Tecon Corporation, Green Construction Company, and Winston Brothers Company, Joint Venture, Trading as Tecon-Green-Winston v. The United States

PER CURIAM. This case was referred to Trial Commissioner Roald A. Hogenson with directions to make recommendation for conclusions of law on plaintiff’s motion for summary judgment and defendant’s cross-motion for summary judgment under the order of reference and Rule 99 (c). The commissioner has done so in an opinion and report filed on October 4, 1968, wherein such facts as are necessary to the opinion are set forth. Plaintiff filed a request for review of the commissioner’s report and the case has been submitted to the court on oral argument of counsel and the briefs of the parties. Since the court agrees with the commissioner’s opinion and recommended conclusion of law, as hereinafter set forth, it hereby adopts the same as the basis for its judgment in this case. It is, therefore, concluded that plaintiff’s specifications ■ of errors of law and fact in the decision of the Corps of Engineers Board of Contract Appeals should not be sustained; that the board’s findings of fact are supported by substantial evidence; that the board’s decision is correct as a matter of law; that plaintiff’s motion for summary judgment should be, and is, denied; that defendant’s cross-motion for summary judgment should be, and is, allowed ; and that plaintiff’s petition should be, and is, dismissed. . OPINION OF COMMISSIONER HOGENSON, Commissioner: On plaintiff’s motion and defendant’s cross-motion for summary judgment, and the supporting briefs of the parties, this case presents 'issues pertaining to the finality of a decision of the Corps of Engineers Board of Contract Appeals, ENG BCA 2704, 67-1 BCA ft 6148. The challenged decision denied plaintiff’s claim presented to the board under the standard changes, suspension of work, and disputes articles of the contract (DA-34-066-CIVENG — 60-1407) entered into on January 11, 1960, by and between plaintiff and defendant, acting by and through the Corps of Engineers, Department of the Army. Under such contract, plaintiff constructed the spillway, power intake, and power substructure of the Eufaula Dam on the Canadian River in Oklahoma. Plaintiff asserts that the board’s decision is not final and conclusive under the standards of judicial review prescribed by the Wunderlich Act, 41 U.S.C. §§ 321-322 (1964), in that such decision contains specified errors of law and in that certain designated findings of fact are not supported by substantial evidence; and that upon the facts established in the record of the administrative proceedings, plaintiff is entitled to recover as a matter of law. Plaintiff moves this court for entry of judgment to that effect, and since the board deferred consideration of issues of damages and made no decision thereon. *1273 requests suspension of proceedings herein to permit the parties to reinvoke administrative remedies for determination of the equitable adjustment of the contract price. Defendant asserts that the board’s decision is entitled to finality, and requests that plaintiff’s petition be dismissed. Plaintiff is a joint venture known as Tecon-Green-Winston,- consisting of Te-con Corporation of Dallas, Texas; Green Construction Company of Des Moines, Iowa; and Winston Bros. Company of Minneapolis, Minnesota. The pertinent contract, employing unit prices on some items with estimated quantities, had an estimated price of $14,672,709. The basic dispute before the board, now on review before this court, is the question of the responsibility for the reworking of the fabrication of certain steel girders, called trunnion anchorage thrust girders, installed near the tops of central piers of the dam, as parts of the structures and mechanisms employed to operate the tainter gates to be raised and lowered to control the quantities of water passing over the spillway. The reworking of such girders was necessitated by cracking of some of the welds' originally placed to secure together various parts of each of such girders. Plaintiff had the required 22 girders fabricated under subcontract with Creamer Industries, Inc., at the latter’s plant in Fort Worth, Texas, and the necessary reworking of all of such girders was also done by such subcontractor at the same plant. Plaintiff’s claim is that the production of such girders was a practical impossibility under the defendant’s original specifications and drawings, that substantial modifications were required and employed to accomplish fabrication free of cracking in the welds, and that plaintiff is entitled under the changes article of the contract to an equitable adjustment of the contract price on account of the increased cost resulting from such defective specifications and drawings. Plaintiff’s claim is further that the reworking of the girders caused delays in the performance of other contract work, i.e., installation and construction of the handrail bridge and non-overflow sections, spillway bridge, tainter gates and other related items, and that under the suspension of work article of the contract, plaintiff is