Experimental Investigations of the Rapid C rack Propagation Performance of Polyethylene Pipes

Transcription

1 Experimental Investigations of the Rapid C rack Propagation Performance of Polyethylene Pipes Mark J. Lamborn and Ashish M. Sukhadia Chevron Phillips Chemical Company LP, USA Bartlesville Research & Technology Center Bartlesville, O K A BST R A C T Good resistance to Rapid Crack Propagation (RCP) is considered to be important to the field performance of polyethylene (PE) pipe. Even though RCP occurrences in PE pipe are very rare, it remains important to understand the RCP phenomenon to ensure that pipes are designed to avoid RCP events. The characterization of RCP performance depends on a variety of factors, such as the test conditions (temperature and pressure), pipe geometry (diameter and wall thickness), and choice of RCP test. Presented here are an accumulation of results from various experimental programs undertaken by us to develop a basic understanding of RCP in PE pipes and the test methods used to characterize RCP performance. Data are presented showing critical pressure critical temperature failure surfaces for MDPE, HDPE, and bimodal HDPE pipe, and their use in the selection of RCP resistant pipe based on specific service conditions is discussed. Data are presented to demonstrate the phenomenon of false arrest and to show how this can lead to misleading and incorrect interpretations of RCP performance from single-point S4 tests at excessively high pressures. The results of experimental work to assess the influence of pipe size on S4 test results are also presented. Analysis of these data, combined with consideration of the effects of shear lips and residual stress effects, indicates that an Irwin-Corten relationship can be employed to provide conservative estimates RCP performance for different size pipe, subject to specific restrictions on the relative geometries between pipes. IN T R O DU C T I O N The long term safe performance of polyethylene (PE) pipe in service has been well established, making it the material of choice for lower pressure gas distribution pipelines. More recently, the development of next-generation PE resins has created interest in using PE pipe in ever more demanding applications, requiring higher pressures and larger pipe sizes. When considering PE pipe for use in these applications, it remains important to consider all design properties, so that confidence in the use of PE pipe is maintained, and its record of safe performance is continued. The of several performance criterions that have generated considerable interest. Although RCP failures are rare, knowledge of the factors contributing to the RCP phenomenon is necessary to continue the safe operation of PE pipe systems and avoid RCP events. Two standard test methods are available to measure the RCP performance of PE pipe systems; specifically, the Full Scale (FS) [1] and Small Scale Steady State (S4) [2] tests. The FS test acts 1

