APPENDIX B
Pipeline Safety Data and Trends in the United States
Pipeline safety data are compiled by the Office of Pipeline Safety (OPS) of the U.S. Department of Transportation (USDOT), the Energy Information Administration of the U.S. Department of Energy, the Bureau of Transportation Statistics, and the pipeline companies themselves. Only the national data were available to the committee. This appendix provides a brief overview of pipeline safety data and trends. It does not provide a comprehensive assessment of pipeline safety.
DATABASES
A number of data sets could be used in reviewing various components of pipeline safety. OPS maintains the Hazardous Liquid Pipelines Accident Report database, which was established in 1970, revised in 1986, and revised again in January 2002 (when the threshold for reporting hazardous liquids pipeline accidents was reduced from 50 barrels to 5 gallons) (NTSB 2002). Irrespective of the volume spilled, any pipeline incidents in which damage exceeded $50,000 or there was an injury, fatality, fire, or explosion must be reported to OPS. The current OPS reporting requirements follow closely a voluntary industry program, the Pipeline Performance Tracking System (PPTS), which became effective January 1, 1999.
OPS maintains the Natural Gas Gathering and Transmission Systems Incident Database, which was revised in 1984 and again in 2001. Finally, OPS maintains the Natural Gas Distribution Systems Incident Database, whose reporting requirements are the same as those for gathering and transmission pipelines. This database, which was last revised in 1984, has recently been redesigned (NTSB 2002). OPS requires natural gas pipeline operators to report each failure that results in fatalities, injuries that require
in-patient hospitalization, property damage (including cost of gas lost, of the operator or others, or both) of $50,000 or more, or ignition of gas (49 CFR 191.3, amended in 2001).
The numbers vary from one database to another because reporting criteria, definitions, incident causation classifications, and other categories differ from one database to another and within a database across time. For example, some pipeline product releases go undocumented because they do not meet the federal requirements for reporting. This results in an underreporting of releases and impacts. In addition, there are reporting errors, missing data, and preliminary data that are not subsequently updated. These problems make it hard to analyze the data and difficult to draw statistically significant conclusions.
INCIDENT DATA AND TRENDS
From 1989 through 1998, 226 people died (a yearly average of 22.6 deaths) and 1,030 people were injured (a yearly average of 103 injuries) in major pipeline accidents.1 In addition, according to GAO (2000), major pipeline accidents from 1989 through 1998 caused about $700 million in property damage. Hazardous liquids pipelines accounted for about $350 million, or 50 percent, of this property damage.2
In 2001, there were a total of 44,969 transportation fatalities, none of which were attributable to liquids pipeline incidents and 7 of which were attributable to natural gas pipeline incidents. In 2002, the number of transportation fatalities increased to 45,098. Of these, 1 was due to a liquids pipeline incident and 10 were related to natural gas pipeline incidents (NTSB 2003). A comparison of the accident rates for the different
methods used to transport petroleum indicates that pipelines are the safest overall method. Only rail is safer in terms of injuries per ton-mile, and only barges are safer in terms of deaths per ton-mile. The rate of fatalities, injuries, and fires and explosions per ton-mile of oil transported for all other modes is typically at least twice—and in some cases more than 10 times—as great as the rate for pipelines. Trucks are, by far, the least safe method for transporting petroleum (see Table B-1).
From 1985 through 2001 there were 1,417 reportable natural gas pipeline safety incidents, an average of 83.4 per year; 1,159 were for natural gas transmission and gathering system incidents. Of these, 910 incidents, or 53.5 each year on average, occurred on onshore transmission and gathering systems (see Table B-2). Accidents on natural gas transmission and gathering pipelines accounted for 58 deaths in 25 incidents; 3 of the incidents accounted for 30 deaths. During the 1985–1987 period, there were 12.1 reportable injuries per year on average. For the 7-year period 1995–2001, the number of reportable injuries dropped to an average of 6.7 per year.