entitled to an equitable adjustment on account of the increased costs resulting from such delays. The girders in dispute are now described as specified in the original contract documents, with the limited modifications as constructed on refabrication later detailed. Each girder consisted of various steel plates, all being 2 inches thick except some 1-inch plates specifically described as such hereinafter. The girder is described in horizontal position, although actually installed inclined. It was about 12 feet long, about 2.2 feet wide, and about 3.3 feet high. It can be described as shaped like a large I beam with added structural members. Flanges. The bottom of the girder, or bottom flange, was a horizontal plate, having the length (12 feet) and the width (2.2 feet) of the overall girder. The top flange (or top plate of the girder) had the same width as the bottom flange, and extended in horizontal plane for most of the length of the girder. However, near each end of the girder, the top flange curved downward (1-foot radius curve) and then continued vertically (for the remainder of, being most of the 3.3-foot height of the girder) to the bottom flange. Thus, the top flange can be described as a continuous flange constituting the top horizontal plate and both vertical end plates of the girder, with an appropriate downward curvature of the continuous flange at each end of the girder from the horizontal to the vertical plane. At the bottom of each end of the girder, the end edge of the top flange was welded to the upper surface of the bottom flange. Actually, each flange consisted of two plates, extended in succession length *1274 wise, with their adjoining ends welded together. After the two plates for the top flange had been welded together, such continuous flange was bent into the curvatures above described. Thus, each flange had a welded joint other than those at the junctures of the top and bottom flanges. This welded joint for each flange was located on the top or bottom of the girder in corresponding position with the other, both of such welds being intermediate between the ends of the girder. Each flange was somewhat wider at one end of the girder than at the other, due to the fact that each flange had one plate about 2.7 feet wide and the other plate 2.2 feet wide. On each flange, the edges of the wider plate tapered where welded to the narrower part of the flange. By way of general description, it can be said that with the flanges welded together, as above described, their overall configuration would resemble a large rectangular box without top or bottom, turned over on one of its long sides, with the upper corners curved and the lower corners squared. There is no evidence in this case of any cracking of the welds either in the joint between the individual plates of each flange, or in the joints between the top and bottom flange. Web. The remaining major plate in each girder was the web. Like the location of the web in an I beam, it was placed vertically between the top and bottom flanges, its vertical plane being centered between the opposite edges of each flange. Its overall shape was rectangular, except that its upper corners were curved to fit the corresponding curvatures of the top flange. Its edges fit snugly against the inside contours of the flanges, at the top, bottom, and both ends of the girder. All of the edges of the web plate were welded to the flanges. Most of the cracks occurred in this weld, and of these cracks, most were in the weld metal at the curvatures of the web and top flange. Stiffeners. Various rectangular plates, called stiffeners, were placed in each girder. They extended vertically between the inside surfaces of the top and bottom flanges and were at right angles to the web. There were 10 corresponding stiffeners on each side of the web plate. Each stiffener extended from the web to the corresponding outer edges of the flanges. On each side of the web plate, seven were 1 inch thick, spaced in the central one-third of the girder; and three were 2 inches thick, one located toward one end of the girder, and two toward the other end. All three contact edges of each stiffener were welded to the web and to the top and bottom flanges. In lesser quantity than the cracks in the web-to-flange weld, some cracks occurred in some of the stiffener-to-flange welds. The contract drawings and approved welding procedure showed the edge preparation of the plates to be welded. For the web-to-flange joint around the entire perimeter of the web plate, the edge of the web had double J grooves, i. e., the edge was gouged out or grooved on each side, with each groove resembling the shape of the main stem of the letter “J”, with the depth of the groove (across the edge of the plate) %-ineh on each side of the plate. Thus, the land zone, or remaining edge of the 2-inch web plate in contact with the flange, was y2-inch thick. In the welding process, each groove was filled with weld metal, and a ,%-inch fillet weld placed alongside. In the approved welding procedure, the contact edges of each 2-inch stiffener (at joints with web and top and