiffs’ motion in limine to exclude evidence of the 1983 plans, which the plaintiffs prepared after the failure of the rebuilt conical heads. It is Bath’s contention that plaintiffs finally achieved in the 1983 plans what they had intended to achieve in the 1979 plans, i.e., duplication of the 1950’s heads, by providing detailed grinding instructions in the 1983 plans that were absent in the 1979 plans. Bath argued that the 1983 plans were admissible: 1) to attack the credibility of plaintiffs’ witnesses who testified that Bath should have known how to grind the critical weld by examining the 1979 plans, since plaintiffs thought it necessary to include grinding instructions on the 1983 plans; 2) to support the credibility of the witnesses who testified that the 1979 plans did not require grinding in the absence of a specific grinding instruction; 3) to show that the 1979 plans did not duplicate the 1950’s heads; and 4) to show that the plaintiffs reeog- *791 nized the inadequacy of the 1979 design and therefore corrected the problem in the 1983 plans. 2 All of Bath’s suggested uses of the 1983 plans evidence really boil down to the proposition that, since plaintiffs included certain details in their 1983 plans that were (arguably) absent from the 1979 plans, and since both sets of plans were designed to produce the same product, the 1979 plans must have been inadequate. The issue of what finishing instructions were included in the 1979 plans, and how those instructions should have been interpreted by Bath, was sharply contested at trial. Plaintiffs’ witnesses testified that Bath should have known how to grind the welds from the 1979 plans; Bath contended that the absence of more detailed grinding instructions was a design failure attributable to Riley. Relevant evidence is evidence "having any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence.” Fed.R.Evid. 401. Applying this standard, evidence that the 1983 plans prepared by plaintiffs (which were designed to produce the same heads as the 1979 plans) did include more detailed grinding instructions certainly appears to make less probable plaintiffs’ contention that Bath should have known how to grind the heads from the 1979 plans without such instructions. Of course, plaintiffs may simply have decided to err on the side of caution in preparing the 1983 plans, given Bath’s previous failure to follow what plaintiffs thought were clear instructions in the 1979 plans. But the fact that different inferences may be drawn from the evidence does not make the evidence irrelevant. Plaintiffs also suggest that, even if the evidence about the 1983 plans was relevant, it should have been excluded under Fed.R.Evid. 407, which makes evidence of “subsequent remedial measures” inadmissible under certain circumstances. 3 That rule provides: When, after an event, measures are taken which, if taken previously, would have made the event less likely to occur, evidence of the subsequent measures is not admissible to prove negligence or culpable conduct in connection with the event. This rule does not require the exclusion of evidence of subsequent measures when offered for another purpose, such as proving ownership, control, or feasibility of precautionary measures, if controverted, or impeachment. The primary ground for excluding evidence of subsequent remedial measures to prove negligence or culpable conduct is the “social policy of encouraging people to take, or at least not discouraging them from taking, steps in furtherance of added safety.” See Advisory Committee Notes to Proposed Rule 407. Bath notes plaintiffs’ failure to cite cases applying this rule in a situation where the evidence concerns a plaintiffs subsequent measures, and where the evidence is offered by a defendant to refute allegations that it was negligent. Bath thus implies that the rule may simply be inapplicable under such circumstances. Certainly many eases paraphrase the rule as excluding evidence of a defendant’s subsequent remedial measures, and the policy behind the rule arguably may not apply where a plaintiff’s activities after an event are at issue (since, for example, a party who is not in danger of being sued may be less likely to be deterred from *792 making subsequent improvements by the possibility that the improvements will be used against it). But the rule by its terms contains no such limitation, and our inability to find cases applying the rule to exclude evidence of plaintiffs’ subsequent remedial measures may have more to do with the fact that plaintiffs’ negligence is not as often at issue than with the inapplicability of the policy underlying the rule. But we need not decide whether Rule 407 can ever be invoked to exclude evidence of a plaintiff’s subsequent remedial measures to prove plaintiff’s negligence, since the rule by its terms allows evidence of subsequent