4
Failure Criteria

OBJECTIVE AND APPROACH

One objective of this committee is to determine whether ductile iron pipe (DIP) with polyethylene encasement (PE) and cathodic protection (CP) can reliably provide a 50-year service life when used in underground buried water projects. As already noted, the Bureau of Reclamation initially requested a failure rate of zero in this regard. Clearly this desire was an indication that Reclamation is very concerned with pipeline failures and with the potential costs associated with pipeline failure and having pipelines out of service.

This committee asked Reclamation to provide a quantitative benchmark against which it could measure the performance of DIP with PE and CP installed in severely corrosive soils. The committee also asked whether failure rates of other systems would be acceptable benchmarks. The objective of these questions was simply to obtain from Reclamation a threshold value other than zero so that a comparative analysis could be performed by the committee.

In responding to the committee, Reclamation noted that an average annual failure rate of 0.000044 failures per mile per year was computed from gas pipeline performance data from the Department of Transportation’s (DOT’s) Office of Pipeline Safety (OPS). Reclamation also noted that this database did not provide information on soil conditions and that “some adjustment to the computed failure



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 99
4 Failure Criteria OBJECTIVE AND APPROACH One objective of this committee is to determine whether ductile iron pipe (DIP) with polyethylene encasement (PE) and cathodic protection (CP) can reli- ably provide a 50-year service life when used in underground buried water projects. As already noted, the Bureau of Reclamation initially requested a failure rate of zero in this regard. Clearly this desire was an indication that Reclamation is very concerned with pipeline failures and with the potential costs associated with pipe- line failure and having pipelines out of service. This committee asked Reclamation to provide a quantitative benchmark against which it could measure the performance of DIP with PE and CP installed in severely corrosive soils. The committee also asked whether failure rates of other systems would be acceptable benchmarks. The objective of these questions was simply to obtain from Reclamation a threshold value other than zero so that a comparative analysis could be performed by the committee. In responding to the committee, Reclamation noted that an average annual failure rate of 0.000044 failures per mile per year was computed from gas pipeline performance data from the Department of Transportation’s (DOT’s) Office of Pipeline Safety (OPS). Reclamation also noted that this database did not provide information on soil conditions and that “some adjustment to the computed failure 

OCR for page 99
corrosion Prevention standards ductile iron PiPe 00 for rate may be warranted to compensate for this uncertainty in soil conditions.”1 The bureau provided no specific guidance on how to adjust the data, but it did state that, regarding an average annual failure rate of 0.000044 failures per mile per year, “we believe this analysis supports Reclamation’s original response to the Committee’s question of what we define as reliably providing a minimum service life of 50 years for our pipelines.”2 This statement indicates that the threshold value that Reclama- tion provided (0.000044) is within its risk-tolerance range. The committee believes that a threshold value must be defined by Reclamation, and whether this information is based on pipe failures of other pipe systems or on an economic risk-based procedure, some value is necessary to allow a comparative analysis to be performed. Without discussing whether the analysis for determining the threshold value used by Reclamation is appropriate, but realizing that Recla- mation accepts the 0.000044 value as within its risk tolerance, the committee first uses the threshold value of 0.000044 failures per mile per year as suggested by Rec- lamation. This value is identified as Threshold 1. This threshold value can change based on other assumptions, so the committee examined two additional possible threshold values derived from the gas pipeline data set for comparison with the performance of DIP with PE and CP in soils with soil resistivity values less than 2,000 ohm-cm. The two additional gas pipeline-based thresholds are listed below with Threshold 1: • Threshold : Gas pipeline failure data from 2002 to 2008. • Threshold : Gas pipeline failure data from 2002 to 2008 (data for Threshold 1) that include four additional leaks associated with external corrosion of welds, elimination of leaks for pipeline in service for more than 50 years, and the use of all mileage for bonded dielectric steel gas pipeline. • Threshold : Gas pipeline failure data from 1974 to 1983, a time period during which regular inspections were not required by law. The threshold provided by Reclamation of 0.000044 (Threshold 1) articulates the bureau’s risk tolerance. The two additional threshold values derived from gas pipeline data are necessarily estimates, owing to the lack of information on the corrosivity of the soils where the failures occurred; they may not be representa- tive of Reclamation’s risk tolerance. However, the committee believed that it was necessary to develop additional threshold values to identify the potential variation 1 Michael Gabaldon, Bureau of Reclamation, letter to Emily Ann Meyer, National Research Coun- cil, re: National Academies Review of the Bureau of Reclamation’s Corrosion Prevention Standards for Ductile Iron Pipe—A Response to the Committee’s Request for Clarification on Project Scope, August 21, 2008. 2 Gabaldon, letter to Meyer, August 21, 2008.