For the 3-year period 1985–1987, annual natural gas consumption in the United States was 16.9 trillion cubic feet (tcf), with an average of 70 reportable safety incidents per year and an average of 5.5 reportable safety incidents per year per tcf (see Table B-3). By 1999–2001, the average annual consumption had increased 29 percent to 21.8 tcf per year. For the same time period, the average annual number of safety incidents dropped to 63 and the number of reportable safety incidents per year per tcf dropped by 40 percent to 3.4. Thus, this downward trend in reportable incidents was accompanied by a 30 percent increase in natural gas consumption.
TABLE B-1 Relative Occurrence of Transportation Accidents per Ton-Mile of Oil Transported, 1992–1997
Event |
Pipelinea |
Rail |
Tank Ship |
Barge |
Truck |
Fatality |
1.0 |
2.7 |
4.0 |
10.2 |
87.3 |
Injury |
1.0 |
2.6 |
0.7 |
0.9 |
2.3 |
Fire/explosion |
1.0 |
8.6 |
1.2 |
4.0 |
34.7 |
a The rates of occurrence are based on a value of 1.0 for pipeline. Values of less than 1.0 indicate a better safety record. SOURCES: Trench 1999; GAO 2000. |
TABLE B-2 Safety Performance of Natural Gas Transmission and Gathering Systems, 1985–2001
|
Total |
Average per Year |
Reportable safety incidents |
1,417 |
83.4 |
Safety incidents (transmission and gathering) |
1,159 |
68.2 |
Safety incidents (onshore transmission and gathering) |
910 |
53.5 |
Safety incidents (offshore transmission and gathering) |
249 |
14.6 |
SOURCE: Trench and Selig 2003. |
Only a small fraction of natural gas pipeline operators experience safety incidents. In 1998, 56 operators reported incidents; by 2000, this number had dropped to 33. No more than 6.5 percent of operators have had reportable incidents in any year. Furthermore, 90 percent of incidents (809 of 899) occur in unpopulated areas, and third-party damage is the single largest cause of onshore incidents at 23 percent.
Hazardous liquids pipeline summary incident statistics show a comparable downward trend (see Table B-4). From 1986 through 1989, there were an average of 201 incidents with 27 injuries and 3 fatalities per year. From 2000 through 2002, the annual average number of incidents had decreased to 140, with 5 injuries and 0.7 fatalities per year. During the same time period the annual average number of gross barrels of liquids lost decreased from 270,000 to just over 100,000 (see Table B-4).
CAUSALITY
NTSB, the USDOT Inspector General, the New Jersey Institute of Technology, and others have reported that OPS data on pipeline incidents and infrastructure are limited and sometimes inaccurate. Until recently, OPS’s
TABLE B-3 Comparison of Consumption and Reportable Safety Incidents for Natural Gas Pipelines for Two 3-Year Periods, 1985–1987 and 1999–2001
|
1985–1987 |
1999–2001 |
Consumption, tcf/year |
16.9 |
21.8 |
Reportable safety incidents per year per tcf (all operators) |
5.5 |
3.4 |
Safety incidents per year (transmission and gathering) |
70 |
63 |
SOURCE: Trench and Selig 2003. |
TABLE B-4 Summary Accident Statistics for Hazardous Liquids Pipeline Operators, 1986–2003
incident report forms used only five categories of causes for incidents on natural gas distribution pipelines, four categories for those on natural gas transmission pipelines, and seven categories for those on hazardous liquids pipelines. As a result, about one-fourth of all pipeline incidents were attributed to “other causes,” which limited OPS’s ability to adequately identify and address causes of incidents. In addition, data on pipeline mileage in various infrastructure categories (such as age or size) are necessary for a meaningful comparison of the safety performance of individual pipeline companies. OPS did not require hazardous liquids pipeline operators to submit this type of data and did not collect complete data from natural gas pipelines. Given these limitations, in 2001 OPS revised its incident report forms for hazardous liquids and natural gas transmission incidents to include 25 categories of causes, and in early 2003 OPS revised its natural gas distribution incident form (GAO 2002, 7).
