THE EFFECTS OF FLAW SIZE AND IN-SERVICE INSPECTION ON CASS PIPING RELIABILITY

Transcription

1 THE EFFECTS OF FLAW SIZE AND IN-SERVICE INSPECTION ON CASS PIPING RELIABILITY T. J. Griesbach, D. O. Harris, H. Qian, D. Dedhia, J. Hayden, Structural Integrity Associates, Inc. A. Chockie, Chockie Group International, Inc. ABSTRACT Managing aging of Cast Austenitic Stainless Steel (CASS) piping per the generic aging lessons learned report (GALL) requires a combination of flaw tolerance evaluations and examinations for nuclear plant primary system piping. It has well been established that the toughness of CASS components degrade with time as a result of thermal aging resulting in reduced flaw tolerance of the components. However, in the thermally aged condition, most CASS components have very adequate toughness to be able to operate. The flaw tolerance of these components is influenced by various factors. However, many studies that have been conducted to date have used deterministic fracture mechanics models to evaluate critical flaw sizes and flaw stability margins. More recently, a probabilistic fracture mechanics approach was developed under an EPRI-funded project to determine probabilities of failure considering the various scatter in unknowns (e.g., material toughness properties, tensile properties, flaws and flaw sizes.), and sensitivity studies have been performed to quantify the contributions of these key input parameters. An area of uncertainty is the effectiveness of examinations using UT techniques. Because of the coarse microstructure, effective UT examinations of CASS components have been a challenge in the industry over the years. In view of the relatively high flaw tolerance of CASS components even in the thermally aged condition, the question becomes how much improvement in examination techniques is required for adequate aging management of these components. Use of improved UT examination techniques are now being proposed in an ASME Section XI Code Case, and corresponding flaw tolerance methods are being developed for analysis of the CASS piping materials. This paper presents some initial results based on the probabilistic fracture mechanics models that can be useful as a technical basis to develop effective aging management programs for CASS piping. BACKGROUND The purpose of this paper is to provide a brief review of probabilistic fracture mechanics (PFM) models of benefits of inspection, and to discuss implications related to cast austenitic stainless steel (CASS) piping in commercial power reactors. There are many advantages to using the PFM approach. The usual tendency is to use deterministic methods and assume worst case (or bounding) toughness and loads and perform an analysis to establish a maximum allowable flaw size. This often results in very conservative results for flaw sizes (depths and lengths). In the case of thermally aged CASS piping materials, such conservatively sized flaws may be unrealistic in terms of flaw detection capabilities. An alternative is to employ PFM and define all inputs as random variables, then perform Monte Carlo simulations to compute a final result as a probability of failure. Once an accepted failure probability is set, the allowable flaw depths and lengths can be defined for similar load conditions. Without having such a tool, it is not always obvious which set of potential optional paths provides the better solution. A logical extension of the PFM work is to consider how inspection reliability can influence the failure probability, which is examined here in more detail. REVIEW OF PFM MODEL WITH INSPECTION The simplest probabilistic fracture mechanics model that includes inspection would consider crack depth and detection probability as the only random variables. The critical crack depth (a c )is deterministically defined, and the probability of failure is the probability of the crack depth exceeding the critical value. Denote the probability density function of initial (as-fabricated) crack depths, a, as p o (a). Consider this to 200

2 be conditional on a crack being initially present, so it integrates to unity. The probability of failure is the probability of having a crack deeper than the critical depth, a c. For a body of thickness h, the failure probability can then be written as h P = p ( a) da (1) f ac Inspection comes in through the probability of detecting a crack as a function of depth. It is more convenient to consider the nondetection probability, (a), which is unity minus the probability of detection. The post inspection crack depth distribution is then po ( a) PND ( a), and the failure probability following the inspection (given that a crack is present) is h o P ND P = p ( a) P ( a) da (2) f ac o Crack growth during service, such as fatigue crack growth, is treated by considering the time (cycles) for a crack to grow from its initial depth to the critical depth. In this simple example, this is deterministic. Figure 1 depicts the procedure. ND Figure 1: Schematic Representation of Procedures Involved in Calculating Failure Probability for a One-Dimensional Crack Problem The as-fabricated crack depth distribution is shown, along with the post-inspection distribution, and the crack depth distribution after a time in service, t s. The process is easily expanded to multiple inspections with intervening crack growth [1]. It is also 201

