Proceedings of the 2012 9th International Pipeline Conference IPC2012 September 24-28, 2012, Calgary, Alberta, Canada

IPC2012-90334

MODEL ERROR ASSESSMENT OF BURST CAPACITY MODELS FOR DEFECTFREE PIPES

W. Zhou Western University London, Ontario, Canada G. (Terry) Huang TransCanada Pipelines Calgary, Alberta, Canada

ABSTRACT The model errors associated with 19 burst pressure prediction models for defect-free thin-walled pipes are evaluated using a total of 76 full-scale burst test data of perfect pipes and pressure vessels collected from the literature. The considered models are based on the Tresca yield criterion, the von Mises yield criterion, or the average shear stress yield criterion. The probabilistic characteristics of the model error, i.e. the mean, coefficient of variation and best-fit probability distribution, are obtained based on the test-to-predicted ratios. The applicability of an empirical equation for estimating the strain hardening exponent in the burst capacity models is also evaluated. The model errors obtained in this study can be used in the structural reliability analysis of energy pipelines with respect to the limit state of burst of defect-free pipes and will facilitate the reliability-based design and assessment of pipelines. INTRODUCTION According to the current oil and gas pipeline design codes in the US and Canada, the pipe wall thickness is determined using the safety factor approach that can be written in a general format as follows:

P = FS

2t SMYS D

(1)

where SMYS is the specified minimum yield strength; P is the pipe design pressure; D and t are the outside diameter and wall thickness of the pipe respectively, and FS is a factor of safety. According to Eq. (1), the pipe wall thickness is determined by applying the safety factor against yielding due to the pipe internal pressure. In practice, yielding of a perfect (i.e. defectfree) pipe due to internal pressure has very little, if any, adverse consequences. On the other hand, there are significant safety and economical implications associated with the burst of a

perfect pipe. The burst pressure of a perfect pipe is defined as the maximum internal pressure the pipe can bear, i.e. the internal pressure at plastic collapse (the occurrence of unbounded plastic deformation in the pipe). Burst of a perfect pipe, which is mainly due to inappropriate operation or failure of the operating equipment, is a relatively rare event in reality. Nevertheless, the burst pressure of a perfect pipe is an important consideration for oil and gas pipelines because it defines the safety margin for defect-free pipelines under the operating pressure. As high-strength steels are being increasingly used for new pipelines, the burst pressure of perfect pipes becomes more important. This is because highgrade pipes typically have higher values of yield-to-tensile strength ratios. Given that the same safety factor is being used in Eq. (1) to determine the pipe wall thickness for both lowgrade and high-grade pipes, the safety margin against burst for high-grade pipes can be significantly less than that for lowgrade pipes. Reliability-based methodologies are becoming more and more accepted by the pipeline industry as an alternative to the safety factor approach in the design and assessment of oil and gas pipelines over the last two decades. Compared with the safety factor approach, reliability-based methodologies are associated with a number of advantages including consistent safety levels, cost savings and ability to deal with unique design situations (e.g. frost-heave and thaw-settlement loadings on pipelines in the Arctic areas) [1]. Recently, a set of target reliability levels and guidelines for the application of reliability-based design and assessment (RBDA) for onshore natural gas pipelines has been developed [2] and adopted by the current edition of the Canadian pipeline standard, CSA Z66207 [3], as a non-mandatory annex, Annex O. RBDA requires that the failure probability of a given pipeline due to all applicable failure causes (e.g. burst of the perfect pipe due to internal pressure and corrosion) not exceed the maximum allowable failure probability or one minus the target reliability level. Therefore, the burst probability of a defect-free pipe