bottom flange) were shown to be beveled on each side to a sharp edge, which would be burned away, resulting in full penetration of the weld metal in the joint, with a fillet weld along each side. The edges of the 1-inch stiffeners were not altered, and only fillet welds were indicated. The fillet welds for all stiffeners were %-inch. In accordance with the requirement of Rules 96(b) (2) and (3) of this court, *1275 plaintiff attacks the finality of the board’s decision on the following specific grounds: 1. Not expressed in the administrative decision, but underlying its reasoning, is the erroneous conclusion of law that where a contractor has shown the Government’s design to be technically deficient and has shown that cracks which appeared in the welded structures, which he was fabricating according to that design, were the probable result of the design deficiency, he may nevertheless not recover if he failed to negate all other possible causes of cracking no matter how remote, speculative, or unlikely they may be. 2. The Board’s ultimate conclusion that the Government design was not a practical impossibility is a mixed question of fact and to the extent that it constitutes a conclusion of law, is erroneous. 3. To the extent that the Board’s finding that the Government design was not a practical impossibility is a question of fact, it is not supported by substantial evidence. 4. The Board’s finding that the cracks which appeared in the welds were the result of poor workmanship is not supported by substantial evidence. In support of finality of the board’s decision, defendant argues that the board was legally correct and supported by substantial evidence in finding: (1) That plaintiff had not proved that the specifications were followed in the initial fabrication of the girders; (2) that plaintiff’s subcontractor deviated from the specifications in regard to the welding sequence, which produced slag pockets and discontinuous welds which are a possible cause of the cracking; (3) that other possible causes of the cracks were poor workmanship in fitting the curved portions of the webs and flanges and inadequate preheating during the initial fabrication; (4) that the repair procedures did not constitute a material change in the girder design; and (5) that the plaintiff had failed to prove practical impossibility of the performance pursuant to the specifications. Plaintiff’s performance of the contract commenced in February 1960 pursuant to issuance by defendant of notice to proceed, and in early March 1960 plaintiff issued and the subcontractor accepted a purchase order for the fabrication of the thrust girders. The subcontractor procured the specified A373 steel in large sheets, a mild steel amenable to welding, and in time proceeded to cut the steel into the sizes of plates required for the girders. Shop drawings of the girders were prepared by the subcontractor, and these were approved by the defendant’s contracting officer on July 15, 1960, subject to certain notes. One of such notes referred to the contract specifications restricting welding and fabrication of the girders until a welding procedure was submitted to and approved by the contracting officer. The first welding procedure submitted was rejected on October 24, 1960, but a revised procedure was approved November 14,1960. The girders were fabricated and shipped from subcontractor’s Fort Worth plant, arriving in shipments of 2 to 4 girders at the Eufaula Dam site in January and February 1961. Prior to each shipment, each girder was inspected by a duly assigned government inspector, who signed a material receiving and inspection report, DD Form 250, under the following statement: I certify that the items listed herein have been inspected by me or under my supervision. They conform to contract. One inspector was regularly assigned to the plant, devoting about 2 hours per day to the girder fabrication, and when he was occasionally absent, one of two other inspectors was sent to the plant to substitute for him. No cracks in the welds of the girders were discovered prior to the shipments. However, the welds were covered with *1276 weld slag and dirt generated in the plant operations. Sandblasting and cleaning of the girders were to be performed by a painting subcontractor at the job site. At the outset of the shipments, the regular inspector informed the subcontractor that he had no authority to make final acceptance of the girders because final fitting up and assembly were to be in the field. Furthermore, the regular inspector asked the subcontractor to clean the girders for inspection before shipment, but the subcontractor’s plant superintendent declined because the girders were to be sandblasted and cleaned in the field for painting. Under the circumstances of this case, it appears that the board correctly ruled that the cracks discovered initially at the job site after sandblasting of the girders had commenced, were latent defects, rendering inapplicable any final acceptance at the fabrication plant under the inspection article of the contract. In any event, such ruling is not in issue before this