«TEST OF THICK VESSEL WITH A FLAW IN RESIDUAL STRESS FIELD* R. H. Bryan P. P. Holz S. K. Iskander1 J. G. Merkle G. D. Whitman Oak Ridge National ...»
— " " " • " ««•
TEST OF THICK VESSEL WITH A FLAW
IN RESIDUAL STRESS FIELD*
R. H. Bryan P. P. Holz
S. K. Iskander1 J. G. Merkle
G. D. Whitman
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37830
Intermediate test vessel V-8, a 152-mm-thick vessel fabricated of SA533,
grade B, class 1 steel, was pressurized to failure at —23°C. The vessel contained a fatigue-sharpened notch adjacent to a half-bead weld repair that had not been stress relieved. Residual stresses and fracture toughnesses were determined before the pressure test by measurements on a prototypical weld, and fracture predictions were made by linear elastic fracture analysis. Predictions agreed well with test results, demonstrating the important influence of high residual stresses on fracture behavior.
By acceptance of this article, the publisher or recipient acknowledges the U.S. Government's right to retain a nonexclusive, royalty-free license in and to any copyright covering the article.
Computer Sciences Division, Union Carbide Corp., Nuclear Division.
situation in which a repair without post-weld heat treatment is considered, the adequacy of the weldment with regard to expected residual stresses and fracture toughness should be assessed.
In the Heavy-Section Steel Technology (HSST) Program six half-bead weld repairs were made substantially in accordance with Section XI of the Code in 152-mm-thick vessels and cylinders fabricated of ASTM A533, grade B, class 1 steel plates; material properties were determined; residual stresses were measured; and flawed vessels were tested to failure in order to provide baseline information for use by persons who make assessments of such repairs.
Four of the welds were within the plate material and extended completely
both plate material and submerged-arc fabrication welds. The weldments in the cylinders were prototypical of three different repair welds in the HSST intermediate test vessels V-7 and V-8.
Three tests were conducted with the repaired and flawed vessels. The first two tests, V-7A and V-7B, were performed at upper shelf temperatures.
The V-7A test  was a pneumatic test with the prepared flaw remote from the repair xveld, which was incidental to the test objective. In V-7B  the sharpened tip of the prepared flaw was placed in the heat-affected zone of the repair weld. This hydraulic test was an attempted replication of the two earlier tests of this vessel, and it resulted in a leak at about the same pressures attained previously. The flaw in the repair zone, unlike the others, exhibited much greater tearing axially than had been previo.sly observed.
This tearing could have been a consequence of the residual stress field near the flaw or of the slightly higher test pressure, which was about 2.2 x design pressure.
The objectives of the test of vessel V-8 were (1) to demonstrate the behavior of a flaw in the residual stress field near a half-bead weld repair, and (2) to evaluate linear elastic fracture mechanics as a means of predicting the combined effects of residual stress and pressure stress on fracture.
In order to realize the second objective it was necessary to attain a low toughness condition in a region of high residual stress.
DESCRIPTION OF VESSELThe 990-mm-OD by 152-mm-thick cylindrical test section of vessel V-8 was fabricated of ASTM A533, grade B, class 1 steel plate. A repair weld 318 mm long by 89 mm deep was made in this cylinder to provide a flaw site in the vicinity of the repair weldment that was not stress relieved. The weld was made in accordance with the sequence proscribed in the ASME Boiler and Pressure Vessel Code, Section XI, Subsubarticle IWB-4420, weld repair procedure without post-weld stress relief.  A flaw was machined and fatigue sharpened in the original vessel fabrication weld, as shown in Fig. 1.
The 'estinghouse Electric Corporation, Tampa Division, performed the halfbead weld repair of the cavity in vesssel V-8, shown in Fig. 2, and in an identical cavity in a 600-mm-long prolongation of the test cylinder.  The prototypical weld in the prolongation provided material for determination of residual stresses and material properties.
The weld cavity was located across one edge of the fabrication weld so as to provide a choice among five different flaw sites near the repair: in base metal, in the base metal heat-affected zone (HAZ), in repair weld metal, in the fabrication weld HAZ, and in the fabrication weld. The axial sides of tne cavity in the vessel were machined along radial planes so that their loca
changes on exposed surfaces of the vessel and prolongation and by the holedrilling technique.  The vessel and prolongation were instrumented with weldable strain gages prior to repair welding and the changes in strain induced by welding were recorded for twenty locations on the prolongation near the repair and for twenty-five locations on the vessel. Subsequently the prolongation was cut into two rings so as to expose the central cros;; section of the repair weldment. Strain changes induced by this cut were measured and used as a basis for adjusting the residual stresses measured on the cut surface. Principal stresses were measured at 46 points en this surface by hole drilling. The same technique was used to measure outside-surface residual stresses at 24 points near the repair on the vessel. Additional surface measurements x^ere made to evaluate the residual stresses induced during vessel fabrication.
The principal results of the residual stress investigation are summarized in Figs. 3, 4, and 5. These data show that r idual stresses are low within the repair weld itself and rise to values near yield in nearby material at the outside surface.
The half of the prolongation not used for residual stress measurements was sectioned for the study of material properties. Hardness measurements were made across the section of the repair shown in Fig. 6. Tensile properties of the repair weld metal were determined from specimens taken from halfbead weld metal test plates made by Westinghouse with the specified E-8018-C3 welding electrodes. Fracture toughness of tht various regions of the repair was evaluated by sloxj bend tests of precracked Charpy specimens and static tests of IT and 2T compact specimens taken from the prototypical weldment.