Copyright 2012 by ASME

under internal pressure can be an important consideration in RBDA. A large number of empirical and theoretical models have been developed to predict the burst pressure of thin-walled and thick-walled pressure vessels since the beginning of the 20th century. Comparative studies have also been carried out [4-8] to evaluate the accuracy of the existing models. Christopher et al. [4] reviewed 12 burst pressure prediction models for pressure vessels and compared the model predictions with a series of burst test data collected from the literature. The test data consist of a variety of metallic materials (i.e. different steels, alloys and aluminums). They observed that there is not a unique model that consistently achieves higher prediction accuracy than the other models for all the different types of materials included in the test data. A review of the original sources for the test data selected by Christopher et al. indicates a number of questionable aspects in their study. For example, the values of the parameters such as the strain hardening exponent and tensile strength used in the prediction for certain test data were not reported in the original sources and therefore appear to be assumed by Christopher et al. Furthermore, they selected a subset of the test data reported in a number of references without providing the rationales for the data selection, whereas the test data being excluded appear to be equally admissible to their study. Focusing on assessing the accuracy of the burst pressure prediction models for thin-walled perfect pipes, Law et al. [5] carried out a comparative study for 16 models against burst test results reported in [9-10]. Law and Bowie [11] further evaluated 23 existing models by using the results of five burst tests they conducted in a previous study [6]. Zimmermann et al. [7] reviewed 11 burst pressure prediction models for thinwalled pipes by investigating the sensitivity of the ratio between predicted and test burst pressures (i.e. predicted-to-test ratio) to the yield-to-tensile strength ratio and the diameter ratio (i.e. ratio between the outside and inside diameters) and by evaluating the mean and standard deviation of the predicted-totest ratios. Zhu and Leis [8] pointed out that there were analytical errors in the study reported in [11]. It is noted that the above-described studies are mainly focused on evaluating and comparing the accuracies of various burst pressure prediction models, as opposed to characterizing the model errors associated with these models. The probabilistic characterizations of the model errors associated with the burst pressure prediction models can have a large impact on the calculated probability of burst of pipelines. The objective of the study described in this paper was to evaluate the model errors associated with several existing burst pressure prediction models for perfect pipes using a relatively large database of full-scale burst test data collected from the literature. The model errors were characterized based on the test-to-predicted ratios, i.e. the ratios between the burst pressures observed in the full-scale tests and corresponding burst pressures predicted by the various prediction models. The means, standard deviations and coefficients of variation (COV) of model errors for the considered models were evaluated. The

probability distributions for the model errors were recommended based on the distribution fitting techniques. The test-to-predicted ratios were also plotted versus different parameters to investigate the influence of key parameters on the model error.
Table 1 Burst Capacity Models for Perfect Pipes
Model No. 1 2a 2b 2c 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Model Name ASME Barlow OD Barlow ID Barlow Flow Bailey-Nadai DNV Fletcher Max. Shear Stress Turner Stewart et al. (Tresca) Bohm Faupel Marin and Rimrott Marin and Sharma Nadai-1 Nadai-2 Soderberg Svensson Stewart et al. (von Mises) Zhu and Leis (1)
Pmax =

Burst Pressure Equation

k 1 Pmax = u , 0.6k + 0.4 for k < 1.5

Group of Solution

Pmax = u
Pmax = u

2t D
2t Di
2t Di

Pmax = flow

Pmax =

1 1 2n k 2 n

2t 2 , f cb = min y , u f cb 1.15 D t 3
Pmax = Di (1 u / 2 ) 2t flow

Tresca

k 1 Pmax = 2 u k +1

Pmax = u ln (k )
Pmax = t u 2 (n1) Dave
n

t 0.25 e 2t 1 Pmax = u 0.227 + n n Di Di

Pmax =

2 y2 y u 3

ln (k )

Pmax =

2 u ln (k ) 3 (1 + u )

Pmax =

( 3 )(
u

4t

n +1)

u
Di

Pmax

2 u ln(k ) = 3
1 1 2 n 3n k

von Mises

Pmax =

Pmax =

4 k 1 u 3 k +1
n

0.25 e Pmax = u ln (k ) 0.227 + n n u 4t Pmax = ( n +1) Dave 3

( )

18

2+ 3 Pmax = 4 3
2+ 3 Pmax = 4 3

n +1

4t u Dave
0.596

(1+0.239((1 / YT )1) )

19

Zhu and Leis (2)

4t u Dave

Average Shear Stress Yield Criterion

BURST CAPACITY MODELS FOR PERFECT PIPES The burst pressure models considered in this study are summarized in Table 1. In Table 1, Pmax is the burst pressure; D, Di, Dave are the outside, inside and average diameter of a