3 possible to consider the loads, crack growth rates, fracture toughness and other inputs to be random, but numerical techniques, such as Monte Carlo simulation, must be employed. Pursuing the simple model, it can be shown [1] that the relative influence of inspection for multiple inspections can be written as P k f ( t) with inspection = PND [ an ( t) ] (3) P ( t) f noinspection In this expression, a n (t), is the depth of a crack at the time of the n-th inspection that would just grow to ciritical size at the current time. Equation 3 expresses the interesting result that the relative benefit of inspection does not depend on the distribution of the initial crack depth. n= 0 RELIABILITY OF INSPECTIONS NDE techniques do not detect all of the cracks all of the time. They can be thought of as detecting cracks with a probability that depends on crack size, as expressed above as P ND (a). Detecting a crack depends on many factors, including the dominant degradation mechanism, defect type and morphology, flaw location (e.g., surface or buried), orientation, material type (static or centrifugally cast) and grain structure, inspector training and experience, and inspection procedures. The Probability of Detection (POD) has been used to define the capability of a specific NDE technique and inspection team. For a given situation, the POD curve is a function of defect size (depth or area). For these reasons, the POD curve is a measure of inspection reliability. The probability of non-detection is determined to be P ND (a) = [1 POD(a)], so the higher the POD the more reliable the inspection technique. Different techniques will have differing PODs. (False positive calls are not considered in this simplified discussion.) In general, the POD curve for a certain NDE method, procedure, and inspection team is estimated based on a statistical methodology using either experimental data, expert elicitation, or both data and elicitation. The POD curves based on expert elicitation [2,3] are used here for illustration. The approach was to postulate four POD curves that represented widely different level of NDE performance. These curves were intended to bound the performance expected from inspection teams operating in the field. To establish parameters for the POD curves, an informal expert judgment was made using staff with knowledge of NDE performance data from inspection round robins and of recent industry efforts in the area of NDE performance demonstrations. It was recognized that a population of inspection teams operating under field conditions can exhibit a considerable range of POD performance, even though all such teams have successfully completed a performance demonstration. The basic premise in estimating POD curves was that teams had passed the ASME Section XI performance demonstration. ASME Section XI requires a minimum of 10 flaws and 20 blanks to be examined. A team who does not detect at least 8 of the 10 flaws (i.e., makes more than two false calls) receives a failing grade. The informal expert judgment also considered information and trends observed in the Pacific Northwest Laboratory (PNL) mini-round robin on UT inspection of wrought stainless steel [2, 4, 5]. NDE experts were asked to define POD curves by estimating parameters for the following form of a PND function: 1 a P ND ( a) = ε + 2 ( 1 ε ) erfc ν ln (4) a * where P ND is the probability of nondetection, a is the depth of the crack, a* is the depth of the crack for 50% P ND, is the smallest possible P ND for very large cracks, and is the slope of the P ND curve. Four POD curves were selected to characterize various levels of NDE reliability: Marginal Performance In the judgment of the PNL experts, a POD performance described by this curve would represent a team using given equipment and procedures that would have only a small chance of passing a performance demonstration. Good Performance A POD described by this curve corresponds to a better performance level in 202

4 the PNL round robin. Very Good Performance This curve corresponds to a team (with given equipment and procedures) that significantly exceeds the minimum level of performance needed to pass a performance demonstration test. Advanced Performance This curve describes a level of performance significantly better than expected from present-day teams, equipment, and procedures that have passed a performance demonstration. This performance level implies advanced technologies and/or improved procedures that could be developed in the future. Table 1 summarizes the input data for the above four POD curves, which are shown in Figure 2. These particular curves assume that POD is a function of the crack depth as a fraction of pipe wall thickness, independent of the actual wall thickness. Parameters indicated in Table 1 are considered appropriate to wall thicknesses of 1.0 inch (2.54 cm) and greater. Table 1. POD Curve Parameters for Four Performance Levels Inspection Performance Depth, a*/ h 1 Level Marginal Good Very Good Advanced h is the wall thickness of the pipe Figure 2. Example Probability of Detection Curves. EFFECT OF NDE ON PROBABILITY OF FAILURE As an example of analysis of benefits of inspection, Figure 3 presents results based on Reference 6 for failure probability with and without inspections with various inspection capabilities. 203

5 1 D:\FortranStuff\DET1.PLT 0.1 failure probability with inpsection good low advanced failure probability no inspection Figure 3: Influence of Inspection on Failure Probability for Various Inspection Performance Levels The low quality inspection provides hardly any benefit, whereas the advanced inspection provides orders of magnitude improvement in the failure probability. PROBABILISTIC FLAW TOLERANCE OF CASS PIPING The tensile and fracture properties of cast austenitic stainless steel exhhibit much scatter. Furthermore, these properties change with time at temperature, as discussed and documented in References 7 and 8. The decrease in toughness and increase in yield and tensile strength reaches a saturation point beyond which no further changes take place. In commercial power reactors, this is usually within a few years. Fully saturated conditions are concentrated upon here, and the CF8M material is considered, because of the piping materials employed, this is the one that degrades the most. The considerable scatter in material properties calls for a probabilistic approach, especially since considerable data is available. References 9 and 10 describe the probabilistic models of CASS piping reliability that have been developed, based primarily on data and correlations drawn from References 7 and 8. Figure 4 provides an example of the results obtained for three failure probabilities. The results are for a circumferentially oriented part-through flaw in a 32-inch outer diameter pipe with a thickness of 2.25 inches. The loading considered imparts a tensile (membrane) stress of 8 ksi and a bending stress of 15 ksi. This loading is believed to be towards the high end of values encountered in these large pipes during normal plant operation. Chemical composition and delta ferrite content of the material in this example are based on typical values, as detailed in References 9 and