OCR for page 99
failure criteria 0 TABLE 4-1 Gas Pipeline Failure Data from the U.S. Department of Transportation, Office of Pipeline Safety Event No. Year Installed Nominal Pipe Size (in.) Pipe Wall Thickness (in.) Years to Incident 1 1990 48 0.42 17 2 1978 20 0.25 29 3 1975 6.62 0.19 30 4 1975 30 0.31 30 5 1971 20 0.26 31 6 1967 4 0.19 35 7 1968 36 0.38 37 8 1967 36 0.33 39 9 1967 12.75 0.25 40 10 1966 36 0.33 41 11 1965 36 0.34 42 12 1965 30 0.31 42 13 1964 16 0.22 42 14 1964 24 0.25 44 15 1960 6.63 0.19 44 16 1960 20 0.28 44 17 1960 24 0.28 44 18 1959 20 0.25 47 19 1956 26 0.28 47 20 1953 24 0.25 50 21 1957 26 0.25 50 22 1947 26 0.3 58 23 1947 26 0.29 58 24 1947 20 0.31 59 25 1944 8 0.19 60 26 1943 24 0.38 60 27 1928 18 0.28 74 28 1964 36 0.42 38 29 1961 16 0.22 42 30 1961 12 0.19 42 31 1962 30 0.31 45 SOURCE: U.S. Department of Transportation, Office of Pipeline Safety. that might result from using different methods. Table 4-1 shows the failure data for the gas pipeline system. Another approach to determining a rational threshold for comparison with DIP with PE and CP is to use the failure rates of pipelines for conditions as reported in Table 2 entitled, “Corrosion Prevention Criteria and Minimum Requirements,” in the Bureau of Reclamation’s Technical Memorandum 8140-CC-2004-1, “Cor- rosion Considerations for Buried Metallic Water Pipe.”3 If the failure rates for 3 Bureau of Reclamation, U.S. Department of the Interior, Technical Memorandum 8140-CC-2004-1, “Corrosion Considerations for Buried Metallic Water Pipes,” Washington, D.C., July 2004.

OCR for page 99
corrosion Prevention standards ductile iron PiPe 0 for pipelines in soils with resistivity values between 2,000 and 3,000 ohm-cm and for pipelines in soils with resistivity values greater than 3,000 ohm-cm could be determined, this same threshold value could be used for pipelines embedded in soils with resistivity values less than 2,000 ohm-cm. The use of the specific pipeline system in these specific soil conditions is an implicit statement of a threshold for pipeline system performance for such soils. It should be noted that the corrosion protection required by Reclamation is modified according to the soil conditions. However, there is no comprehensive database on the performance and failure rates for the different types of pipelines when embedded in soils with resistivity values less than 2,000 ohm-cm, so no quantitative threshold for such data could be obtained. If this information were available, it could constitute a fourth threshold value. VARIATIONS OF GAS PIPELINE THRESHOLD VALUES Threshold values are necessary to establish the probability of pipeline failure. Although Reclamation provided a benchmark, or threshold value, in response to the committee request, it noted that “some adjustment to the computed failure rate may be warranted to compensate for this uncertainty in soil conditions.”4 In addition, changes may be needed so as to include only the failures that occurred prior to 50 years of service life and to include pipelines that had no failures. The following subsections provide a brief review of how Thresholds 1 through 3 were determined. Threshold 1: Reclamation’s Reported Threshold Value The threshold value reported by the Bureau of Reclamation was given in terms of failures per mile per year based on data shown in Table 4-1. As explained by the bureau, this threshold value was based on the number of failures (27) result- ing from external corrosion on the body of the pipe, observed from 2002 through mid-2008, and did not consider the age of the pipe. Also, the length of the gas pipeline used in the calculation included only the lengths of the pipeline systems that exhibited failures (93,523 miles).5 The reporting period was 6.5 years (2002 to mid-2008). The calculation and threshold value provided by Reclamation were as follows: 4 Gabaldon, letter to Meyer, August 21, 2008. 5 Gabaldon, letter to Meyer, August 21, 2008.