measures when offered for purposes other than proving negligence or culpable conduct. The rule does not require exclusion of subsequent measures when offered for impeachment. This exception must be applied with care, since “any evidence of subsequent remedial measures might be thought to contradict and so in a sense impeach [a party’s] testimony that he was using due care at the time of the accident, [and] if this counted as ‘impeachment’ the exception would swallow the rule.” Flaminio v. Honda Motor Co., Ltd., 733 F.2d 463, 468 (7th Cir.1984). For this reason, (assuming Rule 407 is applicable to plaintiffs' subsequent measures) we would be reluctant to allow Bath to present evidence about the 1983 plans merely to show that the 1979 plans were inadequate because they did not include some arguably better feature that was added in the 1983 plans. Nor do we believe such evidence should be used to show that plaintiffs “recognized the inadequacy of the 1979 plans,” in the sense that this implies an admission by plaintiffs on a crucial issue in the case. Such uses may well be inconsistent with the tenor of Rule 407. But Bath’s evidence that both sets of plans were prepared as part of a continuing effort to reproduce the 1950’s heads cannot be ignored. Plaintiffs’ evidence that the 1979 plans were adequate to achieve this objective included testimony that Bath should have known how to grind the welds properly by examining those plans, and that more precise grinding instructions were unnecessary. We think that Bath’s proposed testimony that plaintiffs nevertheless found it advisable to include such instructions for the use of Riley’s fabricators properly could be used to impeach plaintiffs’ witnesses on this point. Thus, we conclude that Rule 407 does not operate to preclude Bath’s evidence about the 1983 plans if used for that limited purpose. 4 *793 Plaintiffs also argue that evidence of the 1983 plans should be excluded under Rule 403 because of the danger of undue prejudice and confusion of the issues. We disagree. This is admittedly a confusing case, but given that the jury will have heard evidence as far back as the 1950’s heads and as far forward as the failure of the rebuilt heads, we do not think that evidence of the 1983 plans will particularly cause confusion. Nor do we think that evidence, while perhaps unfavorable to plaintiffs, will result in “undue” prejudice to them. As noted, the 1983 evidence is probative to impeach testimony presented by plaintiffs about the adequacy of the 1979 plans and we conclude that it should be admitted on retrial. IV. Bath also attacks the method by which PSI’s damages were calculated. Throughout the damages trial, PSI told the jury it was seeking, and presented evidence to support, three distinct and cumulative elements of damages: 1) expenses associated with repair and replacement of the faulty conical head, 2) increased operating expenses and 3) loss of use. Bath does not dispute the first element of damages, but contends among other things that allowing PSI to recover both the second and the third element constitutes a double recovery. In the circumstances which appear to apply here, we think this contention is generally correct. Because the cost incurred by an electric utility in generating power varies as among its different power plants in accordance with their differing capital costs 5 and differing operating or running costs (reflecting, primarily, the cost of fuel), such a utility will generally try to dispatch first the available plant having the lowest cost of fuel. In the hierarchy of plants having the lowest to the highest operating costs, Gallagher Station — a so-called base load plant — had one of the lowest operating costs per kilowatt-hour. When the capacity of Gallagher Station was diminished by the loss of the coal pulverizer, PSI apparently had to dispatch plants having higher operating costs to meet its demand. Thus, PSPs claim for increased operating expenses presumably reflects the fact that it had to replace the electricity that otherwise would have been produced by Gallagher Station at low fuel cost with electricity costing more to produce at other plants. Calculation of these increased costs is based in part on a “derate,” which represents the loss of capacity of the boiler at Gallagher Station resulting from the lack of one coal pulverizer because of the failure of the conical head. This lost capacity, under the mode of calculation we have de- ' scribed, must be made up from higher operating-cost sources. Loss of use, on the other hand, involves a somewhat different calculation based on the value of use of the damaged item lost during the period of its unavailability. Loss of use may be measured by the fair rental value of the property. See, e.g., Farm Bureau Mutual Insurance Co. v. Dercach, Ind.App., 450 N.E.2d 537, 