Use Data to Drive Rail Inspections

Use Data to Drive Rail Inspections

Copyright 2007, Traffic World, Inc.

Nearly 1.7 million carloads of hazardous materials are transported by rail in the United States each year. Although the industry''s record for transporting hazardous materials has been good, the catastrophic consequences that can arise from the release of hazardous materials from rail cars are a significant safety issue.

From 2003 through 2006, the railroads reported 145 rail incidents that involved hazardous materials, resulting in 19 fatalities and 423 injuries. Although these numbers, on their face, are not large, these incidents resulted in the evacuation of 17,384 people from their homes and businesses, caused at least $17 million in track damages and resulted in about $71 million in equipment damages.

The Federal Railroad Administration must continue implementing its safety initiatives since train accidents are still on the rise overall. The rail industry''s safety record has improved, but significant train accidents continue to occur and the train accident rate has not shown substantive improvement in recent years.

To illustrate, even though in 2005 the number of train accidents decreased by 4 percent and the rate of train accidents per million train-miles traveled decreased by 7 percent, the overall data for 1995 through 2005 show that train accidents increased by 31 percent and the rate of train accidents grew by 11 percent.

Since 1998, our audit results have repeatedly shown that the FRA would benefit from an inspection program that places substantially greater emphasis on data analysis to target its inspection resources - a proactive rather than reactive strategy. Such an approach would aid in identifying some of the most prevalent causes of train accidents and enable FRA to devise corrective measures.

Our ongoing analyses show that human factors and track problems were responsible for 72 percent of the train accidents that occurred from 1996 through 2005.

By using trend analysis to track predictive indicators in problem areas, FRA could identify potential safety "hot spots." A number of predictive indicators could yield beneficial preventive measures, including improperly lined switches and unusually high defect ratios resulting from safety inspections.

In February 2005, we reported that FRA''s inspection program was structured to function in a manner that was (a) discretionary with individual inspectors in regard to routine inspections, and (b) reactive in terms of how it conducted focused inspections.

Currently, 385 inspectors oversee the Nation''s vast network of train miles. It is critical that FRA''s limited inspection and enforcement resources be carefully targeted to address those safety problems most likely to result in accidents and injuries. In our February 2005 report, we recommended that targeting be achieved through systematic use of trend analysis, along with other data analysis tools, to examine key indicators of a railroad''s safety condition (for example, its accident rate, defect ratio and employee injury statistics). FRA would benefit from a data-driven inspection program that makes substantially greater use of objective analysis of empirical data and metrics to target its inspection and enforcement activities. This approach would enable FRA to better allocate its inspection resources and decide appropriate levels of enforcement.

Circumstances related to the Jan. 6, 2005, Norfolk Southern Railway accident in Graniteville, S.C., both illustrate and underscore the value of trend analysis. Within five days of the accident, FRA issued a safety advisory to all railroads concerning improperly lined switches stating, "An improperly lined switch invites disaster and can be easily avoided. ? All railroads need to adopt the safety measures outlined in this advisory."

Trend analysis of rail safety data identifies improperly lined switches as the second-leading cause of railroad accidents in general, and the leading cause of accidents resulting from human error. By analyzing its data, FRA will be able to conduct predictive analyses and identify early indicators of problems, such as improperly lined switches, and order corrective action before accidents occur.

Defect ratios are a key indicator of a railroad''s safety that should be used to better target inspections. Defect ratios should be a key factor in determining the number of inspections that railroads receive, but we still see a gap between defect ratios and average inspections. Our 2005 review looked at several rail safety metrics and found that one - safety enforcement data - showed that serious safety problems have long existed for all four major railroads.

For example, Union Pacific had the highest average number of train accidents (weighted per million train miles) of the four major railroads during calendar years 1998-2000 and 2001-2003. Yet, Union Pacific had been inspected proportionally less, ranking third in FRA inspections per million train miles in those periods.

We find it counterintuitive that the railroad with the most track miles and the worst accident rate and defect ratio would be inspected at a lower rate than two of the three other major railroads that had fewer miles and better rates. Trend analysis leading to the targeting of resources on high-risk areas is particularly critical, because FRA inspection resources are limited.

-- This commentary is excerpted from testimony Scovel submitted to the House Transportation and Infrastructure Committee Subcommittee on Railroads, Pipelines, and Hazardous Materials on Jan. 30, 2007.