OCR for page 99
failure criteria 0 number of failures Threshold 1= (reporting period) × (pipe mileage) 27failures = (6.5years) × (93,523miles) = 0.00044failures/mile per year (Eq. 4-1) Threshold 2: Threshold 1 Adjusted for Pipe Age and Length Several potential challenges arise from the values used by Reclamation for determining the Threshold 1 value: 1. The number of failures did not include failures identified on the welds (four additional failures were identified); 2. The number of failures used in the calculation included failures that occurred after or at 50 years (27 failures were reported, but only 19 occurred before 50 years); and 3. Pipelines that did not exhibit failures were not included in the pipe mileage (an additional 206,877 miles were identified with no reported failures). Modifying the number of failures (4 + 19) and the total length of pipeline in service (300,400 miles) results in the following threshold value: 23failures Threshold 2 = (6.5years) × (300,400miles) = 0.000012 failures/mile per year (Eq. 4-2) It should be noted that the total mileage listed is only an estimate, which could vary depending on the amount of pipe in service at a specific time and on the amount of pipe that has been installed for fewer than 50 years. However, this information was not readily available to the committee, and because the objective is to assess the potential variability in the threshold value, the committee believes that this is a reasonable estimate. In addition, Reclamation noted: The DOT database does not include information on the soil conditions in which the pipelines are installed, so we are unable to further screen the data to include only pipe installed in severely corrosive soils. We are not able to quantify the impact this issue has on the calculated performance data noted above, but some adjustment to the computed

OCR for page 99
corrosion Prevention standards ductile iron PiPe 0 for failure rate may be warranted to compensate for this uncertainty in soil conditions across the data set.6 Applying factors to Equation 4-2 could change the threshold but, after significant searching, the committee was unable to determine these factors. Threshold 3: Based on Early Gas Pipeline Failure Rates Current regulations7 require that oil and gas steel pipelines be stringently inspected. This strategy has been found to be very effective at locating potential corrosion problems before failures occur. Advanced technology coupled with rig- orous internal and aboveground inspections has resulted in significant reductions in failure rates of steel pipelines used for gas or hazardous liquids in the past few decades. Because water lines are not inspected at the same level or with the same frequency as pipelines carrying gas or other hazardous liquids, the threshold pro- vided by Reclamation (Threshold 1) could be considered to be low. To assess this, the committee reviewed results from the Eiber study, an evaluation of DOT data collected between 1972 and 1984, inclusive.8 At the time (1972 to 1984) of Eiber’s study, the reporting requirements included any leak that required immediate repair or any incident that required removal from service of any segment of transmission pipeline. Furthermore, while inspection was required at that time, the requirements were minimal, had recently been imposed, and used technologies that were much less effective than those cur- rently available. Thus, the inspections did not have a significant influence on the failure rate. The pipeline inventory included pipe buried in the 1930s through the 1970s; thus most ages of pipe relevant to this study were available (up to 54 years of service life). These pipes consisted of bonded dielectric coatings with CP, but many may not have been cathodically protected for their entire life. Some of the steel gas pipelines, depending on company policy, may not have been adequately cathodically protected for the first 30 or 40 years of their life, until this was required by DOT regulations. The Eiber study eliminated duplicate and clearly invalid reports and reported nearly 400 service failures caused by external corrosion out of 5,872 incidents. The failure rate for these failures is plotted versus the age of the pipe in Figure 4-1. (The 6 Gabaldon, letter to Meyer, August 21, 2008. 7 Pipeline and Hazardous Materials Safety Administration, Code of Federal Regulations, Title 49, Part 192. 8 D.J. Jones, G.S. Kramer, D.N. Gideon, and R.J. Eiber, American Gas Association, “An Analysis of Reportable Incidents for Natural Gas Transportation and Gathering Lines 1970 Through June 1984,” NG-18 Report No. 158 (Washington, D.C.: Pipeline Research Committee of the American Gas Association, 1986).