court because plaintiff has not assigned the same as an error of fact or law committed by the board. The girders remained in outside storage at the job site from February 1961 to October 1961 by which time plaintiff was ready as originally planned to commence their installation. On October 12, 1961, two had already been installed, and two others had been sandblasted in preparation for painting, when cracks in some of the welds of the latter were observed. Inspection of other girders revealed cracks in welds prior to cleaning. In about June 1961, defendant’s assistant resident engineer had rather closely inspected the girders in connection with the visit of various persons undertaking fabrication of similar girders on another dam project, and he observed no cracks. The reasonable explanation in the record for this phenomenon is that minute cracks had probably existed at the time of plant inspection prior to shipment, not observed because of their size or because obscured by weld slag and dirt, and that such cracks were propagated by stresses resulting from exposure of the girders to the wide extremes of temperature at the Oklahoma job site during the long period of their outside storage. At the meeting at the job site on October 13, 1961, between representatives of defendant, plaintiff and the subcontractor, it was conceded that the cracked areas of the welds would have to be repaired, and the subcontractor proposed to accomplish such work at the job site with its own plant personnel. These workmen were nonunion, labor troubles threatened contract performance, and the subcontractor therefore returned all of the girders in pairs to its Fort Worth plant. The initial attempt to repair the girders can reasonably be described as occurring in two efforts, in neither of which was any girder disassembled or completely rewelded. The first effort involved the first two girders returned to the plant on October 19, 1961. In accordance with a repair procedure submitted by the subcontractor and approved by defendant, and under close supervision by the subcontractor, with representatives of plaintiff and defendant in attendance, the cracked weld areas of these two girders were arc-air gouged out and rewelded. After postheating, cracks occurred in the re-welded areas. The second effort to repair concerned the first four girders returned to the plant. Defendant then instructed plaintiff, and plaintiff in turn the subcontractor, that cracked weld areas were to be gouged out and rewelded, and then the girders submitted to spot X-ray testing and finally to stress relieving. Under close supervision, with representatives of plaintiff and defendant in attendance, this gouging out and reweld-ing were accomplished. The X-ray tests revealed no cracks, and these girders were sent to stress relieving, a process of gradually and uniformly postheating a fabricated metal weldment to a high temperature (with gradual and uniform cooling) to realign the molecules and *1277 distribute stresses to prevent cracking. After stress relieving, there were cracks in the repaired weld areas and also in some other weld areas of the girders. The subcontractor then retained two experts for consultation on the welding difficulties encountered, Dr. A. Anthony Toprac, Professor of Civil Engineering and Director of the Structures Fatigue Research Laboratory, University of Texas, and Mr. Cyril S. Adams, consultant and registered professional engineer, Houston, Texas. The latter died before the board hearing in this ease. At a meeting at subcontractor’s plant on October 31, 1961, Dr. Toprac and Mr. Adams explained to representatives of plaintiff and defendant the basis of their conclusions that the cracks had been caused by deficiencies in the girder design and contract specifications. Thereafter, a meeting was held on November 6, 1961, at defendant’s Tulsa District Office, with representatives of defendant, plaintiff and subcontractor in attendance, at which the subcontractor stated its position that the design and welding procedure were not workable, and Dr. Toprac explained in detail his conclusion that the girders were overde-signed and that they could not be successfully fabricated under the existing welding procedure. The subcontractor then orally presented its recommendations for revisions, and defendant requested submission of the same in writing. On November 8, 1961, the subcontractor submitted nine recommendations, seven of which were approved by defendant that same day, the approved recommendations being herein summarized and explained as follows : 1. Permit use of low hydrogen rods E 7018 and E 11018. The contract specifications designated use of E 6010 and E 6011 electrodes (welding rods) in the fabrication of the girders. Plaintiff’s original welding procedure proposed use of E 60 electrodes, a group of 15 different electrodes numbered E 6010 through E 6024, some of which were not suitable for the welding involved in this case. In rejecting plaintiff’s original welding procedure, defendant designated use of the E 6010 and E 6011 rods, and the welding procedure revised, approved and used