13th Annual (International) Mechanical Engineering Conference – May 2005 Isfahan University of Technology, Isfahan, Iran Influence of initial stresses on cleavage fracture of pressure vessel containing a part circumferential surface crack Saeid HadidiMoud 1, David Smith 2 Mechanical Engineering, University of Bristol, BS8 1TR, UK 1,2 Mechanical Engineering, Ferdowsi University of Mashhad, Iran 1 Saeid.HadidiMoud@Bristol.ac.uk , David.Smith@Bristol.ac.uk Abstract One critical issue in the integrity assessment of pressurised vessels to be addressed is characterising the influence of initial residual stresses (arising from fabrication processes especially around welds, where the appearance of cracks is also a likely event) on cleavage fracture. Using a specific crack configuration, this paper explores the stress state at crack tip area under similar loading conditions with and without the presence of initial stresses as well as their interaction and compares the stress fields in the context of stress based statistical modelling of cleavage fracture toughness. A recently developed local stress based approach to failure probability prediction for cleavage fracture of ferritic steels is used to predict toughness distributions. It is found that the presence of initial residual stresses dramatically change the final fracture conditions. Keywords: Interaction, residual stress, cleavage fracture, pressure vessel, finite element method Introduction Proof loading or autofrettage, where no defects or initial residual stresses are assumed, improves the load carrying capacity of pressure vessels. This is principally because of local compressive residual stresses generated at the crack tip. In presence of cracks the benefit from proof loading depends on the combination of crack geometry and loading conditions. If there exist residual stresses in a vessel containing defects, the stress field due to the applied mechanical loading will interact with the residual stresses. Determination of fracture parameters in residual stress fields has received great interest in recent years [1]. Local statistical approaches to predicting cleavage fracture toughness, mostly based on the Weibull distributions, have been developed [2-4]. Recently the authors developed a local stress based model that suggested using parameters calibrated based on material response at the specified test temperature regardless of the crack /loading configuration [5]. This model was used to predict the influence of warm pre-stressing on cleavage fracture in standard fracture mechanics specimens (e.g. C(T) and SENB) [6]. Recently the model was 1 2 examined to explore its applicability to cases where initial residual stresses interact with the mechanical loading [7,8]. Initial RS field were generated by either local compression, proof loading, or in plane compression and the test specimens were subsequently subjected to proof loading prior to final loading to fracture [9] at a temperature lower than the proof load. This paper further explores the role of the type and the nature of initial residual stress fields on the subsequent resistance to fracture for a ferritic steel pressure vessel structure. This study is a continuation of the work reported by authors on use of the local approach to predict failure probability for a cracked vessel with prescribed initial stress [10]. The Finite Element Simulations The model used in this study was a 432 mm diameter pipe with a wall thickness of 19.6 mm. The dimensions were adopted from a test configuration reported by Bouchard and Bradford [11]. Depending on the type of loading, whole or half pipe FE models were produced in ABAQUS/CAE and used in the corresponding analyses. In case of symmetry, only - Assistant Professor, Ferdowsi University, Iran / visiting fellow, University of Bristol, UK; Dr. - Faculty Director of Research and Head of Structural Integrity; University of Bristol, UK; Prof. 13th Annual (International) Mechanical Engineering Conference – May 2005 Isfahan University of Technology, Isfahan, Iran one half of the cylinder was modelled with the plane of symmetry in the crack plane normal to the pipe axis. The whole pipe was modelled for use with general loading schemes where symmetry rules did not apply. In the whole model the partial circumferential crack was introduced artificially by the addition of new nodes at the coordinates coincident with the nodes representing the crack face and assigning those nodes to elements on one side of the crack. For the half model the crack was identified by assigning boundary conditions to the remaining ligament at the crack plane, the plane of symmetry. The partial circumferential crack was introduced on the outer surface of the pipe as shown in Figure 1. The crack front was 50 mm wide and 5mm through the wall at the deepest point. The material used in the analyses was A533 steel for which appropriate models were available from previous research [12]. In all cases the introduction or generation of residual stresses as well as proof loading took place at room temperature, whereas the material was cooled to – 170C before final loading to fracture. Based on the results of an initial displacement controlled analysis, the load level was adjusted for load controlled analyses so that the level of prior loading resulted in significant level of plastic deformation at the crack tip region thus producing a residual stress field of noticeable extent. The initial