1. ILLINOIS - RAILTEC Slide 1 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Pipeline Data Analysis Crude Oil Pipeline Transportation Risk Analysis Jeff LaHucik Rail Transportation and Engineering Center – RailTEC Department of Civil & Environmental Engineering University of Illinois at Urbana- Champaign, U.S.A. http://www.ucsc- extension.edu/sites/default/files/imce/public/images/134550607.j pg

2. ILLINOIS - RAILTEC Slide 2 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Data Source Data was collected from the U.S. Department of Transportation Pipeline and Hazardous Materials Safety Administration (PHMSA) Pipeline Safety Incidents Database, which was current through April 30, 2013. The database contains all pipeline incidents, without a minimum release size, that occurred within the United States as well as in offshore pipeline. http://primis.phmsa.dot.gov/comm/reports/safety/SIDA.html?nocache=4368 http://pipelineonlinereporting.phmsa.dot.gov/images/phmsa.jpg

3. ILLINOIS - RAILTEC Slide 3 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Data Scope The data used was from January 1, 2002 through April 30, 2013. All onshore, crude oil pipeline incidents that fell within the above dates and that had valid latitude and longitude coordinates were considered. Any commodity other than crude oil was ignored since this analysis focused specifically on crude oil incidents. Offshore incidents were ignored since they experience different failure mechanisms than onshore pipeline. Incidents without latitude and longitude coordinates were ignored since they would be incompatible with the analysis method.

4. ILLINOIS - RAILTEC Slide 4 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Data Used for the Analysis 1811 crude oil pipeline incidents occurred during the time period being analyzed. 1616 incidents were considered in this analysis Due to invalid/missing latitude and longitude coordinates, the remaining 195 incidents were not used On average, 165 crude oil pipeline incidents occurred per year. QuantityMeanMinimumMaximum Release size (barrels) 79.56031,322 Total cost (US $) 1.12 million0810 million Exposure radius (miles) 0.108010.71

5. ILLINOIS - RAILTEC Slide 5 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Crude Oil Pipeline Incidents by Year Note: all 1811 incidents were used in this figure.

6. ILLINOIS - RAILTEC Slide 6 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Incident Cause Georeferenced Data The database was split into 7 incident cause categories: Corrosion failure Equipment failure Incorrect operation Material failure External impact Excavation damage Other The GPS coordinate incident reporting procedure was not viable for use in GIS. Therefore, all latitude and longitude coordinates were transformed to a GIS-friendly format (decimal degrees).

7. ILLINOIS - RAILTEC Slide 7 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Crude Oil Pipeline Incidents Note: incidents range from January 2002 through April 2013; 1616 incidents are represented in the figure.

8. ILLINOIS - RAILTEC Slide 8 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Crude Oil Pipeline Incidents by Release Size

9. ILLINOIS - RAILTEC Slide 9 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Frequency/Severity Analysis of Incident Causes A severity vs. frequency analysis was done in order to obtain a basic intuition of the comparative risk of each incident cause. Frequency is the number of incidents per year. Severity is the average cost per incident (U.S. $). Costs included in severity: operator costs socioeconomic costs cleanup costs cost of released crude oil environmental damage costs. The average frequency and severity are weighted averages.

10. ILLINOIS - RAILTEC Slide 10 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Severity vs. Frequency Analysis of Crude Oil Pipeline Incident Causes

11. ILLINOIS - RAILTEC Slide 11 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Incident Cause Probability

12. ILLINOIS - RAILTEC Slide 12 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Incorporating Multivariate Linear Regression Multivariate linear regression was utilized in order to develop an equation relating total cost with population density and barrels lost. For each incident cause, the total cost (not including the cost of the lost crude oil) was regressed against population density and barrels lost. The cost of lost crude oil was omitted from the regression (it was later added to the barrels lost coefficient) in order to avoid creating an overconfidence in the fit.

13. ILLINOIS - RAILTEC Slide 13 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Multivariate Linear Regression: R 2

14. ILLINOIS - RAILTEC Slide 14 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Multivariate Linear Regression: Coefficient p- values

15. ILLINOIS - RAILTEC Slide 15 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Limitations of Data Analysis Only two regressor variables were used (barrels lost and population density). Variables not accounted for include: time of day, environmental sensitivity, depth to groundwater, emergency services response time, etc. Low values of R 2. Limited sample size for each incident cause.

16. ILLINOIS - RAILTEC Slide 16 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Acknowledgements The development of this presentation is partially supported by the following organizations. The opinions expressed here do not necessarily represent the views of these organizations. National University Rail (NURail) Center U.S. DOT RITA University Transportation Center

17. ILLINOIS - RAILTEC Slide 17 © 2013 University of Illinois at Urbana-Champaign. All Rights Reserved Copyright Restrictions and Disclaimer Presentation Author Jeff LaHucik Undergraduate Research Assistant Rail Transportation & Engineering Center Civil & Environmental Engineering Department University of Illinois at Urbana-Champaign B118 Newmark Civil Engineering Lab, MC-250 205 N. Mathews Ave. Urbana, IL 61801 It is the author’s intention that the information contained in this file be used for non-commercial, educational purposes with as few restrictions as possible. However, there are some necessary constraints on its use as described below. Copyright Restrictions and Disclaimer: The materials used in this file have come from a variety of sources and have been assembled here for personal use by the author for educational purposes. The copyright for some of the images and graphics used in this presentation may be held by others. Users may not change or delete any author attribution, copyright notice, trademark or other legend. Users of this material may not further reproduce this material without permission from the copyright owner. It is the responsibility of the user to obtain such permissions as necessary. You may not, without prior consent from the copyright owner, modify, copy, publish, display, transmit, adapt or in any way exploit the content of this file. Additional restrictions may apply to specific images or graphics as indicated herein. The contents of this file are provided on an "as is" basis and without warranties of any kind, either express or implied. The author makes no warranties or representations, including any warranties of title, noninfringement of copyright or other rights, nor does the author make any warranties or representation regarding the correctness, accuracy or reliability of the content or other material in the file.