2 as the referee method for characterizing the RCP performance of pipe systems. However, the FS test suffers from the disadvantages of being difficult to perform, costly, and available from few laboratories worldwide. By comparison, the S4 has the advantages of lesser complexity, lower cost, greater availability of laboratories, and commercial availability of S4 test equipment. Because of these advantages, the S4 test finds frequent use in characterizing the RCP performance of PE pipe systems. Chevron Phillips Chemical acquired the ability to measure RCP performance of PE pipe via the purchase, installation and commissioning of a S4 test unit in The result of our ongoing research efforts and the fact that Performance Pipe, the largest producer of PE pipes in North America, is a division of Chevron Phillips Chemical has provided an unique opportunity to study the RCP performance of a large number of PE pipes. Since 2004, CPChem has performed in excess of 150 S4 critical pressure and critical temperature tests in efforts to probe, understand, and document the RCP performance of PE pipe systems. It is perhaps noteworthy that the large body of PE pipes tested represents a mix of resin classes (PE80 vs. PE100), structure (unimodal vs. bimodal) and a wide range of PE architectures (experimental vs. commercial resins). Furthermore, this database also represents a mix of pipes fabricated under both laboratory and commercial extrusion conditions. The intent of this paper is to share accumulated knowledge and expertise in the measurement of RCP performance of PE resins via the S4 test. R CP F A I L UR E SUR F A C ES The ISO standard [2] for determination of resistance to RCP via the S4 test provides methods for measurement of the S4 critical pressure (P C ) and the S4 critical temperature (T C ) of PE pipe systems. The S4 P C is determined from series of tests typically performed at 0 ºC and variable pressure. It is defined as the largest pressure producing an arrest event, below the lowest pressure producing a propagation event. The S4 T C is determined from a series of tests typically performed at a pressure of 5 bars and variable temperature. It is defined as the lowest temperature producing an arrest event, above the highest temperature producing a propagation event. While the results of the standard S4 P C and S4 T C are typically employed to characterize a combined results represent only surface, defined by the curve of S4 P C as a function of temperature. (Note that an exception to this statement is the case where the S4 P C = 5 bars and the S4 T C = 0 ºC, whereby the S4 P C and S4 T C results represent only a single point on the RCP failure surface.) The RCP failure surface for a pipe system is established by performing S4 P C tests at several different test temperatures. The resulting curve of S4 P C versus temperature defines the conditions of pressure and temperature that the pipe system is susceptible to RCP failures, as well as those at which the pipe system is not. Towards characterizing the RCP failure surfaces for common PE resins using in pipe systems, an experimental program was conducted on 200 mm SDR 11 pipe of MDPE, HDPE, and bimodal HDPE resins. S4 measurements obtained at several different combinations of pressure and temperature were used to map the RCP failure surfaces. The test pressures employed for this study were 0.8, 2.0, and 4.0 bars for the MDPE pipe; 1.0, 2.5, and 5.0 bars for the HDPE pipe; and 1.0, 2.0, 3.0, 5.0, and 10.0 bars for the bimodal HDPE pipe. Test temperatures were varied from 20 ºC to -50 ºC, in increments of 10 ºC. In cases where a specific pressure temperature 2

3 combination resulted in a propagation event, subsequent tests at lower temperatures were not performed at this pressure because it would surely result in a propagation event. The results of this experimental program appear in Fig. (1), where the grid of data points indicates the pressure temperature combinations at which S4 tests were performed and points marked by an indicate pressure-temperature combinations at which actual S4 test were performed. Data points are shaded either green or red data points to indicate arrest and propagation events, respectively. The dashed lines indicate the approximate locations of the RCP failure surfaces for the different PE pipe systems. Note that the locations of these surfaces are approximate, as controlled by the pressure differences employed for the tests. The FS test pressures were determined using the published P C correlation between the S4 and FS tests [2]; note, this correlation provides conservative estimates of FS pressures from S4 test results. The failure surfaces in terms of FS pressures are considered to be more representative of actual service conditions. For a pipe system of a specific size and resin, the RCP failure surface defines the conditions of pressure and temperature at which susceptibility to RCP is expected, or not. Service conditions falling on, or below, the RCP failure surface not expected to produce RCP failures, while service conditions above this surface are susceptible to RCP failures. Comparison of the failure surfaces between resins shows that the bimodal HDPE exhibits the best performance, followed by the HDPE and MDPE resins. Inspection of all failure surfaces shows that the susceptibility to RCP failures increases with decreasing temperature for all resins, as expected. These surfaces eventually level off to an inverted plateau at low temperatures, with the differences in failure surfaces between resins lessening; thus, resin differences play a much lesser role in RCP performance at low temperatures. The plateau pressures may be of particular importance for low temperature applications because their flatness over a broad range of low temperatures suggests that they define the pressure below which RCP events are not expected at any temperature. It is worthwhile to observe that all resins behave essentially the same at temperatures below the standard S4 T C (i.e., 5 bar pressure). The temperature difference between the standard S4 T C and the onset of the plateau region is approximately 20 º C for all resins; thus, below T C the rate of decline of RCP performance is essentially independent of resin architecture, at least for the resins employed in this study. Additional resins need to be tested to determine if these trends are representative in general. E F F E C TS O F PIPE SI Z E O N S4 C RI T I C A L PR ESSUR E One approach to estimating the influence of pipe size on S4 critical pressure has been to employ the Irwin Corten relationship [3], which when applied to S4 RCP events, provides a relationship G D pressure P C, and pipe geometry (outside diameter D o and standard dimension ratio SDR = D o /t, where t = the minimum pipe wall thickness). Lamborn and Sukhadia [4] employed a form of this relationship to estimate the effects of pipe size differences on the S4 P C of thick-wall pipe; this relationship is PC DoR g SDRR CPR C m (1) P D g SDR CR o 3