Data indicate that third-party damage is the leading cause of onshore natural gas transmission and gathering system incidents and liquid transmission system incidents, accounting for 28 percent and 41 percent of all such incidents, respectively (see Tables B-5 and B-6). External corrosion is the second leading cause, accounting for 17 percent of natural gas pipeline incidents and 21 percent of liquids pipeline incidents. Of those causes listed on the reporting form, vandalism and malfunction are the least likely to result in pipeline incidents. Vandalism accounts for fewer than 1 percent of incidents.
On the basis of voluntarily reported liquids pipeline data in PPTS Advisories 8, 9, and 10, 7 percent of 1,882 total spills were caused by
TABLE B-5 Natural Gas Transmission/Gathering Systems—Cause of Onshore Incidents, 1985–2001
TABLE B-6 Liquids Pipelines—Cause of Incidents, 1996–2000
Cause |
Line Pipe (%) |
Tank/Pump (%) |
Third-party damage |
41 |
5 |
Corrosion |
21 |
22 |
Equipment |
4 |
45 |
Unknown |
11 |
0 |
Incorrect operation |
6 |
8 |
Miscellaneous |
1 |
17 |
Manufacturer |
6 |
0 |
Construction/repair |
4 |
1 |
Weather |
3 |
1 |
Previously damaged pipe |
3 |
– |
Vandalism |
0 |
– |
Total |
100 |
100 |
SOURCE: Trench 2002. |
third-party damage. These spills accounted for 100 percent of incidents causing death, 52 percent of incidents involving an injury, 28 percent of incidents involving fire or explosion, 35 percent of incidents affecting public safety, 56 percent of all volumes released from onshore pipelines, and 54 percent of the largest 2 percent of spills (more than 5,575 barrels). For 75 percent of liquids pipeline incidents caused by third-party damage, the failure occurred at the time of the incident. The failure was due to a prior incident in 17 percent of the cases, and the failure was due to “other” damage (e.g., vehicle accident) in 8 percent of the cases. A breakdown of liquids pipeline incidents involving failure at the time of the incident is presented in Table B-7. Finally, the primary cause of third-party damage reported by the operator was failure to use one-call.
Kiefner and Trench (2001) examined oil pipeline characteristics and risk factors for pipelines constructed from before the 1930s to the present. Twenty-three percent of existing liquids pipelines (measured in terms of mileage) were constructed in the 1960s, only 2 percent before 1930, 7 percent in the 1940s, and another 7 percent in the 1990s. For the pre-1930s pipelines, the incident rates were more than four times higher than for all pipelines combined, whereas for all periods thereafter, the incident rates were approximately equal to the representation (i.e., average) rate. There has been a small downward trend in incident rates for pipelines constructed in the 1930s and more recent periods (see Table B-8). In addition, the greatest percentage (more than 4.0 percent) of third-party damage occurs to pre-1930s liquids pipelines. The smallest percentage of third-party damage (0.3 percent) affects pipelines constructed in the
TABLE B-7 Liquids Pipeline Incidents Involving Failure at Time of Incident
|
Incidents |
Barrels Released |
||
Number |
Percentage |
Total |
Percentage |
|
Landowner (subtotal) |
25 |
37 |
37,711 |
44 |
Landowner—farming |
19 |
28 |
18,717 |
22 |
Landowner—homeowner |
6 |
9 |
18,994 |
22 |
One-call partners |
18 |
26 |
19,008 |
22 |
Additional industrial/commercial activity |
17 |
25 |
24,312 |
28 |
Road construction/maintenance |
9 |
13 |
5,534 |
6 |
Total |
69 |
100 |
86,565 |
100 |
TABLE B-8 Pipeline Mileage and Incidents for Liquids Pipelines Constructed Each Decade from the Pre-1930s to the Present