Copyright 2012 by ASME

pipe respectively; Ri = Di/2; k = D/Di; t is the pipe wall thickness, and y, u, flow are the yield strength, ultimate tensile strength and flow stress of the pipe steel respectively. The flow stress is defined as the average of yield strength and ultimate tensile strength, i.e. flow = (y + u)/2 in Fletchers model [5]; u is the uniform strain defined as the engineering strain corresponding to the ultimate tensile strength; n is the strain hardening exponent and is calculated as n = ln(1+u), which implies that the true stress-strain relationship for the material follows a power law characterization well into the plastic region [12], and YT is the yield-to-tensile strength ratio. Zhu and Leis [8] showed that all the models summarized in Table 1 except for the two models developed by Zhu and Leis [13], when applied to thin-walled pipes, can be generally categorized into two groups: namely Tresca and von Mises solutions. It is also worth noting that the Zhu and Leis (1) model in Table 1, which is based on the average-shear-stress yield criterion, is the average of the Tresca and von Mises solutions, and that the Zhu and Leis (2) model has the same basic burst capacity equation as the Zhu and Leis (1) model but employs an empirical equation reported in [12] to estimate the strain hardening exponent (i.e. n) from the yield-to-tensile strength ratio. The empirical equation to estimate n is given by
1 n = 0.239 1 YT
0.596

(2)

BURST TEST DATABASE OF PERFECT PIPES Full-scale burst tests of defect-free pipes and pressure vessels have been carried out by many researchers over the last forty years. Some of the experimental investigations were carried out to evaluate the significance and influence of the yield strength-to-tensile strength ratio on line pipes (e.g. [9] and [14-16]). More recently, full-scale burst tests have been used to validate the accuracy of burst pressure prediction models for thin-walled perfect pipes (e.g. [5-8], [13] and [17-19]) and to investigate the mechanical properties of the newly emerging high-grade line pipes (e.g. [20-22]). In the present study, fullscale burst test data for defect-free pipes were collected from the literature for the purpose of evaluating the model errors associated with the burst pressure prediction models summarized in Table 1. The following criteria were used to select the full-scale burst test data for defect-free pipes from the literature. The pipe specimens are thin-walled, i.e. D/t 20. The pipe specimens are subjected to internal pressure only. The pipe specimens are defect-free. The length of the specimen is greater than twice the outside diameter of the specimen. This criterion ensures that the boundary condition of the test specimen is similar to that of the real pipelines, which are typically long and have negligible axial strains under internal pressure [13].

Sufficient information regarding the actual properties (as opposed to the nominal properties) of the specimens is provided such that a given data point is applicable to at least one of the models summarized in Table 1. It is noted that due efforts were made in the data collection process to obtain, to the extent possible, first-hand information, i.e. the test data reported by the researchers who carried out the tests. This is based on the assumption that the first-hand information is the most reliable and least prone to errors. Furthermore, crosschecking was carried out among different sources for the same set of test data to identify potential mistakes involved in the reporting of the data. Based on the aforementioned data selection criteria, a total of 76 test data were collected from the literature and included in the burst test database of perfect pipes. The sources of the test data include [5-6], [9], [15-18] and [20-29]. Details of the burst test database are presented in [30]. The lower bound and upper bound of the key parameters for the burst test data are 179.4 and 1422.4 mm for the outside diameter (D); 1.448 and 22.5 mm for the pipe wall-thickness (t); 1.011 and 1.111 for the outside-to-inside diameter ratio (k); 451 and 2247 MPa for the ultimate tensile strength (u); 0.632 and 1.000 for the yield-to-tensile strength ratio (YT), and 0.0129 and 0.189 for the hardening exponent (n). Note that the ranges of YT and n are only for a subset of the burst test data because some of the tests did not report the yield strengths or n (or uniform elongation that can be used to calculate n). It follows that for certain burst capacity models only a subset of the test data included in the database is applicable. Most of the test data were collected from first-hand sources. For the data that were not obtained from first-hand sources, most were found in more than one source, and crosschecking between different sources were carried out to select the most reliable data (i.e. no conflicting information for the data from different sources). The predicted burst pressures using various burst models are reported in the references for all the test data. These predictions were repeated in this study, and test data for which the predictions obtained in this study did not agree with those reported in the references were excluded to avoid potential errors contained in the reporting of the data. MODEL ERROR ASSESSMENT The model error for a given burst capacity model is defined as the ratio of the actual burst capacity to the predicted burst capacity calculated using actual geometric and material properties (as oppose to nominal values). Although the burst capacity of a pipe specimen measured in a burst test is not exactly the actual burst capacity of the pipe due to the inevitable errors (e.g. measurement error) involved in the test, it provides a close estimate of the actual burst capacity. Furthermore, there is a lack of information to quantify the errors involved in the test. Therefore, the ratios between the test and predicted burst capacities were used to characterize the model error associated with a given burst capacity model. This