4 where the C /P CR is the ratio of critical pressures between the pipe of interest and reference pipe, D o is the outside diameter, g(sdr) is a function of SDR given by g SDR 2 2 SDR 2 SDR 1.2SDR 1 8SDRSDR 1 and C m is termed the materials property ratio as defined by 4 G E (2) D C m (3) G DR E R In which In general C m between pipes of different sizes; however, the exact value of C m is usually not known. In a typical application of Eqs. (1) to (3), the P CR is known and it is desired to estimate the P C for the pipe of interest. The work to be presented here uses Eqs. (1) to (3) to estimate CPR for pipes of different sizes assuming that C m = 1. Subsequent comparisons with experimental S4 test results permits determination of the circumstances under which these predictions are expected to be conservative. One limitation of using Eqs. (1) to (3) to estimate the pipe size effects on S4 performance is that their derivation is based on pipe that is free of residual stresses. In practice, PE pipe are subject to residual stresses as the result of differential cooling during the extrusion process [5, 6]. For different pipe sizes extruded under similar cooling profiles, the magnitude of residual stresses increases with increasing pipe wall thickness because of the greater through-thickness temperature gradients and increased differential cooling. Residual stresses are inherent to all extruded PE pipe, and are expected to adversely affect RCP performance because residual energy stored in the pipe is available to contribute to the crack driving force during a RCP event. In order to examine residual stress effects, a simple experiment was designed to estimate contributions to the energy release rate, G D. Pipe ring specimens were cut from 200 mm SDR 9, 11, 13.5, and 17 pipes of a single HDPE resin. The pipe ring specimens for 10 hours, to simulate the temperature conditioning for the standard S4 P C test. At the end of conditioning, a longitudinal crack was immediately cut into the specimen using a band saw, and the opening of the crack along the inside pipe wall was measured. The opening measurement used in conjunction with curved beam formulae available in [7] permits the amount of residual energy released by the longitudinal crack and the residual stress contribution to G D to be estimated. Note that the exact asymmetric distribution of residual stresses is likely different than that predicted by curved beam theory; however, as noted by Choi and Broutman [6], this analysis provides the same relative ranking of residual stress distributions as more exact measurement methods. The intent here is simply to estimate the relative residual stress contributions between different size pipe, as a means of estimating the relative effects on predictions provided by the Irwin-Corten analysis. These estimates appear in Fig. (2), where the G D contributions are normalized by the maximum contribution measured for the SDR 9 pipe. Although these data are estimates, they are expected to correctly reflect the relative trends between pipes of different sizes. These estimates clearly show that the residual stress contributions to G D increase with

