PHMSA Transportation Incident Database Reporting
Week of 3-29-24
Since January 13th, I have been trying to report on chemical transportation incident as reported to PHMSA on their Form 5800.1. Back in April I noted problems with the data sets I was using for my reporting, it did not seem that they were complete. In my reporting then, I noted that I was changing the elapsed time that I was using for selecting the week for which I would be reporting. But I still noted that I did not have solid data to support that selection. Since that time, I have been collecting the necessary data. This is a brief look at the data to date.
Background Information
In my earlier post I provided detailed background information on the FORM 5800.1, the Hazardous Materials Incident Report. The regulatory basis for the report is found at 49 CFR 171.16, Detailed hazardous materials incident reports. It requires anyone that is in the possession of a hazardous material at the time of a covered incident {see §171.16(a) for what constitutes a covered incident} is responsible for submitting the form within 30-days of the incident being discovered. This 30-day reporting period causes the majority of the problems that I ran across in my earlier reporting.
There is a more immediate incident reporting requirement at §171.15, Immediate notice of certain hazardous materials incidents. That section requires telephonic notification to the National Response Center (NRC) at 800–424–8802 within 12-hours of a covered incident. The definition of a covered incident {§171.15(b)} from this section is included as a subset of the §171.16(a) definition. As should be expected, the reporting data requirements are much less detailed for the telephonic reporting.
New Data Collection
Starting on April 5th, I began to collect data from the PHMSA database for the week of March 29th through April 4th. I accessed the database and downloaded that week’s data every Friday morning between 8:00 and 10:00 am EDT. The graph below shows the number of incidents per day of the week for each of the eight weeks that data was collected.
There is no change in any of the data between weeks 7 and 8. For the purposes of this analysis I am going to assume that no other incidents will be reported for this week. There is no real expectation that no other incidents will be reported, but the numbers should be minimal and inconsequential for my reporting on transportation hazmat incidents.
The graph below shows the same data expressed as a percentage of the ‘final’ number of incidents reported for each day. At the current reporting period being used in my blog (5 weeks) at least 94.1% of the daily incidents have been reported, or 97.4% of the total weekly incidents.
Amount Spilled
For the purposes of my reporting on chemical transportation incidents, one important aspect of the data reported is the amount spilled in each incident. I wanted to see if the delays in reporting were affected by the size of the spill. To do this I had to come up with a way of aggregating the size of the spills. I arbitrarily took common liquid shipping-container sizes {1-gal, 5-gal, drum (55-gal), IBC (intermediate bulk container or 330-gal), and bulk (>330-gal)} as my groupings.
Now there is a minor problem with this analysis. The database has a single column for the size of the spill, but it is modified by the adjacent column which designates wither the spill was a liquid or solid. For the purpose of this analysis, I ignored that designation and considered all spill amounts to be in gallons. I could do this because I am trying to gauge the broad measure of relative spill size, not statistically analyze the amounts spilled.
The table below shows the final number (Week #8 data) of releases in each category.
The graph below shows how the number of incidents by spill size changed over the 8-week period. The number of IBC sized spills was only 80% of the total reported number by week 5, but it was only off by one incident. All other spill sizes are within 94% of their final total at 5 weeks. The numbers are too low for the IBC data to come to any conclusions about the effect of spill size on reporting delays.
Container Size
Similarly, I looked at the reported size of the container involved in the incident. This data has similar problems with mixed measurement techniques and unit, so I used the same categorization technique that I discussed above. The table below shows the final numbers for the various size categories at Week 8. Note that one of the ‘Bulk’ data points is a 2500-lb plastic box which should more properly be categorized as an IBC not Bulk.
It is interesting that there were 235 instances where no container size was included. The instructions for completing the Form 5800.1 provide the following directions for this (block 27) information:
“Describe the package capacity and the quantity: Enter the total capacity of the inner and outer package. Also enter the actual amount of hazardous material that was shipped in the package, the number of packages in the shipment, and the number of packages that failed. Please include the units of measurement (liter, gallons, pounds, cubic feet, etc.)”
It looks like the entries of ‘0’ reflect block 27, ‘Package Capacity’, actually being left blank. Looking at the other data for these particular incidents, it would appear that, in most cases, the container sizes were probably gallon or smaller in internal packagings.
The graph below shows the incident reporting timelines for each of the container size categories. Again, the numbers represent a percentage of the final reported values.
For Week #5 only 66.67% of the bulk container incidents were reported. Since incidents involving bulk containers are potentially the most dangerous, it would seem that reporting on the PHMSA database five weeks out may be missing critical information. Of course, the data here is looking at just a single, random week. It looks like I am going to have to collect additional data…