U.S. Pipeline Incident Analysis

by

Richard Stover, PhD

A review of a US Department of Transportation / Pipeline & Hazardous Materials Safety Administration document on pipeline safety can be found here. The review is pretty self-explanatory. But email me if you have questions.

Note added Sept. 5, 2013: In the review noted above I show that pipeline safety, as measured by injury rate, has not been improving since 2001. The government document presents the data but fails to note the lack of safety improvements since 2001. It turns out that the pipeline industry is even better at hiding the data. Their publication “Safe and Getting Safer” obscures the facts by showing bar charts that don't even bother to present the yearly data. They just lump years 1999 to 2001 and compare that with the years 2008 to 2010. Done this way, with just two points one gets the (incorrect) impression that things are “getting safer.” Measured by the number of injuries that hasn't been the case for a long time. The industry used number of barrels spilled as an indicator. When all of the data from all years is considered and presented, there is no improving trend in the net number of barrels spilled. The industry claim is worthless because: 1) it is based on a proprietary data set that can't be verified, 2) they use cherry-picked data, 3) two points on a plot doesn't constitute a meaningful trend, 4) they fail to consider all years, and 5) they give no rational definition of “Safe.”

I produced a 1-minute video showing pipeline incidents in the US since 1986. The video can be seen on YouTube at the Center for Biological Diversity. The Center added specific messages to the video addressing Keystone XL issues.

The raw one minute video (same as the Center video but without Keystone messages) can be downloaded and viewed with your own computer software. (Try right-click if your browser does not ask you if you want to Save the file.)

Or a 30-second version of the same video is available for news outlets.

If you play the video, pause it near the end of the video. Then in another browser window take a look at this map from the U.S. Energy Information Administration. Notice any similarity?

Video background information

The pipeline incident video was generated with publicly available data from the Federal Government. The federal Department of Transportation operates the Pipeline & Hazardous Materials Safety Administration or PHMSA (http://phmsa.dot.gov/pipeline). Pipeline operators submit reports to the PHMSA and the PHMSA prepares summary reports and files. The files used to make the video are:



PHMSA file name

What it is

gt1986to2001.csv

Gas transmission incidents from 1986 through 2001

gt2002to2009.csv

Gas transmission incidents from 2002 through 2009

gt2010toPresent.csv

Gas transmission incidents from 2010 through 2013 (partial)

hl1986to2001.csv

Hazardous liquid incidents from 1986 through 2001

hl2002to2009.csv

Hazardous liquid incidents from 2002 through 2009

hl2010toPresent.csv

Hazardous liquid incidents from 2010 through 2013 (partial)

gd1986tofeb2004.csv

Gas distribution incidents from 1986 through February 2004

gdmar2004to2009.csv

Gas distribution incidents from March 2004 through 2009

gd2010toPresent.csv

Gas distribution incidents from 2010 through 2013 (partial)

Gas transmission refers to major pipelines that move gas over long distances and gas distribution refers to the network of pipes that deliver gas to end-users. Hazardous liquid refers to pipelines that carry oil, diesel, and other hazardous liquids.

Each event in the PHMSA files is marked as “Significant” or not-significant. The term Significant is defined by the PHMSA (see http://primis.phmsa.dot.gov/comm/reports/safety/sigpsi.html). The video shows significant events only.

U.S. Census Bureau county map data was also used in making the video. Many of the incidents in the PHMSA files include geographic longitude and latitude information. But many incidents do not have the longitude and latitude (especially the older records) and sometimes the recorded longitude or latitude is incorrect or improperly entered. Fortunately, almost all incidents record the state and county in which the incident occurred. Whenever the longitude and/or latitude is not available the incident is placed within the county using the Census Bureau county map data. (If an incident does not have a recognizable county recorded and no longitude/latitude those incidents are not plotted.)

Along with the geographic data the incident files record the date (and sometimes the time of day) of the incident. This information is used to place the incidents on the map in the correct chronologic order as the video is played. (Incidents that occur within about 0.1 second of each other as the video plays are grouped together so they actually all appear at the same time. But this is only noticeable if the video is played a frame at a time.)

Each incident is shown as a small round dot when the video is played. The incident dots are color coded and the hazardous liquid pipeline events are sized according to the gross barrels of liquid lost.

The color coding is chosen in this manner:

  1. Any incident involving fatalities is colored red.

  2. Any other incident involving injuries is colored yellow.

  3. Any other gas pipeline incident is colored blue.

  4. Any other hazardous liquid incident is colored dark red.

The data used to draw the state borders is approximate. The U.S. Census county map data is more accurate but is not draw because it clutters the map with too many lines.

The program used to process all of the data files and create the video file was written entirely in Objective-C and runs on an Apple iMac. The video file can then be converted to other video formats using standard Apple Mac software.

I have ignored incidents in Alaska, Hawaii, Puerto Rico, the Virgin Islands, and ones without dates. There are 7978 incidents remaining. 7184 have geo-positions and those are shown in the video.