5 decreasing SDR (increasing thickness). These contributions are expected to lower the internal pressure required to sustain RCP because a lesser energy contribution is required from the internal pressure loading. Another consideration neglected in the derivation of Eqs. (1) to (3) is the effect of shear lips on the fracture toughness G D. For pipe fabricated under similar extrusion profiles, G D is expected to increase with decreasing wall thickness due to the increased influence of shear lips at the inner pipe wall during RCP. During an S4 RCP event, the inside pipe wall is subject to plane stresslike deformations, leading to the development of so-called shear lips along the intersection of crack path and inside pipe wall. These shear lips are associated with relatively high values of G D. Inspection of the thickness of shear lip zones for PE pipes of the same resin and different SDR shows that they are essentially constant in size. The expected result is an increase in G D as pipe SDR is increased (thickness decreased), due to the greater volume fraction of pipe consumed in the shear lip zone. This argument is supported by the RCP failure surfaces shown in Fig. (3) for 200 mm SDR 9 and SDR 17 HDPE pipe. The shear lip deformation zones are located at the inside pipe walls and appear as the regions exhibiting greater material drawing, or ductility. These zones are essentially of equal thickness for both the SDR 9 and SDR 17 pipe. As discussed, both residual stresses and shear lips are expected to improve G D as the pipe wall thickness deceases, or equivalently SDR increases. In the work to follow, the combined effects of residual stress and shear lips will be referred to as the thickness effect, and understood to reference an increase in critical pressure (improvement in RCP performance) as pipe wall thickness is decreased. Note that variations in are also expected to affect RCP performance through the material property ratio C m ; however, these effects are not considered in the current analysis. Data to be presented here indicates that deviations from the predictions of the Irwin Corten analysis are predominately due to the thickness effect. To evaluate the effectiveness of the Irwin Corten analysis to estimate the effects of pipe size on the measured S4 P C, S4 tests were performed on pipes of different geometries, but the same PE resins. Appearing in Fig (4) are the CPRs estimated by Eq. (1), compared to the actual S4 P C measurements for SDR 11 pipe of 50 mm and 200 mm diameters. CPR predictions appear as the solid line in Fig. (4). The arrow extending upward illustrates the direction of change in CPR estimates due to the thickness effect; specifically, the arrow shows the direction the CPR estimates would translate if the magnitude of the thickness effect were known and accounted for. The data points in Fig. (4) are the CPRs determined from S4 P C measurements on MDPE, HDPE, and bimodal HDPE, 50 mm and 200 mm SDR 11 pipe. The 200 mm pipes serve as the reference cases for the analysis, as indicated by the collective reference point located at D o /D or = P C /P CR = 1. CPRs for the 50 mm pipes correspond to points with a diameter ratio, D o /D or, of approximately The arrow extending upwards from the data point for the bimodal HDPE resin indicates that the actual critical pressure is greater than that indicated (i.e., P C is greater than that used for the S4 tests). Inspection of these data demonstrates that Eq. (1) provides conservative estimates of the measured S4 P C. Appearing in Fig (5) are similar data and analysis for MDPE and HDPE pipe of the same 200 mm diameter, but different SDR. The SDR 9 pipes serve as the reference for the analysis. The S4 P C measurements for the MDPE pipe show a decrease in P C as the SDR is increased from 9 to 13.5, followed by an increase for the SDR 17 pipe. For the HDPE pipe, S4 P C decreases as the SDR is increased from 9 to 11, followed by an 5