on the original fabrication limited plaintiff (except for a relatively few passes with an E 6024) to the use of such electrodes, as the specifications provided. The expert testimony for both parties established that the specified A373 steel was weldable with the E 6010 and E 6011 rods, but also that the low hydrogen rods were better electrodes for welding of the girders. Among theE 60 series, E 6015 and E 6016 were low hydrogen rods essentially like E 7018, approved for the final reworking of the girders. Plaintiff did not request use of E 6015 or E 6016 for the original fabrication. The reasons why low hydrogen rods were preferable were because less preheating of the welded areas would be required and because less hydrogen would be entrapped in the weld metal to cause hydrogen embrittlement. On the final repair of the girders, the subcontractor used both the E 7018 and the E 11018 rods on the first three girders, but thereafter used the E 7018 rod exclusively on the remaining girders. 2. Preheating of the areas to be welded shall be not less than 300° F., and a continuous heat application shall be made to the work during the entire welding cycle. The specifications and original welding procedure required preheating at a minimum temperature of 200° F. In the original fabrication, each welder used a single jet burner, heated the relatively small area he was about to weld, completed the welding of that area, and then moved on to the next area for preheating and welding. Each welder used a standard tool called a tem-pelstik to determine whether the joint to be welded and adjacent metal had been sufficiently heated. This consisted of applying a chalk mark to the surface of the plate, which mark disappeared if the surface was sufficiently heated. In the final fabrication, the subcontractor erected each girder on a jig *1278 (with the web plate horizontal) about 5 feet above the floor, and placed thereunder a burner assembly, consisting basically of a %-inch pipe, emitting flame the full length of the girder from holes spaced about 4 inches apart. The pipe was placed about 6 inches below the web plate, and the web-to-flange joint was rewelded with preheating and continuous heat application at a minimum of 300° F. in that manner. Apparently the stiffeners were rewelded by use of a single jet burner, with preheating at 200° F. 3. Remove the stiffeners. With respect to the web-to-flange weld, on one side of the web plate, gouge out over one-half of the weld metal around the entire perimeter. Complete all around one-half of the rewelding on that side. Then, gouge out the weld metal on the other side to sound weld metal, and complete one-half of the rewelding on that side. Complete the rewelding on the first side, and then complete the reweld-ing on the second side. In the accomplishment of the final refabrication, the stiffeners were removed, after which the rewelding of the web-to-flange joint was accomplished, as above stated, with the web plate in horizontal position on each step of the operations. Such horizontal welding was required by the specifications. Of course, the girder was reversed, and the rewelding in each step was done from the top side of the web plate, with the burner operating underneath such plate. In the original fabrication, and in the two unsuccessful attempts to repair, the stiffeners were in place in each girder when the welding was done. The first welding of each girder was not done until its various plates had been assembled together by tack welding. Thus, at each point along the web-to-flange weld, where the square corner of a stiffener fit tightly against both the web and the flange, voids occurred in that weld. The contract specifications required continuous welds. The sequence of welding required in the originally approved welding procedure, after each flange had been separately fabricated, was as follows: 1. Weld the web to the bottom flange. 2. Weld the web to the top flange. 3. Weld the top flange to the bottom flange at both ends. 4. Weld the 2-inch stiffeners in place. 5. Weld the 1-inch stiffeners in place. It is reasonable to conclude that this sequence contemplated that the web-to-flange connections be completely welded before the stiffeners were placed in the girder. Otherwise, voids were likely to occur, as they did, at the locations where the corners of the stiffeners interrupted the welding of the web to the flange. 