Copyright 2012 by ASME

results in a conservative characterization of the model error [31]. Given the burst test database, the test-to-predicted ratios for the burst capacity models summarized in Table 1 were calculated. The calculated means and COVs of the model errors as well as the number of applicable data points for each of the burst capacity models are given in Table 2. The results shown in the table indicate that the model errors for all the burst capacity models have mean values close to 1.0 and relatively small COV values (typically between 4% and 6% with the maximum value of 8.1% for the DNV model). This suggests that all the models provide reasonably accurate predictions of the burst capacities for thin-walled perfect pipes. Note that the mean values of the model errors for the models belonging to the Tresca solution are generally very close to or slightly greater than 1.0. On the other hand, the mean values of the model errors for the models belonging to the von Mises solution are all less than 1.0 and typically around 0.9. In other words, the models belonging to the Tresca solution tend to slightly underestimate the burst capacity whereas the models belonging to the von Mises solution generally overestimate the burst capacity. These results are consistent with the fact that the Tresca yield criterion represents the minimum yield surface that satisfies both the convexity requirement and the symmetry conditions associated with an isotropic material, i.e. a lower bound solution [32], whereas the von Mises yield surface is close to the theoretical upper bound [17]. The results summarized in Table 2 suggest that the Zhu and Leis (1) model is the best of all the models considered in this study in that the corresponding model error has a mean value (0.980) close to unity and a COV value (4.3%) that is the lowest of all the COV values. It is also noteworthy that the prediction accuracies of several other models are comparable to that of the Zhu and Leis (1) model, e.g. the Zhu and Leis (2) model, the Stewart et al. (Tresca/von Mises) models, the Bailey-Nadai model and the Svensson model. The probability distributions for the model errors were determined by plotting the test-to-predicted ratios on probability papers and using standard distribution fitting techniques to identify the best-fit distributions. A distribution fitting program, C-FIT, provided by C-FER Technologies was used in the analysis. The best-fit distributions of the test-topredicted ratios for all the burst capacity models are summarized in Table 2. We found that the normal or lognormal distribution fits the test-to-predicted ratios remarkably well for all the models except the DNV, Bailey-Nadai and Nadai-2 models. The Weibull distribution (with the lower bound equal to zero) was selected as the best-fit distribution for the BaileyNadai and Nadai-2 models, whereas the Gumbel distribution was selected as the best-fit distribution for the DNV model. Four representative probability distribution fitting plots are shown in Figs. 1 through 4, for the Zhou and Leis (1) model, Bailey-Nadai model, DNV model and Stewart et al. (Tresca) model respectively. The distribution fitting plots for the other models can be found in [30]

Table 2 Results of the Model Error Assessment

No. 1 2a 2b 2c 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Model ASME Barlow OD Barlow ID Barlow Flow Bailey-Nadai DNV Fletcher Max. Shear Stress Turner Stewart et al. (Tresca) Bohm Faupel Marin and Rimrott Marin and Sharma Nadai-1 Nadai-2 Soderberg Svensson Stewart et al. (von Mises) Zhu and Leis (1) Number of Data Points 76 76 76 55 49 55 49 76 76 49 49 55 49 49 76 49 76 49 49 49 Mean 1.023 1.039 0.999 1.048 0.998 1.046 0.989 1.019 1.019 1.064 0.935 0.886 0.950 0.888 0.883 0.865 0.883 0.934 0.909 0.980 COV 6.0% 5.8% 6.6% 5.3% 4.7% 8.1% 4.8% 6.1% 6.1% 4.3% 4.4% 4.8% 4.7% 4.3% 6.1% 4.7% 6.1% 4.4% 4.3% 4.3% Fitted Distribution Normal Normal Lognormal Lognormal Weibull Gumbel Normal Lognormal Normal Lognormal Normal Lognormal Lognormal Normal Normal Weibull Normal Normal Normal Normal Average Shear Stress Yield Criterion Group of Solution