Decade of Pipeline Construction |
Percentage of Existing Pipelines (in miles) |
Percentage of All Reportable Incidentsa |
Percentage of Reportable Incidents due to Third-Party Damagea |
Pre-1930s |
2 |
>4 |
4.0 |
1930s |
7 |
1.2 |
1.0 |
1940s |
13 |
1.1 |
1.5 |
1950s |
22 |
0.9 |
1.0 |
1960s |
23 |
1.0 |
1.0 |
1970s |
17 |
0.7 |
0.7 |
1980s |
9 |
0.8 |
0.6 |
1990s |
7 |
0.5 |
0.3 |
2000+ |
<1 |
Incomplete |
Incomplete |
a Data are the percentage of all reportable incidents divided by the percentage of existing mileage. 1.0 indicates that the incident data are directly proportional to the amount of pipeline in that age category. |
1990s (see Table B-8). In comparison with past years, recent experience with third-party damage includes the following: steel is now less brittle, encroachments are more frequent, farming techniques are more invasive, and depth of cover is greater. Thus, pre-1930s transmission pipelines have a higher likelihood of problems, whereas the difference from one decade to the next since then is not very significant.
According to FERC (2003), older natural gas pipelines (i.e., those installed in 1950 or earlier) exhibit a significantly higher rate of incidents compared with those installed since 1950. This may be partially due to a higher frequency of corrosion, which is a time-dependent process. However, since July 1971 pipelines have been required to have external protective coatings and cathodic protection to reduce corrosion potential.
The changes in the reporting forms, which have been adopted recently, should enable OPS and others to better understand the causes of incidents so that efforts to improve pipeline safety will be more likely to reduce the number and severity of incidents and failures. In addition, a comprehensive, detailed, viable database on pipeline incidents and an inventory of the pipeline infrastructure can be used to establish quantifiable performance measures by which the effectiveness of the integrity management and other risk management programs may be evaluated (GAO 2002).
REFERENCES
Abbreviations
FERC Federal Energy Regulatory Commission
GAO General Accounting Office
NTSB National Transportation Safety Board
OPS Office of Pipeline Safety
FERC. 2003. Grasslands Pipeline Project. Final Environmental Impact Statement. FERC/ EIS-0154F. Washington, D.C.
GAO. 2000. Pipeline Safety: The Office of Pipeline Safety Is Changing How It Oversees the Pipeline Industry. GAO/RCED-00-128. Washington, D.C.
GAO. 2002. Status of Improving Oversight of the Pipeline Industry. Testimony before the Subcommittee on Energy and Air Quality, Committee on Energy and Commerce, House of Representatives. GAO-02-517T. Washington, D.C., March 19.
Kiefner, J. F., and C. J. Trench. 2001. Oil Pipeline Characteristics and Risk Factors: Illustrations from the Decade of Construction. Report prepared for API. Washington, D.C.
NTSB. 2002. Safety Report: Transportation Safety Databases. NTSB/SR-02/02. Washington, D.C.
NTSB. 2003. National Transportation Safety Board 2001–2002 U.S. Transportation Fatalities. Washington, D.C., Oct. 2. www.ntsb.gov/Pressrel/2003/trans_fatalities_ 2001-2002.htm.
OPS. 2003. Hazardous Liquid Pipeline Operators Accident Summary Statistics by Year, 1/1/1986–6/30/2003. Washington, D.C., Sept. 17. ops.dot.gov/stats/lq_sum.htm.
Trench, C. J. 1999. The U.S. Oil Pipeline Industry’s Safety Performance. Allegro Energy Group, New York, May.
Trench, C. J. 2002. The U.S. Oil Pipeline Industry’s Safety Performance. Allegro Energy Group, New York, March.
Trench, C. J., and B. J. Selig. 2003. The Safety Performance of Natural Gas Transmission and Gathering Systems. Allegro Energy Group, New York, April.