4. Cut out the curved corners of the web where they fit the curvatures of the top flange (as shown in a drawing submitted by the subcontractor), in order to relieve the internal stress loading of the curved portion of the web welds due to thermal stresses. As permitted by this approved recommendation, the subcontractor cut out a portion of each web plate at each of its curved corners, prior to rewelding the web to the flange, and this left at those locations no contact between the edge of the web plate and the flange. At each such cutout, of course, there was a termination of the web welds, and the open ends of such welds were seal welded. The expert testimony for both parties was in agreement that the welding of the curved corner of the web to the curvature of the flange created triaxial stresses which could contribute to the cracking of the weld at that location. Such triaxial stresses were also present in each of the girders as originally fabricated at each of the locations in the girder where an inside (square) corner of a stiffener fitted snugly against both the web and the flange, i.e., where three welds met, the web-to-flange, and the stiffener-to-flange, and the stiffener-to-web weld. However, the latter triaxial *1279 stresses were eliminated in the final reworking of the girders. This was done by clipping the inside corners of each stiffener, a procedure not expressly authorized by any welding procedure involved in this case or by the contract documents. However, the evidence is un-contradicted that clipping of the corners of such plates is common practice in the welding trade, usually done even though not expressly authorized in the specifications, drawings, or welding procedures. The prior placing of the continuous web-to-flange weld, with its fillet weld, made it impossible to fit the edge of a stiffener, with square corners, against the web plate, and clipping of such corners was necessary, if the fillet weld was to be continuous. 5. In order to reduce the welding stresses in the structure, 1" stiffener members shall be welded using %" fillet welds. This was not a change from the originally approved welding procedure. 6. For the same reasons the 2" stiffener members shall be beveled and welded with a resultant %" fillet weld. The fillet weld remained the same size as specified in the original welding procedure, but the contact edges of the 2-inch stiffeners had 1.5-inch land zones, whereas complete beveling to a sharp edge was prescribed in the orginally approved welding procedure. Actually, however, the subcontractor left land zones varying from x/i inch to 1 inch on these stiffeners in the original fabrication, as was shown when each girder was disassembled in the final repair operations. 7. As indicated by subcontractor’s procedures, the welds between the web and the end pieces uñll be gouged and repaired as found necessary, as the end flanges in themselves only constitute a stiffener. In the final repair work, however, the web-to-flange joints were completely gouged out and rewelded in the manner described above. The two rejected recommendations of the subcontractor were substitution of a weaving method for the single stringer welding prescribed in the original welding procedure, and a change in the structure of the girder by replacement of the 2-inch stiffeners with 1-inch stiffeners. Under the specifications and welding procedure as modified by the approved recommendations, the subcontractor successfully refabricated the girders, commenced shipments to the job site in late November 1961, and such girders were thereafter installed on the dam piers. This is a close case in which a review of the cold administrative record leaves one with the impression that there may be merit to plaintiff’s claim, and had one had the advantage of observing the witnesses, the conclusion might have been reached by preponderating the evidence that plaintiff would be entitled to recover. Dr. Toprac, obviously well qualified in the field of stresses created in welded steel structures, seemed firm in his opinion (detailed in his testimony for plaintiff before the board) that the subject girder was overdesigned both from the standpoint of its structural members and the quantity of welding required, and that under the specifications, drawings, and original welding procedure, it was a practical impossibility to fabricate such girder without cracking in the welds. But the basic question in this ease is whether there is substantial evidence in the administrative record to support the board’s ultimate conclusion that plaintiff has failed to prove that performance was a practical impossibility. Obviously, this is essentially an issue of fact, as the parties are not in dispute as to the meaning of pertinent specifications, drawings, or welding procedures. River Constr. Corp. v. United States, 159 Ct. Cl. 254, 262-266 (1962). Of course, consideration must be given to all of the evidence, that which militates against and that which supports the board’s decision, but the board’s decision must be upheld if there is substantial evidence to support such decision on the record as a whole. Carlo Bianchi & Co. v. United States, 167 Ct.Cl. 364, 368 (1964), cert. *1280 denied 382 U.S. 841, 86 S.Ct. 32, 15 L.Ed.2d 82 (1965). If there was such evidence as might convince a reasonable man to reach the conclusion made by the board, then the board must be sustained, even though the reviewing court might have decided otherwise. T. C. Bateson Constr. Co. v. United States, 149 Ct.Cl. 514, 518 (1960). Furthermore, since the board had the opportunity to observe the witnesses, and to assess their credibility and also the weight to be given to their testimony, this court should accept the board’s evaluation of the conflicting conclusions of expert witnesses on the issue of practical