Tresca

von Mises

19

Zhu and Leis (2)

55

0.970

4.5%

Normal

Figure 1 Fitted probability distribution of test-to-predicted ratios, Zhu and Leis (1) model

Copyright 2012 by ASME

Figure 2 Fitted probability distribution of test-to-predicted ratios, Bailey-Nadai model

We further evaluated the accuracy of the empirical equation developed by Zhu and Leis [12] (i.e. Eq. (2)) for predicting the strain hardening exponent by comparing the actual values of n with the predicted values for 42 data points included in the burst test database. The ratios between the actual and predicted values (denoted by ) are plotted versus YT in Fig. 5. The mean and COV of these ratios were calculated to be 1.364 and 36.6%, respectively. Figure 5 shows that tends to become larger as YT increases. This observation suggests that Eq. (2) will likely overestimate n for high grade materials, which are typically associated with high YT values. Despite the relatively large COV value for , it is noted that using Eq. (2) to estimate n only slightly reduces the accuracy of the Zhu and Leis (1) model, given that the COV of the model error for the Zhu and Leis (2) model is marginally higher than that of the model error for the Zhu and Leis (1) model. This suggests that the Zhu and Leis (1) model is relatively insensitive to n.

Figure 5 Ratios between the actual values of n and values of n predicted using Eq. (2) versus the corresponding yieldto-tensile strength ratios Figure 3 Fitted probability distribution of test-to-predicted ratios, DNV model

To further evaluate the applicability of Eq. (2) in the burst capacity prediction, the test-to-predicted ratios for the other models that employ n or u were also calculated based on the estimated n values obtained from Eq. (2), and the corresponding means and COVs are summarized in Table 3. The results in the table suggest that the predictions given by these models are all insensitive to the value of n and that using Eq. (2) to estimate n is acceptable for the burst capacity models considered in this study.
Table 3 Mean and COV of the model error based on the

actual and estimated n values

Based on Actual n Values Mean COV 0.998 4.7% 0.989 4.8% 1.064 4.3% 0.935 4.4% 0.950 4.7% 0.888 4.3% 0.865 4.7% 0.934 4.4% 0.909 4.3%

Model Bailey-Nadai 1 Fletcher Stewart et al. (Tresca) Bohm 1 Marin and Rimrott Marin and Sharma Nadai-2 1 Svensson 1 Stewart et al. (von Mises)

Based on Estimated n Values Mean COV

0.996 1.001 1.049 0.921 0.930 0.878 0.862 0.920 0.898 4.7% 4.5% 4.4% 4.4% 4.7% 4.6% 4.7% 4.4% 4.4%

Figure 4 Fitted probability distribution of test-to-predicted ratios, Steward et al. (Tresca) model

Copyright 2012 by ASME

CONCLUSIONS This paper describes a model error assessment carried out for 19 burst pressure prediction models for defect-free thinwalled pipes using a total of 76 full-scale burst test data of perfect pipes and pressure vessels collected from the literature. The considered models were categorized into three groups, namely the Tresca solution, the von Mises solution, and the average shear stress yield criterion. The probabilistic characteristics of the model error, i.e. the mean, coefficient of variation and probability distribution type, were obtained based on the test-to-predicted ratios. The model errors obtained in this study are critical to the structural reliability analysis of energy pipelines with respect to the limit state of burst of defect-free pipes and will facilitate the reliability-based design and assessment of pipelines. Specific observations obtained from the study are as follows: The model errors of all the burst capacity models considered have mean values close to 1.0 and COV values generally between 4 and 6%. This suggests that all the models provide reasonably accurate predictions of the burst capacities for thin-walled perfect pipes. The models belonging to the Tresca solution tend to slightly underestimate the burst capacity whereas the models belonging to the von Mises solution generally overestimate the burst capacity. The Zhu and Leis (1) model is considered the best of all the models considered in this study, as the corresponding model error has a mean value (0.980) close to unity and a COV value (4.3%) that is the lowest of the model errors for all the models considered. The normal or lognormal distribution fits the test-topredicted ratios with remarkable agreement for all the models except the DNV, Bailey-Nadai and Nadai-2 models. The Gumbel distribution best fits the test-topredicted ratios for the DNV model, whereas the Weibull distribution with the lower bound equal to zero best fits the test-to-predicted ratios for the Bailey-Nadai and Nadai-2 models. There is a relatively large uncertainty associated with the accuracy of the Zhu and Leiss empirical equation (i.e. Eq. (2)) for predicting the strain hardening exponent (n) from the yield-to-tensile strength ratio. However, it was found that the prediction accuracies of the burst capacity models considered in this study that employ n (or equivalently u) are all insensitive to the values of n. Therefore, the use of Eq. (2) to estimate n is considered acceptable in these models. ACKNOWLEDGMENTS Financial support provided by the Natural Sciences and Engineering Research Council and Canada (NSERC) through the Discovery Grant program and by the Faculty of Engineering at Western University is gratefully acknowledged.

