Collisions and derailments

The National Transportation Safety Board (NTSB) has found much evidence to indicate the probable cause of the January 6, 2005, collision and derailment of Norfolk Southern train 192 in Graniteville, South Carolina, was the failure of the crew of Norfolk Southern train P22 to return the main line switch to the normal position after the crew completed work at an industry track. Contributing to the severity of the accident was the puncture of the ninth car in the train, a tank car containing chlorine, which resulted in the release of poison gas. 

Because of changes in indexing of the dollar threshold for reporting accidents, consistent data on derailments and collisions are available only since 1975. The number of collisions and derailments was 50% higher in 1979 than in 1975. However, by the mid-1990s, the munber of collisions and derailments had decreased to about 40% of the number in 1975. Much of this improvement occurred in period between 1980 and 1985. 

Update of the data used as the basis of the analysis that resulting in a reduction in the number of train-to-train collisions, buffer stop collisions and derailments predicted per year. 

These actions are broken down into elementary operations that must be undertaken safely , to avoid collisions and derailments. The signal box allows an operator to perform all these operations. 

There was consequently agitation for some government intervention, not least from the labor unions whose members jobs were under threat as railroads attempted to improve their financial situation. The unions argued, for example, that the increase in collisions and derailments was linked to the removal of firemen from diesel locomotives in the mid-1960s. (See the academic paper by Fisher and Kraft (1971) whose econometric argument, in the fullness of time, appears to be misleading.)... 

This chapter provides a summary of the major issues so far. It identifies the hazards posed by railroads, assesses the casualty rates, looks at trends in those rates, makes comparisons with comparable hazards in other industries or elsewhere in society, and reflects on how people react to the hazards. The five major railroad hazards considered are (in no particular order) fatalities to highway users at grade crossings trespasser fatalities, fatalities to train crews in collisions and derailments and during coupling operations occupational injuries to maintenance employees and releases of hazardous materials. 

Most collisions and derailments are of little concern to the general public as they involve freight trains and occur in sidings and on yard track. However, there seems to be a great fear of those small number of collisions and derailments that result in the release of hazardous materials. These materials can affect the communities surrounding the accident site due to contamination of ground water, explosion or release of poisonous gases. 

The remainder of the book deals with operational safety. That is the prevention of collisions and derailments. This chapter discusses how much safety should be provided. It also introduces five possible market failures which may result in railroads providing non-optimal levels of safety. Subsequent chapters will investigate whether railroads are susceptible to these failures, and the possible public policy responses to the failures. 

The c(x) function is composed of two sub-parts. The first is the cost of purchasing the safety inputs. A cost fiinction for safety inputs can be obtained by duality theory from the safety production fiinction. The second is the cost of destruction to railroad property and injury to railroad employees when an accident occurs. Under FELA, railroads are strictly liable to employees who are injured in accidents resulting from violations of federal safety rules, which will be the case in most collisions and derailments. 

Another source of published data is the annual FRA Accident / Incident Bulletin which provides individual information for the largest forty-four railroads. The data on collisions and derailments per million train miles in 1996 are shown in table 13.2. As can be seen, collision and derailment rates do vary markedly between the large. Class I, railroads. Shippers could draw some conclusions from these data, although they would have to build in allowances for differences in the nature of operation of individual railroads, such as the amount of switching, that affect accident rates. 

It is more difficult, but not impossible as explained in a later chapter, to draw meaningful comparisons between the accident rates of the smaller Class II freight railroads. These twenty-four railroads average eight-and-one-half collisions and derailments each a year, and accident rates will vary markedly from year to year due to statistical fluctuations explained by the Poisson distribution. For the small. Class III, railroads meaningful statistical inference is impossible, even if the data were published individually by railroad by the FRA. The 275 different corporate entities average only 0.7 collisions and derailments each a year. 

Table 13.2 Collisions and Derailments per Million Train Miles 1996...