analysis was also used to identify the appropriate node/element sets at the crack front area in order to obtain the stress data files containing the crack front plastic zone for use in the local approach to cleavage fracture [5]. Thus stress and plastic strain data from ABAQUS results were only extracted over the region required and were stored in a data file for use with the failure prediction model. Throughout the load-controlled analyses an appropriate amplitude function was used to control the load application scheme. The FE simulations were conducted in two stages. Firstly the as-received (AR) or cool-fracture (CF) and load-unload-cool-fracture (LUCF) analyses were performed. In the next stage, initial residual stress fields were introduced and the local approach was to predict fracture probability in presence of initial stresses. A series of analyses were undertaken as summarised below. Trial analysis: an un-cracked pipe with and without the initial residual stresses (IS) subjected to internal pressure until plasticity spread throughout the wall. As received (AR) or cool-fracture (CF): Cracked pipe with no residual stress subjected to axial loading to fracture at low temperature Load-unload-cool-fracture (LUCF): Cracked pipe with no residual stress subjected to axial loading to fracture at low temperature following WPS Initial stress-cool-fracture (ISCF): Cracked pipe with IS subjected to axial loading to fracture at low temperature Initial residual stress-load-unload-cool-fracture (ISLUCF): Cracked pipe with IS subjected to axial loading to provide proof loading at room temperature, unloaded and then reloaded to fracture at low temperature (CF) Finally the results for a whole range of analyses were used in conjunction with a stress based local approach to fracture [5], to explore the predictions for probability distributions and the consistency of the predictions with the stress distributions. Description of analyses and results The residual stresses introduced to the cracked pipe were produced at two levels. “Short range” or “local” IS fields were generated by carrying out separate finite element simulations of load-unload cycles and editing the resulting stress distributions for use as initial conditions in the main analyses. Prescribed IS fields were also artificially generated using an arbitrary IS file and re-distributing over an initial analysis step. The latter procedure was used to introduce IS fields of significant level and extent. Analytical stress intensity factors for crack configurations in cylindrical vessels are available from Anderson [13] and others [14-16]. An approximate solution taking into account the finite size effect was used to estimate K values. Fracture toughness was calculated from incremental fracture loading at low temperature during the last step of each analysis for the specific crack configuration. The local approach predictions for fracture in the cylinder used the same Weibull parameters calibrated for the C(T) specimen in as-received conditions [5]. The failure probability distribution was calculated using the routine “local”, and plotted against stress intensity (or fracture toughness in the context of probability). The predictions were promising especially with regard to the validation of calibrated Weibull parameters for various configurations of geometry and crack for the same material. The distribution obtained for cylinder was found very much consistent with those of previous research for standard fracture mechanics specimens (C(T) and SENB) [5]. This is illustrated by comparing the as received distributions for the two problems for A533B steel loaded to fracture at – 170°C in figure 2. 13th Annual (International) Mechanical Engineering Conference – May 2005 Isfahan University of Technology, Isfahan, Iran Various analyses as described were performed and required data was extracted for the examination of the local approach. The routine “local” used the material yield point in all cases to evaluate the Weibull stress hence the failure probability distributions. Consistent with C(T) test results, significant increase in fracture toughness distribution resulting from LUCF was observed (figure 2). A series of analyses were performed for the ISCF and ISLUCF cycles. For low levels of initial residual stresses at the crack tip area, the shift in failure probability distributions were only marginal. However, the influence of initial residual stresses became more obvious where sufficiently high level of tensile stress normal to the crack plane and around the crack front was introduced. The crack front initial stress field remained significant when rebalanced through an analysis step. The results of these analyses were also used to predict fracture distribution using the local approach. Significant shift in predictions was observed when compared with the predictions corresponding to the effect of low-level initial stresses. This is shown in Figure 3. The required data files for the routine local were also prepared for all the analyses and the probability distributions were predicted. As figure 3 indicates the predictions are apparently different from the predictions found earlier in presence of secondary residual stresses in C(T) specimens [7]. Interaction of RS distribution with subsequent fracture loading Interaction of initial residual stresses (distribution / redistribution) with the crack development process was also studied. Autofrettage