REFERENCES [1] Nessim, M., Zhou, W., Zhou, J. and Rothwell, B. (2009). Reliability Based Design and Assessment for LocationSpecific Failure Threats with Application to Natural Gas Pipelines. Journal of Pressure Vessel Technology, ASME, 131(4), 0417011-0417015. Nessim, M., Zhou, W., Zhou, J. and Rothwell, B. (2009). Target Reliability Levels for Design and Assessment of Onshore Natural Gas Pipelines. Journal of Pressure Vessel Technology, ASME, 131(6), 0617011-06170112.

[2]

[3] CSA (2007). CSA-Z662 Oil and Gas Pipeline Systems. Canadian Standards Association, Rexdale, Ontario. [4] Christopher, T., Rama Sarma, B. S. V., Govindan Potti, P. K., Nageswara Rao, B. and Sankarnarayanasamy, K. (2002). A Comparative Study on Failure Pressure Estimations of Unflawed Cylindrical Vessels. International Journal of Pressure Vessels and Piping, 79(1), 53-66. [5] Law, M., Bowie, G., Fletcher, L. and Piper, L. (2004). Burst Pressure and Failure Strain in Linepipe: Part 1-3. Journal of Pipeline Integrity, 3(2), 95-113. [6] Law, M. and Bowie, G. (2006). Failure Strain in High Yield-to-Tensile Ratio Linepipes. Journal of Pipeline Integrity, 5(1), 25-36. [7] Zimmermann, S., Hohler, S., and Marewski, U. (2007). Modeling Ultimate Limit States Burst Pressure and Yielding of Flawless Pipes. Proceedings of the 16th Biennial Pipeline Research Joint Technical Meeting, Canberra, Australia, Paper 13. [8] Zhu, X. K., and Leis, B. N. (2010). Prediction and Comparison of Burst Pressure for Linepipes. Proceedings of the 8th International Pipeline Conference, ASME, Calgary, Canada, Paper No. IPC2010-31581.

[9] Gaessler, H. and Vogt, G. (1988). Influence of Yield to Tensile Ratio on the Safety Of Pipelines. Proceedings of the 7th Biennial Joint Technical Meeting on Line Pipe Research, NG-18/EPRC, Calgary, Canada, Paper 8. [10] Manucci and Harris (2000). Fracture Properties of API X100 Gas Pipeline Steels. Technical Steel Research, Contract #7210-PR/058. [11] Law, M. and Bowie, G. (2007). Prediction of Failure Strain and Burst Pressure in High Yield-to-Tensile Strength Ratio Linepipe. International Journal of Pressure Vessels and Piping, 84(8), 487-492. [12] Zhu, X. K. and Leis, B. N. (2005). Influence of Yield-toTensile Strength Ratio on Failure Assessment of Corroded Pipelines. Journal of Pressure Vessel Technology, ASME, 127(4), 436-442.