6 increase for SDR 13.5 and 17. The reductions in CPR with increasing SDR are consistent with the predictions of Eq. (1). The increases in CPR with increasing SDR are consistent with the expected results of the thickness effect. In this case the direction of P C changes via Eq. (1) and due to the thickness effect are opposite in direction; thus, the exact direction of P C change cannot be estimated. Regardless, inspection of the data shows that Eq. (1) provides conservative estimates of the measured S4 P C. Although not shown, reanalysis of the data in Figs. (4) and (5) using the pipe of the least thickness as the reference pipe provides overestimates of the CPR, indicating that conservative P C measurements are only obtained when the pipe of greatest thickness is used as the reference pipe. This is consistent with the expected consequences of the thickness effect. The good agreement between the experimental data, predictions of Eq. (1), and the mechanism presented for the thickness effect, suggests that Eq.(1) may be employed to provide conservative estimates of S4 P C, subject to limitation that the thickness of the reference pipe must be greater than that of the pipe of interest. An example of such estimates appears in Fig. (6), where 200 mm SDR 11 reference pipe is used to estimate the critical pressure ratio, CPR, for other pipe sizes. As indicated, Eq. (1) is expected to provide conservative estimates of CPR for pipes of lesser thickness relative to the reference pipe. For pipes of greater thickness, the CPR estimates are non-conservative and should be avoided. It should be noted that the experimental data supporting these predictions is limited to either pipe of constant SDR with different diameters, or pipe of constant diameter, but different SDR. The apparent agreement with the suggested consequences of the thickness effect suggests that these predictions are valid for more general pipe geometry differences. Further experimental work is needed to verify, or disprove, that the analysis conservatively predicts P C between pipes of both different diameter and SDR. F A LSE A RR EST The terminology false arrest refers to a specific outcome from the S4 test, whereby the occurrences of arrest events occur at pressures above the S4 critical pressure. This has also been termed the cloche effect, because the resulting curve of crack length versus pressure is bell shaped. Leevers [8] reports that false arrests are the result of pipe expansion and pipe wall flaring along the crack opening during RCP, both serving to aid in decompression ahead of the crack. For S4 tests performed on thin wall pipe, the extent of flaring may be such that the pipe makes contact with the specimen cage, thus limiting deformation and promoting arrest. Kosari et. al. [9] reported false arrests for S4 critical temperature tests performed on a PE pipe at internal pressures of 5 bars and 10 bars. In this study, the S4 critical temperature test at the 10 bar pressure was lower than that for the 5 bar pressure, indicating the occurrence of false arrest. The occurrence of false arrest in the S4 test is promoted by relatively high pressures and thin pipe walls, and in the case of thin pipe walls, false arrest may occur at pressures below 5 bars. To address the occurrence of false arrest, the ISO S4 standard limits the internal pressure to 10 bars and requires that test be performed in 2 bar increments below the candidate P C. In this manner, false arrests can be detected, but not prevented. Regardless of the guidelines of the S4 standard, it has been common practice to report S4 critical pressures of 12 bars, or greater, on resin datasheets. This practice can be problematic, particularly if the S4 test is performed on a single test condition, pass/fail basis. 6

7 The pass/fail method of performing the S4 test appears to be a quick cost-effective manner of determining if a pipe/resin is susceptible to RCP at a specified pressure. In this case, only a single S4 test is performed at the pressure of interest. If the test results in an arrest event, the S4 P C is reported as being greater than the pressure employed for the test. Although the S4 standard and literature warns against this practice, it remains attractive primarily due to cost and time considerations. As an example of this problem, in 2007 Chevron Phillips Chemical performed both S4 P C and S4 T C tests on 200 mm SDR 11 pipe of a HDPE resin; S4 P C results for this pipe appear in Fig. (7). The initial S4 test s resulted in an arrest event, providing the initial result that the S4 P C was greater than 12 bars. Subsequent S4 T C tests performed at a 5 bars determined a S4 critical temperature greater than, thus identifying the initial 12 bar P C result as a false arrest. Additional S4 tests at 0 ºC provided the results appearing in Fig. (7), where S4 tests performed at 0 ºC resulted in propagation events at pressures well below 12 bars. These data serve as an example of an incorrect S4 P C result being obtained as the result of false arrest during a single S4 test performed on a pass/fail basis. Another factor that can conceivably contribute to the occurrence of false arrest is inadequate stiffness of the S4 internal baffle system. Baffle stiffness could conceivably affect S4 test results because it controls deflection of the baffle during the S4 test. Crack propagation during the S4 test results in sudden depressurization on one side of the baffle, producing a dynamic event that results in deflection of the internal baffles. This in turn increases the gap between the baffle and pipe wall, providing the opportunity for increased outflow and possibly false arrest. An example of extreme baffle deflection appears in Fig. (8), where the baffle system was permanently deformed during a S4 critical pressure test performed at a 20 bar pressure. It is noted that this test resulted in an arrest event, possibly due to false arrest. C O N C L USI O NS The intent of this paper is to share experimental observations of factors affecting the measurement of RCP performance via the S4 test. Specific conclusions from these observations are as follow: 1. RCP failure surfaces measured by the S4 test are a useful tool for defining the pressure and temperature service conditions at which PE pipe systems are susceptible to RCP failures. While they are time consuming to generate, they provide the most comprehensive picture of 2. The RCP failure surfaces for the PE resins employed in this study displayed essentially the same behavior at temperatures much below the S4 T C. All exhibited a temperature difference of approximately 20 ºC between the S4 T C and onset of the low temperature plateau where no further rapid crack propagation was observed at sufficiently low pressures. This pressure was observed to be between 1-2 bars for all three resin types tested here and further work is required to determine if this behavior and pressure threshold is representative of PE resins in general. 3. An Irwin Corten relationship appears to be capable of providing conservative estimates of S4 critical pressures for pipe of identical resins, but different sizes, provided the S4 P C for a 7