OCR for page 99
failure criteria 05 FIGURE 4-1 Pipe age versus failure rate for older steel pipe with bonded dielectric coating and cathodic protection. age of the pipe was deduced from the decade of installation, so there are only 5 points—one per decade—with each point representing about 80 failures.) The figure shows that, as expected, the failure rate from external corrosion (a time- dependent degradation) increases as the pipes age. Furthermore, this increase is well predicted, with an exponential increase with time. Note that the rate is rapidly increasing as 50 years is approached. This is one reason that coated steel with CP may not eliminate the need for additional mitigation methods to minimize loss of service. The mileage-weighted average age of the 353 miles of DIP with PE and CP in corrosive soil (Table 4-2) was determined to be 22.5 years.9 Using the best-fit curve shown in Figure 4-1, the equivalent-age failure rate for the older steel pipes with an average age of 22.5 years was determined to be 0.000041 failures per mile per 9 Thismileage-weighted average can be calculated using the following equation:  Σ(mi ai ) mileage-weighted average age =   , where mi is the mileage of pipe age ai. ′  Σmi   

OCR for page 99
corrosion Prevention standards ductile iron PiPe 0 for TABLE 4-2 Data from Table 3-8 in This Report on All Relevant Ductile Iron Pipe with Polyethylene Encasement and Cathodic Protection in Corrosive Soils Abbreviated Total Project Average Length, li Failure Rate ti × li Project Name Age, ti (years) Age (years) (miles) Failures (failures/mile/year) WEB 24 12 150 3,600 0 0 Southwest 25 20.5 75.8 1,895 1 0.000528 Mid-Dakota 12 9 49.6 595.2 0 0 Montrose 27 27 27 729 0 0 Hanna 20 20 7 140 0 0 Santa Margarita 23 23 7.5 172.5 0 0 Vacaville 24 24 3 72 0 0 Akron 7 7 0.25 1.75 1 0.571429 Trinidad 29 27.5 10 290 0 0 California City 25 16.5 1.5 37.5 1 0.026667 Denver 24 24 9 216 0 0 Sheridan 16 15 12 192 0 0 TOTALS 352.65 7,940.95 SOURCE: In Chapter 3 of this report, see Table 3-8, “Partial List of Cathodically Protected Polyethylene- Encased Ductile Iron Pipelines.” year. This threshold value is only slightly lower than that provided by Reclamation (Threshold 1). Probability Analysis for Threshold Value The thresholds estimated in the previous subsections can be used to calculate the probability of having no failures over a 50-year period. Also, the data can be used to estimate the probability of having one or more failures. Assuming that the annual probability of failure per mile remains constant over time10 and that each year is statistically independent, the probability of having zero failures over a 50- year period for a given mile of pipe can be determined as follows: P50 years, 1 mile, 0 failures = (1 − Benchmark Failure Threshold)50 (Eq. 4-3) Using the threshold values determined in the previous subsections and Equa- tion 4-3, the probability of having no failures for a given mile of pipe over a 50- year period for each threshold value is shown in Table 4-3, and the trend in the probability with threshold is linear, as shown in Figure 4-2. The probability of 10 The failure rate generally increases over time. However, the failure rate here is for pipe having an average age near 50 years. Therefore, calculating the probability of failure using this rate will be conservative (i.e., it should overpredict the number of failures in 50 years).

OCR for page 99
failure criteria 0 TABLE 4-3 Summary of Possible Failure Rates and Resultant Probabilities Threshold Failure Rate Probability of No Probability of One or More No. (failures/mile per year) Failures in 50 Years Failures in 50 Years 1 0.000044 99.78% 0.22% 2 0.000012 99.94% 0.06% 3 0.000041 99.80% 0.20% FIGURE 4-2 Trend of probability of no failures in 50 years, with failure rate. having at least one failure on a given mile over a 50-year period is then 1 minus the probability of having no failures; these values are also shown in Table 4-3. They are used in this report as benchmark probability values for failures. If the estimated probability of failure of DIP with PE and CP is greater than the values shown in Table 4-3, it will be concluded that the DIP with PE and CP does not meet Reclamation’s performance requirements. Alternatively, if the probability is lower than the values shown in Table 4-3, it will be concluded that the DIP with PE and CP meets Reclamation’s performance requirements. Although the committee evaluated alternative methods for assessing failure rates, it does not know whether these failure rates fall within Reclamation’s risk tolerance. Reclamation did note that it believes that its analysis supports the origi- nal response to the committee’s question of what it defines as reliably providing