impossibility of performance, unless the testimony accepted is inherently improbable or discredited by uncontrovertible evidence or physical fact. Williamsburg Drapery Co. v. United States, 369 F.2d 729, 733, 177 Ct.Cl. 776, 783 (1966). Professor Roy B. McCauley, Chairman of the Department of Welding Engineering and Director of Welding Research at the Engineering Experiment Station, Ohio State University, was the expert witness for defendant in the board proceedings. He considered the contract specifications and drawings concerning the subject girders, and also the originally approved welding procedure, and testified squarely that in his opinion the girders so described could be fabricated under such conditions without difficulty without cracking in the welds. He agreed strongly with Dr. Toprac’s recommendation that the E 7018 electrode should be used on the final reworking of the girders, but asserted that at the time of the issuance of the subject specifications and original welding procedure, ‘ the prescribed E 6010 electrode was the “hand maiden” of the welding industry. He testified that preheating to allow distribution of stresses was common knowledge in the industry, and regarding the originally specified preheating to a minimum of 200° F., stated that for heavier structures like the subject girders, large burner bars should have been used to heat a large portion of the underside of a plate, while the welder applied supplementary heat and welded on the top surface. He criticized the application of surface heat only as not permitting penetration of the heat sufficiently through the material to accomplish absorption of energy and reduction of residual stresses in the welding. Professor McCauley agreed with Dr. Toptrac’s recommendation (not accepted by defendant) that the 2-inch stiffeners be eliminated and certain 1-inch stiffeners substituted to reduce the weld metal and accomplish less restraint and less residual stress in the fabricated girder. He also agreed with Dr. Toprac that the curved portions of the flange were unloaded, and that it was not necessary to maintain contact with the web at those points. In other words, he conceded that cutting out of the corners of the webs was an improvement, eliminating the geometric restraint resulting from welding of the curve of the web to the curve of the flange. He stated that poor preheating, i.e., the use of single jet burners, and a poor fit-up between the curved corner of the web plate and the corresponding curvature of the flange would have quite adequately supplied reasons for the cracking of the web-flange sections. It was his opinion that the constraint caused by welding the web plate to the curvature of the flange did not alone create the problem with generated cracking. Professor McCauley conceded that triaxial stress was generated at each junction of the square edge of a stiffener with the web and flange, but stated that clipping of the corners of such plates (as was done on final reworking) was standard practice in the welding industry, which is done unless the designer specifically directs otherwise. It was his belief that the tack-up procedure used by the subcontractor, assembly of all parts prior to welding, made it extremely difficult, if not impossible, to weld the web to the flange at each point where the square corner of a stiffener was located, and that lack of welding at any such location would result in a notched geometry, which would aggra *1281 vate the fit-up problems at the flange curves and the poor preheating procedures, and facilitate cracking of the welds. He stated that cracking would tend to initiate at the voids in the web-to-flange weld underneath the square corners of the stiffeners. Professor Mc-Cauley conceded that the subject girder was a restrained structure, as most girders are, but stated that fabrication of the girders was well within the province of good workmanship. Professor McCauley flatly disagreed with Dr. Toprac’s statements to the effect that the originally approved welding procedure, coupled with structural design defects, made the girder susceptible to cracking, even though good commercial fabrication techniques were used. Professor McCauley relied upon the testimony of defendant’s regularly assigned plant inspector as to the preheating procedure used by the subcontractor on the original fabrication of the girders. His assumption that there was a poor fit-up in the original welding of the web to the curvature of the flange was based on the testimony of defendant’s district coordinator of shop inspection, Robert M. Sutter. Mr. Sutter visited the subcontractor’s plant, before there had been an approval of the original welding procedure, and observed that the fabrication of the girders was under way to the extent that layout had been completed on numerous plates and some of the plates (flanges, stiffeners and webs) had been cut into final shapes. He saw several of the girders assembled by tack welding. He observed that the curvatures of top flanges had been formed on a device other than a permeated roll, which he thought should have properly been used, whereas