loading was applied to the cylinder using the boundary conditions that represent the un-cracked cylinder. Using a loadunload (LU) cycle, a global RS field was generated. The RS field was then used to create a file containing stresses at significant level, both in magnitude and the extent. This stress field was then used as initial conditions. Subsequently, the crack was introduced and the residual stresses were allowed to rebalance. Cool-fracture (CF) and proof load –cool-fracture (LUCF) loading cycles were then applied in presence of the incorporated initial RS fields. For all the cases described above axial and hoop stresses throughout the analyses were obtained to explore the differences in fracture response based on changes in the stress-state. Results are discussed in the next section in relation to the local approach predictions for the corresponding analyses. Finally a separate analysis was performed similar to ISCF as described above, but using an initial RS field defined intentionally with significant level of axial stress and no other stress component. Introducing such stress field into a cracked cylinder and redistributing the RS through one analysis step using the elastic material response, resulted in appearance of hoop stresses of significant level and also dramatic reduction in the initially introduced axial RS component when the RS field was redistributed due to the crack introduction. Comparison of the stress level at similar loading with the analyses that used low level of residual stresses, suggests that much further influence of the initial global RS field on the proof load effect is expected. This was the case both when initial RS was followed by LUCF and when it was followed by CF. probability distribution predictions presented in figure 3, suggested that unlike the secondary RS field, a global initial RS field with sufficient level and extent does influence the final fracture conditions. The completed analyses include fracture loading of the cylinder with or without initial residual stress field and in both cases with or without proof loading. The residual stresses as well as the stress distributions on final loading step for various analyses were monitored and compared. Normal to the crack face residual stresses, axial for cylinder, and the hoop residual stress distributions through the thickness of the pipe wall are plotted in figures 4a and 4b respectively. Similarly stress distributions on reloading for identical load levels for various analyses are shown in figures 5a and 5b. Discussion of results and conclusions The performed analyses indicate that the nature of IS introduced has significant influence on subsequent fracture. This is believed to be due to the nature of the RS field that was introduced as initial condition. Where the RS field was generated by means of autofrettage of un-cracked cylinder, the level of axial RS was forced to remain significantly lower than hoop RS to maintain equilibrium conditions. The analyses results were used to explore the stress distributions throughout various load cycles and to extract the required input data for the routine ”local” to investigate the effect of global RS field on the proof loading effect and to compare the effect with previous analyses. In earlier study it was shown that secondary residual stresses used as initial RS field had effectively no influence on the final fracture conditions following proof loading as the RS field was completely wiped out as a result of pre-loadunload stage of the loading cycle [7]. The analyses performed based on the introduction of initial low 13th Annual (International) Mechanical Engineering Conference – May 2005 Isfahan University of Technology, Isfahan, Iran level residual stress field (created from load-unload of cracked cylinder as described) are essentially considered as a verification of previous findings obtained from the C(T) model Analyses [7]. However, the analyses performed based on the introduction of initial residual stress field of significant extent, i.e. arbitrary introduced high axial RS field to the un-cracked cylinder followed by crack introduction and rebalancing, suggested dramatic decrease in the probability distribution as shown in figure 3. From figure 4 it can be concluded that there exists consistent residual stress distributions, especially in the normal to crack direction, and their corresponding failure probability distributions in all analyses as shown in figure 3. From figure 5 it can be concluded that this consistency is also the case for stress distributions for similar load levels for various analyses and their corresponding probability distributions. The present study only looks at a specific crack configuration whereas experimental observations suggest that cracks with different geometry and direction are likely to appear in pressure vessels. A comprehensive study is underway that considers further simulations for various crack configurations to further explore the stress and failure issues. These investigations when put together would provide a framework for possible suggestions regarding the improvements in currently established integrity assessment procedures. Currently available integrity assessment codes include design curve approaches, FAD approach according to R6/BS 7910, fully plastic solutions of EPRI type, CEGB reference stress method, the engineering treatment model and the European SINTAP procedure [17]. The results of these studies may be compared to those of the established methods in order to assess the credibility of the local approach in fracture prediction. References Lei, Y., “J-integral evaluation for case involving non-proportional stressing”, Engng. Fract. Mech., 2004, in press. 