OCR for page 99
corrosion Prevention standards ductile iron PiPe 0 for a minimum service life of 50 years for its pipelines, and this desired value was reported to be 0.000044 failures per mile per year. To put this number in perspec- tive, using the 150 miles of South Dakota’s Walworth, Edmunds, and Brown (WEB) Project as a hypothetical system, the failure rate would be 0.0066 failures across the system per year, or 0.33 failures in the system over Reclamation’s desired 50-year service lifetime. Probability Analysis of DIP with PE and CP The committee deliberated extensively regarding the number of corrosion failures on the DOT gas steel and DIP pipelines and ultimately decided that the failure data for both types of pipelines should be treated in the same manner. The committee did not have any specific information (e.g., level of CP, date of CP instal- lation, soil corrosivity, or disbonded coatings) about individual gas pipeline failures in the DOT data set that could be used to disqualify specific failures. The committee thus applied the same standards to the failure of DIP pipelines with CP and PE. As has been noted, if the cathodic protection systems on the gas pipelines were working correctly, the number of failures should theoretically be near zero. However, because of other factors, there have been—and will continue to be—corrosion failures on dielectrically bonded steel pipelines with CP. Therefore, the committee elected to treat the DIP failure data in a manner consistent with the treatment of the DOT gas pipeline failure data. A detailed review of case histories was performed earlier in the report that included an assessment of all known pipelines of DIP with PE and CP. Within this data set, pipeline failures of DIP with PE and CP were identified. It should be noted that there were not widespread failures of DIP with PE and CP. Table 3-8 in Chapter 3 documented that the total pipeline length buried in highly corrosive soils was approximately 353 miles. Table 3-9 indicated that a total of nine cases of failures on DIP with PE and CP from 1983 to 2006 occurred at four differ- ent locations. Of the nine failures reported, six occurred at the Vernal, Utah, site (Chapter 3, Table 3-9, Rows 3a, 3b, and 3c). This pipeline was a DIP sewer line and was protected with PE and CP. Because this was a sewer line and the possible effects of sewage leaks under the PE on the corrosion could not be determined, the committee could not reach agreement on whether to use these failures in the analysis, and they were therefore excluded. There was a consensus among the committee members that the Southwest Pipeline leak (Table 3-9, Row 1) should be included as a valid failure location. For the Akron, Ohio, location (Table 3-9, Row 2), there was some concern that the CP may not have been functioning correctly and that the side of the joint with the leak may not have been cathodically protected. There is an equal chance that the side of the pipe joint with the leak was connected to the same side as the

OCR for page 99
failure criteria 0 anode(s) and therefore connected into the CP system. Alternatively, if the leak was located between the anodes, there is an equal chance that the joint was protected from both directions. The type of joint bond used in Akron appears to be a solid- or rigid-type bond wire, as shown in the photos in the failure report (see Figure 3-8 in Chapter 3). This type of joint bond, once recommended by DIP manufacturers, has historically had problems (breaking or coming loose from pipe) if the pipe moves. It looks as though it broke in this instance, but the timing of the break is not known. Because the committee could not identify how or when the joint bond broke, it also could not ascertain what effect this break may have had on the level of CP at the leak location. To discount this location because of these reasons would be inconsistent with how the DOT pipeline data (which did not eliminate any breaks) was inter- preted. Therefore, the majority of the committee members agreed that this location should be considered a failure. For the California City site (Table 3-9, Row No. 4), some committee members argued that because the CP was not installed concurrent to the pipe installation (4 years difference), this failure location should be discounted. However, as noted by Graham Bell in his presentation to the committee,11 it is normal to have some delay (up to several years) in installing CP on water pipelines. For the DOT gas pipelines with which these data are being compared, many of the DOT gas pipelines were not cathodically protected until the DOT regulations mandated the use of CP in the 1970s. Thus, it is likely that the DOT gas pipeline database includes some older pipelines (pre-1970) that did not have CP for 30 to 40 years, until the DOT regulations went into effect. Based on this reasoning, the initial failure analysis will be performed assuming three failures (failures in Rows No. 1, 2, and 4 from Table 3-9). It should also be noted that the same approach was used by Reclamation to determine the failure rate for the DOT gas pipelines that is used for this initial analysis. As described below, the committee used an alternative analysis to verify the procedure. Employing the methodology used by Reclamation, the number of failures (three), the reporting period (23 years), and the pipe length in corrosive soils (353 miles), the probability of having at least one failure on a given mile over a 50-year period can be determined using Equation 4-3. It should be noted that the three failures occurred at 6 years, 8 years, and 19 years, respectively—all significantly less than 50 years—and it is known from Table 3-8 that all 353 miles of DIP with PE and CP were embedded in highly corrosive soils. Using this information, the annual probability of failure for a given mile of DIP with PE and CP embedded in highly corrosive soils can be determined as follows: 11 Graham E.C. Bell, Schiff Associates, “Measurements of Performance of Corrosion Control Mecha- nisms on DIP,” presentation to the committee, Washington, D.C., July 29, 2008.