540 (1983); Maddox v. Yocum, 114 Ind.App. 390, 52 N.E.2d 636 (1944). In the case before us, PSI measured the ‘reasonable rental value’ of its lost property by first applying the derate to measure the amount of its lost generating capacity. The value of this lost capability was in turn measured by PSI’s “capacity charge,” which was the charge made by PSI to other utilities for committing to them capacity equivalent to the amount lost. 6 *794 The parties have made various (to us confusing) arguments about the appropriate measure of damages for loss of use, particularly with respect to what constitutes adequate proof of rental value. This confusion stems at least in part from the parties’ valiant attempts to conform statements in Indiana cases about rental value as applied to such things as land or automobiles to the peculiar circumstances of an electric utility which loses generating capacity. Rather than undertaking an extended discussion of the parties’ arguments (only to demonstrate the inappropriateness of a number of them), we believe the better approach is for us to set forth what we believe are the applicable general principles governing damages in these circumstances. We think these are fully consistent with Indiana law. If part of a power plant’s capacity (in kilowatts) is lost, one of two courses may be taken to replace it and thus restore the output of the system and the revenues generated by it. First, one may call into use less efficient and idle power plants to make up the lost kilowatts of capacity and kilowatt-hours of energy. The cost of doing this to the system is essentially the increased operating (essentially fuel) cost of the less efficient plants called into service. Under this “increased operating cost” method, there is generally no element of capital or capacity cost as such because the capital costs of the plants called into service are fixed or “sunk” and are not affected by whether or not these plants produce electricity or are idle, see note 5, supra. Another method of replacing a utility’s lost capacity is for it to contract with another utility to supply the necessary kilowatts (and kilowatt-hours). The cost of replacement by this method would generally involve payment of a capacity charge for kilowatts of capacity and an energy charge to cover the fuel costs of the purchased kilowatt-hours (and would be adjusted for fuel cost saved by the receiving utility). The charges by the supplying utility would usually be generally equivalent to corresponding charges the first utility would make to make its capacity and energy available to its neighbors. 7 Presumably, the inter-utility charges for borrowing or lending capacity (and associated energy charges) would fit into the “rental value”, scheme of loss of use. In general, however, it would be duplicative both to replace the lost capacity from one’s own less efficient plants and to borrow the same capacity from a neighboring utility. The damages methodology should provide for replacement of the lost capacity only once— either internally or externally or from both sources in some appropriate proportion. There should not be an assumption, however, that the lost capacity can be wholly replaced by running one’s own less efficient plants (an operating cost calculation) and replaced again in full by borrowing capacity from a neighboring utility (a rental value calculation). As long as the capacity can be wholly replaced, it is generally not relevant whether the utility suffering loss of capacity has been merely meeting its own loads or has, in addition, been making off-system sales. The measure of damage is generally the additional cost of restoring the lost capacity, which will then meet the same system loads or will meet the same off-system demands as before. Revenues thereby *795 generated will be equivalent to those originally received. The appropriate facts may be, however, that the lost capacity was in fact committed in whole or part to off-system sales which were not made while the lost capacity was being restored. On those assumptions the measure of damages would in general be the revenue lost from the off-system sales (presumably based on inter-utility capacity and energy charges) less whatever fuel costs are saved by not having to run the plant which is damaged. 8 If PSI used this measure of damages, it could not, of course, also recover increased operating expenses or the cost of replacing capacity from the outside (because under these factual assumptions such costs would not be incurred). 