6 Figure 4: Flaw Sizes That Would Fail with a Given Probability When the Loads Considered Are Applied Figure 4 shows that the flaw size at which failure would occur increases with increase in failure probability, and that the tolerable flaw sizes generally exceed about 40% or the wall thickness (at probability levels exceeding 10-6 ). INFLUENCE OF INSPECTION The results of Figure 4 show that depending on the loading, the cast austenitic stainless steel piping can be quite flaw tolerant; for the example considered, flaws approaching half way through the wall and of considerable circumferential extent would fail with a probability of 10-6 when the loads considered are applied. This value can be reduced by inspection. Considering a single inspection, Equation 3 can be used to estimate the reduction in failure probability due to inspection. Figure 4 gives the failure probability and crack sizes for no inspection, and Figure 2 gives the detection probabilities for various crack sizes for different inspection performance levels. Results from these two figures can be combined to provide the results in Figure 5 for the failure probabilities for θ / π of 0.2 and 0.7 for the four inspection performance levels and three failure probabilities in these figures. 205

7 Figure 5: Failure Probability for Different Inspection Qualities and Values with no Inspection SUMMARY The considerable scatter in toughness and tensile properties of thermally aged cast austenitic piping materials necessitates a probabilistic approach, rather than the more conventional conservative deterministic approach that results in the need to detect small flaws. The probabilistic fracture mechanics analysis predicts that, for PWR cold leg pipe under normal operating conditions, flaws half-way through 206

8 the thickness and considerable circumferential extent would result in a 10-6 failure probability. Under such conditions, even a marginal inspection would be adequate for effective management of CASS components to assure that piping integrity is maintained. For cases where the loading conditions are more severe, or if lower probabilities of failure must be maintained, then further improvements in the inspection techniques may be required for effective aging management of CASS components. REFERENCES 1) D. O. Harris and E. Y. Lim, Applications of a Probabilistic Fracture Mechanics Model to the Influence of In-Service Inspection on Structural Reliability, Probabilistic Fracture Mechanics and Fatigue Methods; Applications for Structural Design and Maintenance, ASTM STP 798, J. M. Bloom and J. C. Ekvall, Eds. American Society for Testing and Materials, pp , (1983) 2) M. A. Khaleel, F. A. Simonen, The Effects of Initial Flaw Sizes and Inservice Inspection on Piping Reliability, PVP-Vol. 288, Service Experience and Reliability Improvement: Nuclear, Fossil, and Petrochemical Plants, ASME (1994). 3) M. A. Khaleel, F. A. Simonen, D. O. Harris, and D. Dedhia, The Impact of Inspection on Stress Corrosion Cracking for Stainless Steel Piping, PVP-Vol. 296/SERA-Vol. 3, Risk and Safety Assessment: Where is the Balance?, ASME (1995). 4) P.G. Heasler, T. T. Taylor, J. C. Spanner, and S. R. Doctor, Ultrasonic Inspection Reliability for Intergranular Stress Corrosion Cracks: A Round Robin Study Effects of Personnel, Procedures, Equipment and Crack Characteristics, NUREG/CR-4908, U. S. Nuclear Regulatory Commission, Wash. D.C. (1991). 5) T. T. Taylor, J. C. Spanner, P. G. Heasler, S. R. Doctor, and J. D. Deffenbaugh, An Evaluation of Human Reliability in Ultrasonic Inservice Inspection for Intergranular Stress Corrosion Cracking Through Round Robin Testing, Materials Evaluation, Vol. 47, p. 338 (1989). 6) P. Dillstrom, et al., A Combined Deterministic and Probabilistic Procedure for Safety Assessment of Components with Cracks Handbook, 2008, available at 7) W.F. Michaud, P. T. Toben, W. K. Soppet and O. K. Chopra, Tensile Property Characterization of Thermally Aged Cast Stainless Steels, U.S. Nuclear Regulatory Commission Report NUREG/CR- 6142, ) K. Chopra, Estimation of Fracture Toughness of Cast Stainless Steels During Thermal Aging in LWR Systems, U.S. Nuclear Regulatory Commission Report NUREG/CR-4513, Rev. 1, ) H. Qian, D. Harris and T.Griesbach, Probabilistic Models of Reliability of Cast Austenitic Stainless Steel Piping, 2011 ASME Pressure Vessels and Piping Division Conference, Paper PVP ) H. Qian, D. Harris and T.Griesbach, Probabilistic Models of Reliability of Cast Austenitic Stainless Steel Piping, 2012 ASME Pressure Vessels and Piping Division Conference, Paper PVP