OCR for page 99
corrosion Prevention standards ductile iron PiPe 0 for 3 failures Failure Rate DIP with PE and CP = 23 years × 353miles = 0.0003 failures/mile per year (Eq. 4-4) The probability of having zero failures in a 50-year period for a given mile of DIP with PE and CP embedded in highly corrosive soils can then be determined as follows: P (0 failures)DIP, PE, CP 50 years, 1 mile (%) = 100 × (1 − Failure Rate)50 = 100 × (1 − 0.00037)50 = 98.2% (Eq. 4-5) The percent probability of having zero failures for DIP with PE and CP is less than all values shown in Table 4-3, indicating that DIP with PE and CP will likely not meet Reclamation’s requirement. However, to assess this further, the committee considered an alternative approach. An alternative approach to compare the performance of the DIP with PE and CP with the Reclamation benchmark is to estimate a weighted average of the failure rate of the DIP with PE and CP. Data from Table 3-9 are shown again in Table 4-2, with individual failure rates for each pipeline. Given the small mileage of DIP that is relevant to this analysis and the fact that only three failures were deemed valid, a determination of failure rate is difficult. The committee considered this problem for some time to arrive at the following alternative approach to estimate the fail- ure rate. The failure rate for each pipeline project can be calculated based on the number of failures, N (1 or 0 here), its length, l, and the time from its installation to the present date or date of removal from service, t: N Failure Rate DIP with PE and CP = ′ l ×t (Eq. 4-6) For example, for the Southwest Pipeline, 1 Failure Rate DIP with PE and CP = ′ 75.8 miles × 25years = 0.000528failures/mile per year (Eq. 4-7) For the Akron pipeline,

OCR for page 99
failure criteria  1 Failure Rate DIP with PE and CP = ′ 0.25miles × 7 years = 0.571failures/mile per year (Eq. 4-8) And for the California City pipeline, 1 Failure Rate DIP with PE and CP = ′ 1.5miles × 25 years = 0.0266failures/mile per year (Eq. 4-9) For the remaining mileage, RM, of pipeline, 0 Failure Rate DIP with PE and CP = = 0 failures/mile per year ′ 1×t (Eq. 4-10) The various failure rates can then be averaged using a weighting based on the length of each pipeline (li) multiplied by its years of service (ti) divided by the sum of the products of the length of each project the age of that project and (Σ(li × ti)), to obtain an average failure rate: li × t i Failure Rate′ with PE and CP = × (Failure Rate) Σ(li × t i ) DIP (Eq. 4-11) Using these data, Equation 4-11 was evaluated to obtain a length-time- weighted average failure rate for DIP with PE and CP in corrosive soils: 1, 895 =× Failure Rate1 + Failure Rate′ 7, 941 Average DIP with PE and CP 1.75 37.5 6, 607 × Failure Rate 2 + × Failure Rate 3 + × Failure Rate 4 (Eq. 4-12) 7, 931 7, 941 7, 941 Failure Rate′Average DIP with PE and CP = 0.00126 + 0.000126 + 0.000126 = 0.000378 failures/mile per year (Eq. 4-13)

