The suggested values of accident rates and conditional release probabilities for use in the chlorine transport risk analysis will be analyzed and specified.
3.5.1 Accident Rates
In a well-defined, focused transport risk assessment, especially one that will result in quantification of both frequency and consequence, it is important to use accident rates that are:
1. Representative of the actual carrier performance expected for the transport activity being analyzed.
2. Appropriate for the failure scenarios under consideration.
In this thesis, the QRA will focus only on failure of the tank trailer shell or heads due to puncture or impacts associated with a transport accident. (Other scenarios, such as valve shearing during an accident, have already been estimated to be minimal contributors to the overall risk.)
Therefore, for the purposes of this thesis, the accident rate of interest is one that may involve sufficient forces to puncture, tear, or otherwise fail the tank. Accidents such as “fender benders,” that involve mild contact between automobiles or scooters with the chlorine transport unit tractor or trailer, are unlikely to generate such forces, and therefore are not included in the accident rate calculation.
The U.S. DOT counts as “recordable” accidents those that involve a) a fatality, b) an injury requiring immediate treatment away from the scene, and c) damage sufficient to require one or more of the involved vehicles to be towed away from the scene. The latter part of this definition is useful for risk analysis, because damage that is severe enough to require a tow-away of a vehicle may be assumed to have involved more forces than in a simple fender-bender. It is important to note that damage caused to an automobile involved in an accident with a heavy truck may still not involve forces sufficient to puncture a cargo tank shell. However, for the purposes of the current analysis, we will adopt the U.S. DOT definition of a recordable accident, with an adjustment for local conditions,
as discussed below.
3.5.2 Adjustment of Taiwan Car Accident Rates
A major difference between the roadway conditions in the U.S. and Taiwan is the presence of a great number of scooters on the Taiwan roads. A common occurrence in Taiwan is for a scooter rider, in an attempt to move more quickly through traffic, to cut around and in front of other vehicles in an unsafe manner. As a result, thousands of accidents involving scooters occur every year, resulting in hundreds of scooter operator and passenger fatalities. The forces in such accidents may be devastating to the relatively unprotected scooter riders, but are highly unlikely to cause any damage to a chlorine tank truck. Therefore, fatalities to scooter operators and passengers resulting from an accident with a chlorine tank truck, is not included in the accident rates developed for use in the current study. Refer to Table 3.5 and Table 3.6 for Company C’s accident rates in Taiwan.
Table 3.5 Company C chlorine fleet recordable accident Rate, using the definition of the U.S. DOT without scooter accidents
Year Trips Recordable Accident (chlorine fleet of Company C)
Mileage
(km) Accident Rate
1995 420 0 307,440 0.000
1996 580 0 424,560 0.000
1997 693 0 507,276 0.000
1998 634 0 464,088 0.000
1999 737 0 539,484 0.000
2000 836 1 611,952 1.634
2001 867 0 634,644 0.000
2002 889 0 650,748 0.000
2003 996 0 729,072 0.000
Total 6,652 1 4,869,264 0.205
Year Recordable Accident (entire
fleet of Company C) Mileage / km Accident Rate
1995 1 4,326,088 0.231
Table 3.6 Company C entire fleet recordable accident rate
3.5.3 Accident Rate Development
In this thesis, there is only one carrier, Company C Transport, involved in the transport of Chlorine from Kaohsiung to Company A. Two accident rates are to be derived: one representative of their performance in 1995 (the “base” case), and one representative of their performance currently, after the implementation of several safety initiatives (the “mitigated” case). Upper and lower boundaries are also provided.
From 1995 through 2003, the Company C chlorine fleet drivers logged a total of 4,869,264 kilometers driven round-trip, see Table 3.5.3.1. The chlorine fleet kilometers driven have increased annually from 307 thousand in 1995 to 729 thousand in 2003. Only 1 recordable accident occurred during this 9 year period (in 2000), for an overall rate of .205 accidents per million kilometers.