8 specific size pipe is known, and employed to predict the S4 P C for pipes of lesser wall thickness. Experimental confirmation of these predictions for pipes of different diameter and SDR is needed. 4. False arrest is problematic for single S4 tests performed on a pass/fail basis. This method of estimating RCP performance should be avoided by adhering to the requirements of the ISO standard. 5. Insufficient baffle stiffness is expected to contribute to the occurrence of false arrests. R E F E R E N C ES 1. Determination of Resistance to Rapid Crack Propagation (RCP) national Organization for Standardization (ISO). 2. Determination of Resistance to Rapid Crack Propagation (RCP) Small-Scale Steady- International Organization for Standardization (ISO). 3. Irwin, G. P. and Corten, H. T., Report to Northern Natural Gas Co. and El Paso Gas Co., (1968). 4. Factors Affecting the Rapid Crack Propagation Performance of Polyethylene Pipesdings, (2010). 5. pages , York (1982). 6. lymer (Korea), Vol. 21(1), pages (1997). 7. McGraw Hill Book Co., 6 th ed. (1989). 8. Proc. Plastics Pipes XII, April 2004, Milan, Italy. IOM Communications Ltd, London, UK 9. October2006, Washington, DC 8

9 S4 test Pressure, bars Green indicates arrest & Red indicates propagation indicates measured point Temperature, ºC MDPE HDPE Bimodal HDPE Figure 1 - S4 RCP failure surfaces measured for MDPE, HDPE and bimodal HDPE 200 mm SDR 11 pipe FS Test pressure, bars Residual Stress Contribution to G D, G D /G D,SDR= SDR Figure 2 - Residual stress energy contributions to the crack driving force for 200 mm pipe of different SDR and the same HDPE resin.. Figure 3 - Plane stress-like fracture zones located along the longitudinal crack, adjacent to the (wall thickness C ritical Pressure Ratio = P C /P C R Thickness Effect Critical Pressure Data for SDR 11 Pipe Reference Nominal Dia. = D or = 8 inch Diameter Ratio = D o /D or Predicted Critical Pressure Ratio 2 inch M DPE Pipe 2 inch H DPE Pipe 2 inch Bimodal H DPE Pipe 8 inch Reference Pipe Specimens Figure 4 Estimated and measured critical pressure ratios (CPR) for SDR 11 pipes of different diameters in the range of Do/Do R C ritical Pressure Ratio, P C /P C R P C data for 8 inch IPS Pipe Reference SDR = SDR R = 9 Thickness Decreasing Predicted Critical Pressure Ratio MDPE Pipe H DPE Pipe Thickness Effect Standard Dimension Ratio for the Pipe of Interest, SDR Figure 5 Estimated and measured critical pressure ratios for 200 mm pipes of different SDR in the range of SDR/SDR R 9

10 Critical Pressure Ratio Conservative Non-Conservative Reference Pipe SDR IPS Pipe Size Figure 6 Predicted critical pressure ratios (CPRs) for different size pipe, made using 200 mm SDR 11 pipe as the reference Propagation Event Crack Length/Diameter Critical Pressure = 2.2 Bars First test performed as pass/fail determination Arrest Event Pressure, Bar 8 inch SDR 11 Bimodal HDPE Resin Requirement for Propagation Figure 7 - S4 critical pressure test results for Results indicate the occurrence of false arrest. Figure 8 - Deformed internal baffle system resulting from a S4 test performed at a 20 bar pressure 10