OCR for page 99
corrosion Prevention standards ductile iron PiPe  for This is also the failure rate that would have been obtained if all of the pipelines had been combined into one 353-mile pipeline and three failures had occurred during a period of 22.5 years (the length-weighted average age of all the pipelines). Comparing this average failure rate to the Reclamation benchmark threshold of 0.000044 failures/mile per year indicates (Threshold 1) that DIP with PE and CP has a significantly higher failure rate than that desired by Reclamation. The average failure rate for DIP with PE and CP in corrosive soils is about the best estimate that the committee could deduce given the paucity of data. As noted in the earlier discussion, there were other failures of DIP, but they were not considered. Probability estimates were also made to determine if the three failures could be reasonably expected if DIP had the same behavior as the benchmark. In fact, the committee examined many facts that could not be well quantified. Comparison of DIP with PE and CP with Threshold Values Using the data in Table 4-3 for steel pipe with bonded dielectric coating and CP and the probability of having no failures over a 50-year period determined in Equation 4-5, the committee can provide a reasonable comparison of likely perfor- mance. For the DIP with PE and CP, the probability of having at least one failure on a given mile over a 50-year period is 1.8 percent (100 percent – 98.2 percent). Because 1.8 percent is much greater than all alternate values reported in Table 4-3, the probability of having at least one failure on a given mile of DIP with PE and CP is likely significantly higher than the benchmark threshold values reported in Table 4-3 for steel pipe with bonded dielectric coatings and CP. In addition, 1.8 percent is significantly higher than the 0.22 percent value determined from Reclamation’s desired failure rate (Threshold 1). This suggests that DIP with PE and CP does not provide a reliable 50-year service life, as defined by Reclamation, when embedded in highly corrosive soils (less than 2,000 ohm-cm). It should be noted that the assumptions used in this analysis are based on the data identified during the time of this study. Even though considerable effort was spent on identifying relevant data, there is limited information on pipe failures. DIP with PE and CP is relatively new, with the oldest such pipeline being 29 years old, with only 353 miles in corrosive soils, and three failures in water transmis- sion pipelines. Although the committee used different approaches to evaluate the threshold failure rate provided by Reclamation, it realizes that the overall assess- ment is based on limited data. To assess how small variations in the data used to assess the probability of having at least one failure on a given mile of DIP with PE and CP over a 50-year period affect the comparison with the benchmark values, a sensitivity analysis was performed. A sensitiity analysis is a tool used to determine how changes in the input variables (in this analysis, the number of failures, the reporting period, and

OCR for page 99
failure criteria  the length of pipe in highly corrosive soil conditions) can influence the outcome of an analysis (in this case the probability of having one failure in 50 years). To perform a sensitivity analysis, baseline values are used. The baseline values for this analysis were three failures, a 23-year reporting period, and 353 miles of DIP with PE and CP embedded in highly corrosive soil. Baseline values are varied as a percentage from the original value and plotted against the outcome, reported as a percentage of the original value—the original value in this case being 1.8 percent. Figure 4-3 shows the sensitivity analysis.12 Note that if only two failures were valid, the number of failures would be reduced by 33 percent:  3− 2   3 × 100   which would in turn reduce the probability of one failure by 33 percent to approxi- mately 1.2 percent—if one failure were valid the probability of failure would be approximately 0.6 percent. Of significance is that both values (one and two failures) result in greater probabilities of failure than the threshold value of 0.22 percent defined by Reclamation. The initial calculation used the maximum known length of DIP with PE and CP and 23 years as the reporting period. It can be seen that if the length of pipe embedded in highly corrosive soil is reduced or the reporting period is shortened (while keeping the other variables constant—for example, the three failures) the probability of failure increases at a significant rate, indicating that the probability of failure is sensitive to these variables. Reducing either of these two values makes it less likely that DIP with PE and CP will meet the required threshold limits shown in Table 4-2, and more specifically that set by Reclamation. Alternatively, increasing the length of pipe embedded in highly corrosive soil or increasing the reporting period decreases the probability of failure. However, it can be seen that the probability of failure is less influenced as the length of pipe and reporting period increase (i.e., the slope of the lines are lower). As already noted, the number of failures is a significant variable for both positive and negative changes and, as expected, is directly correlated to the probability of failure. What is important to note is that the failure rate of the DIP with PE and CP would have to be decreased by almost 90 percent (to 0 failures) to meet the high- est threshold value (Threshold 1) shown in Table 4-3. The threshold values from 12 As an example of how to use Figure 4-3: assume that an additional 353 miles of DIP with PE and CP embedded in highly corrosive soils were identified and that this pipe exhibited no failures; this would increase the pipe length from 353 to 706 miles, a 100 percent increase in pipe length; from this 100 percent change in variable point on the abscissa, draw a line up to the curve that represents “length of pipe”; at this intersection draw a horizontal line to the ordinate, indicating that the original probability of failure of 0.018 will be reduced by 50 percent to 0.009.