Copyright 2012 by ASME

[13] Zhu, X. K. and Leis, B. N. (2006). Average Shear Stress Yield Criterion and Its Application to Plastic Collapse Analysis of Pipelines. International Journal of Pressure Vessels and Piping, 83(9), 663-671. [14] Kiefner, J. F., Maxey, W. A. and Duffy, A. R. (1971). The Significance of the Yield-to-Ultimate Strength Ratio of Summary Report to Pipeline Line Pipe Materials. Research Committee, American Gas Association. [15] Maxey, W. A. (1986). Y/T Significance in Line Pipe. Proceedings of the 7th Symposium on Line Pipe Research, Houston, U.S.A., Paper 9, 1-18. [16] Amano, K., Matsuko, M., Ishihara, T., Tanaka, K., Inoue, T., Kawaguchi, Y., and Tsukamoto, M. (1986). Significance of Yield Ratio Limitation to Plastic Deformation of Pipeline in High Pressure Proof Test. Proceedings of the 7th Symposium on Line Pipe Research, Houston, U.S.A., Paper 8, 1-21. [17] Stewart, G., Klever, F. J. and Ritchie, D. (1994). An Analytical Model to Predict the Burst Capacity of Pipelines. Proceedings of International Conference of Offshore Mechanics and Arctic Engineering, ASME, V, 177-188. [18] Paslay, P. R., Cernocky, E. P., and Wink, R. (1998). Burst Pressure Prediction on Thin-Walled, Ductile Tubulars Subjected to Axial Load. Proceedings of Applied Technology Workshop on Risk Based Design of Well Cashing and Tubing, Woodlands, U.S.A. [19] Rajan, K. M., Deshpande, P. U. and Narasimhan, K. (2002). Experimental Studies on Bursting Pressure of Thin-Walled Flow Formed Pressure Vessels. Journal of Materials Processing Technology, 125-126, 228-234. [20] Hillenbrand, H. G., Liessem, A., Knauf, G., Niederhoff, K. and Bauer, J. (2000). Development of Large-Diameter Pipe in Grade X100 State-of-the-Art Report from the Manufacturers Point of View. Proceedings of the 3rd International Conference on Pipeline Technology, Brugge, Belgium, 1, 469-482. [21] Papka, S. D., Stevens, J. H., Macia, D. P., Fairchild, D. P. and Petersen, C. W. (2003). Full-Size Testing and Analysis of X120 Linepipe. Proceedings of the 13th International Offshore and Polar Engineering Conference, Honolulu, Hawaii, U.S.A., 50-59. [22] Okaguchi, S., Makino, H., Hamada, M., et al. (2004). Development and Mechanical Properties of X120 Linepipe. International Journal of Offshore and Polar Engineering, ISOPE, 14(1), 28-35.

[23] Royer, C. P. and Rolfe, S. T. (1974). Effect of Strain Hardening Exponent and Strain Concentrations on the Journal of Bursting Behavior of Pressure Vessels. Engineering Materials and Technology, Transactions of the ASME, 96(4), 292-298. [24] Margetson, J. (1978). Burst Pressure Predictions of Rocket Motors. AIAA/SAE 14th Joint Propulsion Conference, Las Vegas, U.S.A., AIAA paper no. 78-1569. [25] Mok, D. H. B., Pick, R. J., Glover, A.G. and Hoff, R. (1990). Behaviour of Line Pipe with Long External Corrosion. Material Performance, 29(5), 75-79. [26] Liessem, A., Grdef, M. K., Knauf, G. and Marewski, U. (2004). Influence of Thermal Treatment on Mechanical Properties on UOE Linepipe. Proceedings of the 4th International Conference on Pipeline Technology, Ostend, Belgium, 3, 1262-1281. [27] Cronin, D. S. (2000). Assessment of Corrosion Defects in Pipelines. Ph.D Thesis, Department of Mechanical Engineering, University of Waterloo, Canada. [28] Benjamin, A. C., Vieira, R. D., Diniz, J. L. C., Freire, J. L. F., and De Andrade, E. Q. (2005). Burst Tests on Pipeline Containing Interacting Corrosion Defects. Proceedings of the 24th International Conference on Offshore Mechanics and Arctic Engineering, OMAE, 3, 403-417. [29] Chouchaoui, B. A. (1993). Evaluating the Remaining Strength of Corroded Pipelines. Ph.D Thesis, Department of Mechanical Engineering, University of Waterloo, Canada. [30] Huang, G. (2011). Model Error Assessment of Burst Capacity Models for Energy Pipelines. M.E.Sc. Thesis, The University of Western Ontario, London, Ontario, Canada. [31] Mendelson, A. (1986). Plasticity: Theory and Application. MacMillan, New York. [32] Mirza, S. A. and MacGregor, J. G. (1989). Slenderness and Strength Reliability of Reinforced Concrete Columns. ACI Structural Journal, 86(4), 428-438.

Copyright 2012 by ASME