9 There appears to have been a great deal of confusion about the relationship among the various elements of PSI’s damage claims at trial. As noted, throughout PSI’s arguments to the jury, it contended that it was entitled to recover both for increased operating expenses (fuel costs) and for loss of use, defined as rental value. It proposed an instruction to this effect. When counsel for Bath objected that such an instruction would appear to allow double recovery, the trial court seemed to agree, and gave an instruction, supposedly collapsing the two elements, which provided that in addition to repair costs PSI was also seeking “the additional cost incurred by PSI because it was required to use more expensive equipment or more expensive methods of generating electricity which includes the loss of the reasonable rental value of the equipment during the time it was not in service.” (Emphasis added.) But the court also instructed the jury that PSI was seeking increased operating costs and the loss of use of the pulverizer. In a separate instruction the court told the jury that “the proper measure of damages because of the deprivation of the use of property is the reasonable value of such use [which is] ordinarily measured by the property’s fair rental value.” On appeal, PSI adopts the position that it sought “damages for loss of use of the IB'pulverizer based upon a reasonable and fair rental value calculated from a capacity charge and increased operating expenses.” 10 PSI Br. at 41 (emphasis added). Given the instructions, PSI’s arguments to the jury and the size of the verdict it seems quite likely that the jury may well have believed it was instructed to award both increased operating expenses and an essentially duplicative amount for fair rental value. PSI notes that Bath did not object to the instructions eventually given. But this point is irrelevant given our conclusion that a new trial is required; new instructions will be given and they should be correct. 11 *796 V. Prejudgment Interest The parties also raise various issues pertaining to prejudgment interest. Given our conclusion that the case must be retried on both liability and damages, we need not discuss this issue extensively; however, since it will probably recur on retrial, we set forth the appropriate standards to be applied. Under Indiana law, prejudgment interest is proper where damages are ascertainable in accordance with fixed rules of evidence and accepted standards of valuation at the time damages accrue, but is not appropriate when the jury must use its judgment to assess an appropriate amount for past and future injury or there are elements not measurable by fixed standards of value. Simmons, Inc. v. Pinkerton’s, Inc., 762 F.2d 591, 607 (7th Cir.1985) and cases cited therein. A. Riley Riley presented evidence respecting its damages for breach of the fabrication contract with Bath indicating that its total damages were approximately $100,000. Of this amount, the majority (about $77,000) represents payments Riley made to another fabricating firm, Babcock & Wilcox, to repair the defective head no. 3. Approximately $2,500 was incurred for shipping the head to Babcock & Wilcox, and returning head no. 4 to Bath. Approximately $5,000 was paid for materials which Riley claimed were necessary for repair of the defective head. Finally, Riley paid about $2,400 for engineering expenses associated with the repair procedure. Except for the engineering expenses, Riley supported its claims with actual invoices for its expenses as well as testimony explaining the need for the expenditures and showing that the invoices had in fact been paid. The jury awarded Riley $70,225.09. The trial judge refused to allow Riley prejudgment interest on that amount, because of its conclusion that interest is not includable where damages are unliquidated and cannot be ascertained until judgment. In reaching this conclusion, the trial court apparently relied heavily on the fact that the jury award was substantially less than the amount Riley requested. But we think it is clear that under current Indiana law, damages do not have to be liquidated, but only ascertainable as of a particular time according to known standards of value. See Simmons, Inc. v. Pinkerton’s Inc., 762 F.2d at 607-08; Courtesy Enterprise, Inc. v. Richards Labs., Ind.App., 457 N.E.2d 572 (1983). The fact that an award differs from the amount of damages sought does not in and of itself indicate that prejudgment interest is inappropriate. See, e.g., Indiana Industries, Inc. v. Wedge Products, Inc., Ind. App., 430 N.E.2d 419 (1982). Prejudgment interest on damages may be appropriate even though the fact finder has to use some degree of judgment in measuring damages; the question in each case must turn on its facts depending on whether the jury appears to have been able to make a computation “according to known standards of value,” or whether its award was by necessity largely a matter of subjective judgment about the value of certain losses. See Simmons, Inc. v. Pinkerton’s Inc., 762 F.2d at 607-08. In resolving this question, factors such as the difference between the jury award and the amount