OCR for page 99
corrosion Prevention standards ductile iron PiPe  for FIGURE 4-3 Sensitivity plot for probability of failure of ductile iron pipe with polyethylene encasement and cathodic protection. Table 4-3 are shown in Figure 4-3 as the dashed horizontal lines. The length of pipe embedded in highly corrosive soil would have to increase by over 750 percent (to almost 3,000 miles) to meet the highest threshold value shown in Table 4-3 (Threshold 1) and almost 2,900 percent (11,000 miles) to meet the lowest thresh- old value shown in Table 4-3 (Threshold 2). The reporting period would have to range between approximately 200 and 700 years to meet the three threshold values shown in Table 4-3. Based on the pipeline data described in Chapter 3, it is unlikely that any of the variables (number of failures, pipe length, or reporting period) would change suf- ficiently to make the probability of having one failure less than the threshold values shown in Table 4-3. Moreover, it is less likely that the data would change sufficiently for the probability of having one failure to be less than that originally required by Reclamation (Threshold 1). This indicates that the analysis is sufficiently robust and not sensitive to the assumptions made in this analysis. Or, stated otherwise, reasonable changes in the input variables (three failures, 353 miles of DIP with PE and CP, and a 23-year reporting period) will not reduce the probability of failure of DIP with PE and CP below the threshold values shown in Table 4-3, or, more importantly, below the threshold values desired by Reclamation.

OCR for page 99
failure criteria 5 Comparison of Corrosion Rates To further examine the finding that DIP with PE and CP does not meet the Reclamation threshold, the committee considered the maximum observed linear- ized corrosion rates from Tables 3-2, 3-4, 3-7, and 3-9 in Chapter 3. Table 3-9 treats data for DIP protected with PE and CP. Most of the data available were for failed pipes, for which the corrosion rate was calculated by dividing the wall thickness by the years of service before failure. Table 3-9 includes data on the failed wastewater lines not used in the failure analysis earlier, but those data represent the lowest corrosion rates in the table. When the maximum corrosion rate data available for bare and as-manufactured DIP (Table 3-2), DIP with damaged PE (Table 3-4), and DIP with undamaged PE (Table 3-7) are examined, it is found that in these categories as well as in the subject category of DIP with PE and CP above, the worst-case corrosion rate can well exceed the corrosion rate thought to be acceptable for reliable service. The committee recognizes that all of these data were selected as the maximum observed corrosion rates and therefore are not typical of the expected performance of most pipeline segments; they represent the tails of the failure distributions. All data in these tables suggest that there are conditions in which these pipes will corrode at undesirable rates and that these data reinforce the conclusion that it cannot be ensured that DIP with PE and CP will achieve the reliability expected by Reclamation. SUMMARY The committee used various assumptions and methods to calculate different threshold values. These threshold values were then used to compare the perfor- mance of DIP with PE and CP and to determine whether DIP with PE and CP meet the various threshold requirements, but more specifically the threshold require- ment provided by the Bureau of Reclamation. The probability analysis based on annual failure rates indicated that DIP with PE and CP does not meet any of the alternative threshold values as reported in Table 4-3 and that it is almost an order of magnitude higher than the threshold value requested by Reclamation. Two sen- sitivity analyses based on three valid failures supported the conclusion that DIP with PE and CP likely exhibits poorer performance than the threshold requirement provided by Reclamation. Comparison of maximum observed pitting rates for vari- ous DIP systems indicates that rates at the tail of the distribution can be similar for bare DIP pipe systems, as-manufactured DIP pipe systems, DIP with damaged PE, DIP with undamaged PE, and DIP with PE and CP. In summary, based on the information collected for this report, DIP with PE and CP will not “reliably” (quot- ing the committee’s statement of work) provide a minimum service life of 50 years and will not meet the threshold requirement set by Reclamation.