Since 1995, Company C has continually implemented safety programs and initiatives, that should have both a qualitative and quantitative effect on risk reduction. However, the presence of zero numerators makes it very difficult to identify any trends that have occurred during this 9 year period.
Looking at the entire Company C fleet (including the chlorine tank truck operation), a total of 50,941,045 kilometers were logged from 1995 through 2003. The entire fleet kilometers driven have increased annually from 4.3 million in 1995 to 6.5 million in 2003. Eleven recordable accidents occurred during this 9 year period, for an overall rate of 0.216 accidents per million kilometers. This rate is very similar to that experienced just by the chlorine fleet. It is reasonable to assume that the
performance of the chlorine fleet drivers would be equivalent to or better than the entire fleet, since they receive even more training and have enhanced provisions concerning work hours. In order to increase the sample size, however, we will use data for the entire Company C fleet as the basis for our accident frequency analysis, recognizing that it may be conservative.
The accident rate from 1995 through 1998 for the entire fleet was approximately 2 times higher than the rate from 1999 through 2003 (0.308 vs. 0.159). This transition period roughly coincides with the implementation of additional safety programs and measures by Company C (which due to gradual phase in and enhancements cannot be pinpointed exactly), and may also reflect an overall improvement in the Taiwan freeway system in recent years.
Accordingly, for the “base” case rate, we will use 0.308 accidents per million kilometers.
1. In order to estimate the effects of uncertainty, analyses should be run using this value as well as high and low estimates of ±50% (0.462 and 0.154). This is an arbitrary but reasonable approach since adequate data to develop actual confidence ranges are not available.
2. For the “mitigated” case rate, which accounts for the progressive programs that Company C has put in place, we will use 0.159 accidents per million kilometers.
3. As with the “base” case, analyses should be run using this rate as well as high and low estimates of ±50% (0.239 and 0.078). The lower figure may also better represent the true performance of the chlorine fleet drivers, given their advanced training and improved operating conditions.
3.5.4 Variation by Road Type and Segment
In most cases, including the current one, it is extremely difficult to generate actual accident rates experienced by a carrier along certain segments of road or even along certain types of roads.
3.5.5 Expressway vs. Local Roads
In general, local roads are more congested than expressways and therefore a higher accident rate may be expected. However, the lower speeds likely lead to a lower probability of release when an accident occurs, because the impact forces generated in the accident are usually not great. In the current study, since the focus is on failure of the tank trailer shell or heads due to puncture or impacts associated with a transport accident, only expressway corridors will be examined. Again, generally speaking, this is where higher speeds (and therefore higher impact forces) may be attained.
The accident rates provided in the previous section will be assumed to apply to expressway movements, although some of the accidents that contributed to the rate development may have occurred on local roads. This assumption will result in a more conservative, but still reasonable estimate.
3.5.6 Variation by Freeway Number and Along Freeway Segments
Ideally, the accident rates developed earlier could be adjusted to account for expected variation along different freeways or segments of those freeways. For this thesis, a corridor rather than a full route approach is being used. Table 3.7 shows the data for Expressway Number 1 from the National Police Administration for 2003.
Accident No. Accident Rate
Table 3.7 Accident numbers and rates of three different road segments
The definitions for A1 (fatality within 24 hours), A2 (injury), and A3 (property damage) do not correspond directly with the definition we are using for the current analysis. However, it can be seen that the A3 accident rate in the Taichung corridor may be slightly less than in the Kaohsiung or Chungli corridors. While we could adjust our accident rate to account for an expected difference, the rate variation is not large enough to justify this added detail. (Also, the variation among different accident outcomes is not consistent within corridors, which further complicates any adjustment.)
Perhaps of greater interest for evaluation of route alternatives would be accident rate differences between the various National Expressways that might be utilized. Since Expressway Number 3 has only recently been in operation, it may require some time before adequate data are collected.
Development of a table similar to above might also be valuable in evaluating potential differences between accident rates on the “base” and “mitigated” case corridors (Changhwa and Hsinchu).