sought, the nature and completeness of plaintiffs proof and the nature of the defendant’s proof on damages, if any, are all relevant. In this case, virtually all of Riley’s proof on damages consisted of invoices showing direct expenses that Riley in *797 curred as the result of its attempts to repair the conical head after it failed. Bath argues that the jury had to determine whether these amounts were reasonable and so had to use its subjective judgment, thus precluding prejudgment interest. But juries often have to make determinations about whether claimed damages are “reasonable,” and if this were determinative, prejudgment interest would rarely be awarded. Here Riley broke down for the jury what charges were incurred for what purpose, and showed the specific amounts that it had paid. Bath of course cross-examined Riley’s witnesses, but presented no evidence of its own to indicate that some or all items of damages were not properly attributable to the costs of repair or were unreasonable in amount. The jury obviously decided that some items were unreasonable for one reason or another. Nevertheless, under all the circumstances, we believe that Riley’s damages were ascertainable as of a particular time according to known standards of value. Therefore, on retrial, assuming the evidence remains substantially the same, Riley would be entitled to prejudgment interest on the damages the jury may assess. B. PSI The trial court originally denied PSI’s claim for prejudgment interest, but later determined that interest was appropriate on part of the jury’s award, equivalent to the amount of damages shown on one of PSI’s exhibits. That exhibit concerned the portion of PSPs damages claimed for expenses related to restoration of the failed conical head. The trial court stated that the exhibit (received without objection) “set forth a liquidated amount upon which, if only that amount had been awarded by the jury, prejudgment interest would have been appropriate.” Since the jury’s general verdict was substantially more than the amount set forth in the exhibit relating to repair costs, the court said it could only conclude that the repair costs amount was a part of the jury’s award, and so it ordered prejudgment interest on that portion of the award. We agree with Bath that it is impossible to determine from the jury’s general verdict whether the jury in fact awarded all or some of the amounts represented in the repair cost exhibit. The jury did not award PSI the entire amount of its claimed damages, and it is possible (though perhaps unlikely) that the portion the jury ‘deducted’ included the repair costs. However, given that the case has to be retried, we nevertheless find it appropriate to consider whether PSI could properly recover interest on the damages awarded for repair costs. On retrial the district court can make use of special verdicts to indicate what part of any award the jury makes to PSI represents repair expenditures, so that if interest is appropriate on this amount but not on other elements of PSPs damages, the court can order prejudgment interest on that part without speculating about the components of the jury’s award. We think that prejudgment interest would be appropriate on the amount of PSPs damages attributable to repair costs, under the same rationale which enabled us to conclude that Riley would be entitled to interest on damages awarded for its repair costs. PSPs claim for these costs was supported by detailed evidence of its expenses, and Bath did not present evidence disputing the validity or amount of these items. Indeed, on appeal it has not raised any issue relating to the appropriateness of PSPs repair cost damages. Presuming that the evidence on this part of PSPs claim remains substantially the same on retrial (and depending, of course, on the jury’s verdict) we conclude that prejudgment interest would be appropriate on PSPs repair cost damage claim. Finally, PSI cross-appeals, claiming that the court erred in refusing to award prejudgment interest on damages for increased operating expenses and loss of use. Given our conclusions that substantial errors occurred in presenting these claims to the jury, however, the evidence on retrial may well be substantially different from that which we are now asked to review. *798 Since we cannot know precisely what this evidence will entail (nor, of course, do we have the benefit of the trial court’s ruling on such evidence) we find it inappropriate to consider whether PSI would be entitled to prejudgment interest on nonrepair elements of damages at this time. 12 For the foregoing reasons, the case is Reversed and Remanded for a new trial to be conducted in accordance with the principles set forth in this opinion. Footnotes [1] PSI contends that Bath's proffered evidence, even if relevant to causation, was not relevant to other