2. Margolin B Z, Kostylev V I, Keim E, “Prediction of brittle fracture of RPV steels under complex loading on the basis of a local probabilistic approach”, Int. J Pressure Vessels and Piping, 81, 2004, pp 949-959 3. Gao X, Zhang G and Srivatsan T S, “Prediction of cleavage fracture in ferritic steels: a modified Weibull stress model”, Materials Science Engineering, 2005, in press 4. Bordet S R, Karstensen A D, Knowels D M and Wiesner CS, “A new statistical local criterion for cleavage fracture in steel. Part I: model presentation”, Engineering Fracture Mechanics, 72, 2005, pp 435-452 1. Hadidi-Moud, S., Mirzaee-Sisan, A., Truman, C.E., and Smith, D.J.,“A local approach to cleavage fracture in ferritic steels following warm prestressing,” Fatigue Fract. Engng. Mater. Struct., 27, 2004, pp 931-942 6. S Hadidi-Moud, A Mirzaee-Sisan, C E Truman and D J Smith, “Comparison of Global and Local Approaches to Predicting Warm Pre-stress effect on Cleavage Fracture toughness of ferritic steels”, Key Engineering Materials, Volume 261, 2004, pp 69-74. Advances in Fracture and Failure Prevention, ISBN: 0-87849-938-5 7. S Hadidi-Moud, A H Mahmoudi, C E Truman and D J Smith: “Combined Effect of Residual Stress and Loading History on Brittle Fracture”, ICM9: 9th Int. Conference on Mechanical Behaviour of Materials, Geneva, Switzerland, 25-29 May, 2003, in proceedings 8. A Mirzaee-Sisan, S. Hadidi-Moud, CE Truman and D.J Smith, “Application of the Local Approach to predict brittle fracture following Local Compression”, in proceedings, 15th European Conference on Fracture (ECF15), August 2004, Sweden, in proceedings 9. Mirzaee-Sisan A, Mahmoudi A H, Truman C E and Smith D J, “Application of the local approach to predict load history effects in ferritic steels”, Proceedings of PVP2005: ASME Pressure Vessels and Piping Division Conference, July 17-21, 2005, Denver, Colorado, PVP2005-71606 10. S. Hadidi-Moud, CE Truman and D.J Smith: “Interaction of mechanical loading with residual stresses in pressure vessels”, APCFS04, Oct 2004, Jeju, S Korea, to appear in Key Engineering Materials, 2005 11. Bouchard, J. and Bradford, C., “Validated Axial Residual Stress Profiles for Fracture Assessment of Austenitic Stainless Steel Pipe Girth Welds,” PVP2001, Vol. 422: Fracture and Fitness, Atlanta, USA, 2001 12. Smith DJ, Hadidi-Moud S and Fowler H, “The Effects of Warm Pre-Stressing on Cleavage Fracture, Part 1: Evaluation of Experiments”, Engineering Fracture Mechanics, Volume 71, Issues 13-14, 2004, pp 2015-2032 13. Anderson T L, Fracture Mechanics (2nd edition), CRC Press, 1994 14. Lin X B and Smith R A, “ Stress intensity factors for semi elliptical internal surface cracks in autofrettaged thick-walled cylinders”, Journal of Strain Analysis, vol. 32, No 5, 1997, pp 351-363 15. Fett T, “Estimation of stress intensity factors for semi-elliptical surface cracks”, Engineering Fracture Mechanics, 66, 2000, pp 349-356 16. Brighenti R, “Surface cracks in shells under different hoop stress distributions” Int. J Pressure Vessels and Piping, 77, 2000, pp 503-509 5. 13th Annual (International) Mechanical Engineering Conference – May 2005 Isfahan University of Technology, Isfahan, Iran 100 AR; C(T) AR 80 Failure Probability; % 17. Zerbst U, Schwalbe K H and Ainsworth R A, “Comprehensive structural Integrity,” vol. 7, pp.289347, Edited by Ainsworth, R.A., Schwalbe, K. –H., Elsevier Ltd., 2003 ISCF LUCF ISLUCF 60 40 20 0 0 20 40 60 80 100 120 Fracture Toughness; MPa-m0.5 Figure 3: Failure probability predictions for various analyses for A533B steel cracked pipe at -170C Figure 3: Failure probability predictions for various analyses for A533B steel cracked pipe at -170C 600 Initial RS-no crack RS-Cracked-no LU RS-Cracked-LU no RS-Cracked-LU Axial RS, initial; MPa 400 200 0 -200 -400 -600 -800 Figure 1: Geometry and crack details for the finite element model 0 1 2 3 4 5 6 7 8 9 10 distance from outer surface; mm Fig 4a. Axial residual stresses for various analyses 100 AR; C(T) AR or CF ; Pipe LUCF; Pipe 300 Initial RS-no crack 60 RS-Cracked-no LU RS-Cracked-LU no RS-Cracked-LU 200 40 Hoop RS; MPa Failure Probability; % 80 20 0 0 20 40 60 80 100 0 120 Fracture Toughness; MPa-m0.5 Figure 2: Comparison of AR predictions for CT and pipe 100 Figure 2: Comparison ARsteel predictions for and the effect of WPSof for A533B at -170C CT and pipe and the effect of local RS and WPS for A533B steel at -170Cf -100 -200 0 1 2 3 4 5 6 7 8 distance from outer surface; mm Fig 4b. Hoop residual stresses for various analyses 9 10 13th Annual (International) Mechanical Engineering Conference – May 2005 Isfahan University of Technology, Isfahan, Iran 1000 AR or CF AR or CF LUCF 800 LUCF 400 ISCF ISCF Hoop stress; MPa Axial stress; MPa ISLUCF ISLUCF 600 400 200 0 200 0 -200 -200 0 1 2 3 4 5 6 7 8 9 10 Distance from outer surface; mm Fig 5a: Comparison of Axial stress field on reloading Fig 5a: Comparison of Axial for various analysesstress field on reloading for various analyses 0 1 2 3 4 5 6 7 8 9 10 Distance from outer surface; mm Fig 5b. Comparison of Hoop stress on reloading Fig 5b. Comparison of Hoop stress field on for various analyses reloading for various analyses