the equipment used was a hydraulic press. Instead of there being a true curve, there was on each curvature a series of straight bends. Also, he observed that the curvature of the web had the appearance of having been prescribed in layout and cut by hand with the use of a burning torch, and was not a true curve, but was scalloped along the curved edge. He stated that there was a lack of matching of the curvature of the web with the curvature of the flange, to the extent that at that location, the web was in contact with the flange in some places but in other places there was a gap more than there normally should be for welding. Of course, at the gaps, more weld metal would have been required than at places where the edge of the web plate contacted the flange. Mr. Sutter testified that he requested the subcontractor to correct the poor fit-up at the curvatures of the flange. There was no direct evidence that such corrections were or were not made. Plaintiff argues that the success of the final repair proves that the fit-up must have been adequate at all times, minimizing the fact that most of the curve of the web was cut away from the curve of the flange in the final repair operations. From a review of the entire administrative record in this case, it is concluded that the board was supported by substantial evidence in finding: (1) That plaintiff had not proved that the specifications were followed in the initial fabrication of the girders; (2) that plaintiff’s subcontractor deviated from the specifications in regard to the welding sequence, which produced slag pockets and discontinuous welds, which were a possible cause of the cracking; (3) that other possible causes of the cracks were poor workmanship in fitting the curved portions of the webs and flanges and inadequate preheating during the initial fabrication; (4) that the repair procedures did not constitute a material change in the girder design; and (5) that plaintiff had failed to prove practical impossibility of performance pursuant to the specifications. Plaintiff concedes that the board in its decision correctly summarized the doctrine of practical impossibility as follows: Even though performance is not actually impossible, a contractor is entitled to relief if it can be shown that the contract requirements are impracticable because of extreme and unrea *1282 sonable difficulty, expense, injury or loss involved in meeting them. Plaintiff argues that the board, having thus recognized that the law does not require proof of absolute or literal impossibility, misapplied the rule of law and held in effect that plaintiff had the burden of showing absolute impossibility. In support of such argument, plaintiff relies upon the following language extracted from the board’s decision: The above statement of practical impossibility requires affirmative proof of extreme and unreasonable difficulty. To prove this, other possible sources of difficulty, such as lack of know-how, and poor planning or workmanship, must be negated. In the instant matter the evidence shows that possible sources of the cracks in the welded joints are slag pockets behind the stiffeners, discontinuous welds in the web to flange joint, and possibly inadequate preheating. Appellant has not produced proof that the existence of these deficiencies did not cause the cracks in the welds. This paragraph of the board’s decision was concluded with the following sentence : * * in fact, appellant does not deny the existence of the slag pockets and that discontinuous welds were the result of the fit up procedure used by the suppliers. Aside from the fact that plaintiff in making this argument relies upon only a small part of the board’s decision, all of which must be read to comprehend the legal basis of the decision, plaintiff is really arguing (as shown by its brief) that plaintiff “did effectively negate each specific instance of poor workmanship alleged by the Government,” and that plaintiff did “show that from all the evidence presented the overwhelming probability was that the fault lay in the structure design and mandatory welding procedures.” Thus, it is apparent that plaintiff’s argument fails on the determination that there was substantial evidence supporting the board’s decision. Moreover, a careful reading of the entire decision demonstrates that the board applied the rule that plaintiff had the burden of establishing that fabrication of the girders was extremely and unreasonably difficult, not absolutely impossible. The board’s decision should be sustained. CONCLUSION OF LAW Upon review of the entire administrative record herein, it is concluded that plaintiff’s specifications of errors of law and fact in the decision of the Corps of Engineers Board of Contract Appeals should not be sustained; that the board’s findings of fact are supported by substantial evidence; that the board’s decision is correct as a matter of law; that plaintiff’s motion for summary judgment should be, and is, denied; that defendant’s cross-motion for summary judgment should be, and is, allowed ; and that plaintiff’s petition should be, and is, dismissed.