issues, such as PSI’s strict liability claim or Bath’s contributory negligence/failure to inspect defense. Assuming for the sake of argument that PSI is correct on these points, the appropriate solution is for PSI to request a limiting instruction on retrial as to the purpose for which the evidence may be considered, not exclusion of the evidence entirely. This same observation applies to virtually all of Riley’s arguments on appeal. Riley has argued that its claim against Bath is solely for breach of contract, not negligence and strict liability as is PSI's suit, and evidence about improper design is not a defense to a breach of contract claim. Since Bath admitted that it did not comply with the 1979 design specifications, Riley argues, Bath clearly breached its contract and that is the end of the matter. We leave it to the trial court in the first instance to determine whether or not Bath’s improper design evidence can constitute a proper defense to the breach of contract claim itself. However, assuming, as the trial court instructed the jury and Bath seems to concede, that it cannot, Bath’s evidence on causation at least is relevant to Riley's damage claim: clearly, a determination that Riley’s design, not Bath’s production deficiencies, actually caused the failures would be relevant to the amount of damages Riley could receive as compensation for Bath’s breach of contract. As with PSI, Riley can request a limiting instruction (as it did in this trial) on the appropriate purpose for which Bath’s evidence may be considered. [2] PSI contends that Bath waived the issue of the 1983 plans because it failed to make a timely offer of proof during the course of the trial. Bath on the other hand argues that it made its position and proposed evidence clear in various trial memoranda, and also actually offered the evidence through exhibits and Mr. Ouellette’s deposition summary. We think Bath sufficiently made its record on this point to enable us to consider the issue, and that Bath’s argument is not waived. [3] We apply federal rather than state law in determining the admissibility of subsequent remedial measures in a diversity case. Flaminio v. Honda Motor Corp., Ltd., 733 F.2d 463 (7th Cir.1984). But see Moe v. Avions Marcel Dassault-Breguet Aviation, 727 F.2d 917 (10th Cir.), cert denied, — U.S. -, 105 S.Ct. 176, 83 L.Ed.2d 110 (1984). [4] Bath also suggests that evidence of the 1983 plans is relevant to the causation issue, apparently on the theory that since, by examining the 1950’s heads, plaintiffs finally were able to successfully duplicate them in the 1983 plans, Riley’s failure to make such an examination before preparing the 1979 plans was the reason that those plans were unsuccessful. Bath cites the case of Quaker Oats Co. v. Davis, 33 Tenn. App. 373, 232 S.W.2d 282 (1950) to support this argument. In that case a chicken farmer claimed that her chickens died because the defendant supplied mislabeled feed bags. The defendant contended on the other hand that the mislabeled feed was not the cause of the chickens’ deaths, offering evidence that after the feed was no longer used, the chickens continued to die until after the farmer stopped washing the floors, and thereafter the farmer had no further untimely chicken deaths. The court held that the evidence regarding plaintiff’s actions (discontinuing floor-washing) was relevant among other things to show causation. Although Quaker Oats does concern the applicability of a rule barring evidence of subsequent remedial measures to a plaintiffs actions, the case does not support Bath’s argument. The evidence that was relevant to causation in Quaker Oats was the result of plaintiff’s subsequent actions — that chickens stopped dying when the farmer stopped floorwashing — not the action itself— that the farmer stopped floor-washing. Thus the evidence was not used to imply an admission on plaintiff's part (that she recognized the feed might not be the cause of the deaths) but to show the cause directly. Here, the only information we have is that plaintiffs changed their plans. We do not know whether that change was successful. For all we know the heads built from the 1983 plans also failed in operation, which would tend to suggest that even if the extra detail had been provided in the 1979 plans, the .head would have failed anyway. Thus, to make evidence of the 1983 plans relevant to causation, we would also need to know the result of the implementation of those plans, which would bring the evidence quite far afield from the real issue in the case, which only concerned the 1980 failure. To the extent that the plaintiffs are concerned that Bath may make improper use of the 1983 plans, they may re *793 quest and should be granted appropriate limiting instructions. [5] Capital or capacity costs have been incurred to build the plant, are fixed and "sunk” and are not affected by whether the plant is operating or left idle. [6] A sort of reciprocal “loss of use” measure is sometimes employed, wherein loss of use is calculated by the cost of renting a replacement item. At one point after trial the district court referred to the reasonable rental value as being made up in part by "the purchased power actually used by PSI,” which we take to be a reference to PSI’s cost to secure substitute capacity (i.e. to "rent” replacement capacity). Although there are some references in the record that might relate to a "replacement cost” theory of *794 rental value, the record as a whole and PSI's briefs on appeal suggest'that it was not seeking loss of use damages specifically on this theory, but rather on the theory of loss of use measured by the fdfr rental value of its property to another, and we treat its position as such. As a practical matter the fair value of renting the lost item to another would presumably be substantially equivalent to the cost of renting a replacement. In any event, a party could not recover both measures. [7] Of course, the goal is to arrive at a fair net rental value. See Jay Clutter Custom Digging v. English, 181 Ind.App. 603, 393 N.E.2d 230 (1979); Maddox v. Yocum, 114 Ind.App. 390, 52 N.E.2d 636 (1944). See also A.T. & T. Co. v. Connecticut Light and Power Co., 470 F.Supp. 105, 109 (D.Conn.1979) (Connecticut law). [8] PSI would have to demonstrate some basis for the assumption that it would or could make off-system sales from the lost generating capacity. The clearest case is where such sales were in fact being made before the outage. [9] A combination of damage measures might be appropriate, for example, if PSI replaced a portion of the lost capacity, but lost revenue from off-system sales it did not make because it did not replace all the capacity. But the point is that PSI can recover for any given unit of lost capacity only once — if it was replaced, by one measure of replacement cost, or if it was not, and would otherwise have been sold off-system, by the revenue lost less saved fuel costs. [10] Perhaps, PSI merely means that in calculating "rental value” derived from off-system sales, both a capacity and an energy charge might be involved. This would seem appropriate, if PSI demonstrated that an off-system sales measure was proper and that the "increased operating expenses” merely represented a reasonable energy charge. Our review of the briefs and record does not make this interpretation an obvious one. [11] Bath also contends that the district court abused its discretion in submitting PSI’s damage claims to the jury, since PSI’s proof of damages was too speculative to allow an award based on that proof to stand. PSI claims that Bath’s arguments as to sufficiency of the evidence on damages are waived. Given our conclusion that the damages issues will have to be retried, we do not find it necessary to discuss these claims in detail. Some of Bath’s arguments have already been dealt with in (or are mooted by) our discussion of the duplicativeness issue. It appears that most or all of Bath's other arguments do not have merit. For example, Bath argues that PSI’s evidence, presented through its expert, as to the appropriate calculation of the *796 derate was too speculative since PSI employed a uniform derate of 40 megawatts during the entire outage, rather than examining the actual diminution in output hour by hour (which apparently varied from 20 to 50 megawatts). But PSI presented evidence that this was an appropriate uniform derate. Bath did not attempt to calculate the actual hour by hour diminution, either, but merely averaged the varying losses of output to arrive at a uniform derate figure of 35 megawatts. The fact that the parties disagreed about the appropriate figure does not mean that PSI’s evidence was too speculative to go to the jury. [12] PSI and Riley contend that they are entitled to prejudgment interest on the separate ground that all parties stipulated that plaintiffs would be entitled to prejudgment interest in the event of a jury award against Bath. Although it is clear that at various points Bath took the position that prejudgment interest would be appropriate, and should be computed by the court in the event of an award against it, the trial court concluded that no stipulation to that effect had occurred. In the absence of such a stipulation, the court said, it simply had to make a decision about the appropriateness of prejudgment interest based on the law of Indiana, not statements previously made by the parties. The trial court was obviously in a much better position than we are to determine whether certain (in part unrecorded) statements by counsel constituted a stipulation regarding prejudgment interest. We find no error in its conclusion that such a stipulation did not in fact occur.