The most notable difference is in the rate of collisions and derailments. The rate of collisions is similar to that of class II railroads, but twice that of the Class I railroads. Derailments occur almost four times as frequently as on Class I railroads and over twice as frequently as on Class II railroads. Of course part of the explanation is that the smaller railroads are engaged in switching operations, which is highly susceptible to collisions and derailments in comparison with line-haul operations. Another explanation is poor track condition, occasioned by years of low investment prior to the sale by the large railroads. Fortunately, the consequences are mitigated by low operating speeds. 

There are also genuine concerns that some small new railroads may be myopic due to inexperience. Albeit, that there is little empirical evidence that they pose a serious safety threat. Small railroads represent 3.2 percent of national train miles, and account for 3.8 percent of total railroad fatalities. While they do have higher rates of collisions and derailments, these are not translated into higher fatality rates. Low speeds of operation mitigate the consequences of many accidents. While some individual small railroads might give cause for concern, it is likely that the inexperience of new short-line railroads would be far down the priority list of railroad safety problems that need to be attended to. 

How important is the harm to bystanders as compared with harm borne by the railroad, its customers and employees An estimate of the annual social costs of collisions and derailments is made in table 16.1. The costs have been classified... 

Dennis estimate of 31 million will underestimate the total property damage to bystanders, but it should not be a substantial underestimate. Based on these figures, the harm suffered by bystanders amounts to about eight percent of the social cost of collisions and derailments. 

Collisions and derailments, especially those involving hazardous materials usually require the attendance of emergency services. At the very least, ambulances and the fire department will attend. If an evacuation is necessary, police will be called to help in the evacuation and provide security for the affected area. Assistance of the state police or the national guard may be necessary in major incidents. Local schools may have to be opened to provide temporary accommodations for those displaced, and the Red Cross may be called to provide bedding and food service for the residents and emergency workers. 

An obvious response to the problem of uninformed customers making incorrect decisions is to mount a public-information campaign. In general, all the public needs to know are historical data on the accident performance, or safety outputy of individual railroads. Much of this information is already available. The FRA s Accident / Incident Bulletin for a calendar year is available with a delay of about nine months. Midyear data are available with a delay of a couple of month. The information is available in hardcopy and on the FRA s World Wide Web site. An interested customer can quite quickly obtain information on the rate of collisions and derailments for the largest Class I and II railroads, and can observe recent trends in these rates for individual railroads. 

A quarter of a century later, it is possible to assess effectiveness by looking at the time-series of the rate of collisions and derailments due to track defects and of variables that might affect this rate. These data series are shown in figure 19.1 in the form of an index with the value in 1975 set equal to 100. The accident rate per track mile is shown by the line with the squares. The rate increased by two-thirds between 1975 and 1978 and then began to decline and is now only a quarter of the level in the peak year. Some of the reduction in track-caused collisions and... 

Figure 19.1 Analysis of Track-Caused Collisions and Derailments...

Collisions and derailments due to equipment failures Coupling, uncoupling and handbrake operations Getting on or off locomotives or cars While on a locomotive... 

In addition to the prevention of injuries, the locomotive and car standards are supposed to prevent serious collisions and derailments caused by defective equipment and inadequate braking. The effectiveness of these regulations can be investigated by a time-series analysis of data relating to collisions and derailments due to equipment defects since 1981. A graph of relevant data is shown in figure 19.2, with all data shown as an index with the value in 1981 set equal to 100. The collision and derailment rate per train mile is shown as the line with the squares. The rate has declined continuously and is now a third of what it was in 1981. Some... 

Measuring different risks Separate measures should be developed for each of the major risks associated with railroading. It would seem sensible to separate out the risks of collisions and derailments from those of en >loyee injuries or gradecrossing accidents or trespasser fatalities. Each of these different types of risk have different causal factors and demand different responses. 

Of course, almost any level of observed number of occurrences is possible, and is consistent with the mean given the inherent variability in the Poisson process. However, the further the observed number is from the mean, the less likely it is to occur. For example, consider a railroad that averages 100 collisions and derailments a year Equation (20.1) determines that this railroad will have more than 107 accidents one year in every four, will have more than 117 accidents one year in every twenty, and more than 120 accidents one year in every forty. 

Collisions and Derailments Employee Fatalities and Injuries Trespasser Fatalities Grade Crossing Accidents... 

---