Additional fracture specimens were obtained from the fabrication weld in another prolongation (V-9) fabricated with vessel V-8 and its prolongation.
Hardness measurements made along the three traverses across the HAZ's shown in Fig. 6 indicate that both base metal HAZ's are similar in hardness and are harder than any other region. The results along traverse A are presented in Fig. 7. These data imply that the repair weld-base metal HAZ has substantially higher yield and ultimate strengths than the plate or weld materials.
The tensile properties of the repair weld metal are compared in Table 1 with those of plate material from another prolongation (V-7) in this series.
The relative fracture toughnesses of the various zones of the repair weldment were evaluated by numerous slow-bend tests of precracked Charpy specimens. The values low in the transition range were of greatest interest in
Specimens from 107 mm depth.
planning the V-8 test because low values were needed for a successful test.
In specimens used to measure HAZ toughness the sharpened precracks were carefully placed within the HAZ. The lowest toughness values were found in the fabrication weld, while the repair weld exhibited the next lowest values and the HAZ's were at least as tough as the base material. These conclusions were also supported by the earlier HSST vessel test, V-7B.  Further evaluations of toughness were concentrated on the submerged-arc fabrication weld, particularly in the region close to the repair weld. (See Fig. 1.) Pretest estimates of fracture toughness depended primarily upon data from IT and 2 f compact specimens taken from prolongations of vessels V-8 and ' V-9. Although these two pieces were fabricated with vessel V-8, cne possibility of variability was considered, since high variability would suggest a greater uncertainty in attributing the measured toughnesses to the vessel itself.
Results of the toughness study are presented in Fig. 8. The work of Canonico, Crouse, and Stelzman had shown a correlation between precracked Charpy data and the microstructure of the weld. [6,7] Hence, it was considered possible that the behavior of the larger fracture toughness specimens was controlled by the lower toughness structures within the weld. If that were the case, a sharp crack in the submerged-arc weld in the vessel (in the location shown in Fig. 1) would also tend to be controlled by these structures. Consequently, the lowest value of toughness (VLOO MN-m"3''2) shown in Fig. 8 in the -25 C C to —20°C range was chosen as an appropriate lower bound value of K for pretest analysis.
Experience in flaw preparation and testing favored the selection of a flaw size and geometry that were similar to those used in the first series of HSST intermediate vessel tests.  Preliminary calculations confirmed the suitability of this selection. Accordingly, a notch 206 mm long and
50.5 mm deep was cut into the fabrication weld and fatigue sharpened to a total depth of 62.5 mm (Fig. 1 ). The fatiguing was accomplished by sealing the notch opening and pressurizing the notch cyclically. Acoustic emission detectors were used to sense crack activity and ultrasonic equipment measured the fatigue crack growth. The crack growth rate was significantly elevated by the residual stress field.
FRACTURE ANALYSIS METHODSPrevious intermediate vessel tests had demonstrated tliat analyses by m°ans of linear elastic fracture mechanics (LEFM) accurately predict crack initiation pressure, provided initiation preceeds the onset of gross yielding in the test section. In order to gain more assurance that the comparisons of LEFM and test results would not be confused by competing phenomena, it was decided that the test pressures should be limited to a range in which yielding would not occur in an unflawed cylinder. The upper limit on the test pressure was about 160 MPa, or 2.A x the design pressure.
Exploratory LEFM calculations of K, for the combined loading from pressure and residual stress were based upon the work of Smith and Alavi  on partcircular surface flaws. Computations were made by means of Merkle's closed form solution for K for a linear radial stress distribution.  Finiteelement calculations using the ADINA program with a 3-dimensional mesh were also made.  The crack geometry assumed for these calculations is given in Fig. 9 and the residual stress distribution in Fig. 10.
The residual stresses present in the uncut cylinder were treated in the caicuiaLiun oi K as though they resulted from a remotely applied load, which was believed to be a reasonable assumption for a crack that did not extend far with respect to the size of the residual stress field. The pretest distribution shown in Fig. 10 fits the measured values at the circumferential centerlinc of the flaw and was assumed to be constant axially.
The results of the pretest calculations shown in Figs. 11 and IP. gave an indication of the margins for uncertainty that could exist in a comparison of theory and experiment. The "measured o " line in Fig. 11 represented our best pretest prediction of K for the initial crack, with residual stresses taken into account; while the "a = 0" line was considered a reliable estimate of K in the absence of residual stresses. Hence, if the effect of residual stresses were grossly overestimated, but K was correctly assessed, the failure should occur at point A, for which the pressure is below the initial yield pressure. However, an underestimation of K could result in yielding before failure (point B) and an inconclusive test result. This situation demanded care in measuring K and in assuring that the vessel was cooled to —18 C C or below.
The closed form solution for K is compared with the finite element results in Fig. 12 for the same crack depth and the same residual stress distribution and load.
VESSEL TESTThe test vessel was instrumented with strain gages, crack-opening displacement gages, pressure transducers, crack propagation gages, thermocouples, and acoustic emission detectors. The vessel was cooled to about —20°C before the start of pressurization; the temperature of the test section was maintained within ^3K of —23°C during the test. The pressurizing medium was a mixture of ethylene glycol and water.
The pressurization proceeded as shown in Fig. 13. The crack popped in and arrested without causing leakage at 26.3 MPa. Continued pressurization resulted in a second pop-in and fluid leakage at 65.3 MPa.