Significant Changes in Chlorine Dispersion Models

As I mentioned on Thursday the Chlorine Institute has published an updated version (Edition 6) of their Pamphlet 74. That pamphlet used to be titled “Guidance on Complying with EPA Requirements Under the Clean Air Act by Estimating the Area Affected by a Chlorine Release”. The new version removes reference to the EPA and is more simply called the “Guidance on Estimating the Area Affected by a Chlorine Release”. This reflects the more widespread use of the modeling data.

I do not have a copy of Edition 5 of this pamphlet, but I do have Edition 4 that I have used for a reference in a number of blog posts on chlorine releases that I have done over the years. So I am going to be comparing the downwind hazard predictions from these two versions of the Pamphlet.

Background

Chlorine is a very common hazardous material and it is widely used in this country. It is certainly the largest volume toxic inhalation hazard (TIH) chemical shipped in the United States. It is typically shipped in the following types of containers:

∙ 150-lb. Cylinder;

∙ 1-Ton Container;

∙ 17-Ton Tank Truck Trailer; and

∙ 90-Ton Rail Tank Car

Chlorine is used as a disinfectant for water treatment and waste water treatment facilities. It is also used in the manufacture of a large number of industrial chemicals and pharmaceuticals. It is generally produced via the electrical disassociation of common salt, NaCl; producing both chlorine (Cl₂) and caustic soda (NaOH). A less common, but still important, source of chlorine is as a byproduct of the production of magnesium from natural ores.

Chlorine gas was used as the first modern chemical weapon in World War I. The CDC reports that the immediately dangerous to life and health (IDLH) concentration of chlorine gas is 10 ppm while reporting that lethal effects have been seen at 34 to 51 ppm when exposed for over an hour. The OSHA personal exposure limit is 1 ppm with a short term exposure limit of 3 ppm.

The Chlorine Institute has been using the values of 3 ppm and 20 ppm as their standards for downwind exposure reporting in Pamphlet 74. In a release scenario exposure levels at less than 3 ppm are considered to be relatively safe in the short term. Exposures between 3 and 20 ppm will have some immediate medical effects that will probably require medical attention, but will probably still allow an otherwise uninjured person to self-evacuate from the affected area. Exposures over 20 ppm will almost certainly cause immediate health effects of a severity that would stop a person from self-evacuating from the affected area and may result in death.

Worst Case Scenarios

The table below describes the worst case scenario reporting from both versions of Pamphlet 74 and the RMP*Comp tool used to report downwind distance of concern for both the EPA Risk Management Program (RMP) and the DHS Chemical Facility Anti-Terrorism Standards (CFATS) program. All distances have been converted to feet.

	Pamphlet 74 – Ed 6		Pamphlet 74 – Ed 4		RMP*Comp
	20 ppm	3 ppm	20 ppm	3 ppm	.0087 ppm
150 lbs	108	210	3,168	8,976	4224
1 ton	331	988	11,616	35,376	15840
17 ton	1,152	1,857	41,712	121,440	58080
90 ton	1,184	1,765	78,144	219,920	>132000

This obviously marks a serious departure from the previous guidance on downwind hazards from chlorine gas releases. According to both publications (pgs 16-17 in both Editions) the following conditions of the release are the same in both studies:

• The ambient relative humidity is 50%

• The ambient temperature is 77°F (25°C)

• Liquid chlorine is at 77°F and 100 psig before the release

• Surface roughness is 3.94 inches (0.1 meters). Such a surface roughness corresponds to a relatively flat, grassy, rural setting. A release in hilly terrain, forested area, and urban environment or over water may have significantly different terrain results.

• The wind speed is 3.36 miles/hour (1.5 meters/sec). The reference height for measuring the wind speed is 32.8 feet (10 meters).

• The contents of the container are released instantaneously and evaporated at a constant rate over a ten-minute period

• The atmospheric stability is class F (night, < 50% cloud cover)

• The release occurs at ground level

• Solar radiation is assumed at 0 Btu/hr/ft2

• Averaging time is 10 minutes

• Receptor height is 0 feet (ground level)

Modeling Differences

What is clearly different is that the two Editions are reporting results from different dispersion models. Edition 4 uses TRACE 8.0 developed by Safer Systems, Inc. Edition 6 uses Hazard Prediction and Analysis Capability (HPAC) 5.0 developed by DOD’s Defense Threat Reduction Agency (DTRA). There is a brief discussion about the model used in each report as the first appendix. What is missing, however, is any discussion in the latest edition about why the model changes were made and how that impacted the huge change in predictions reported.

What is mentioned in the press release announcing the latest version of the Pamphlet is that it “incorporates information obtained from the DHS ‘Jack Rabbit I’ chlorine release field tests”. I think that it would be safe to assume that TSA did not run tests meeting exactly the conditions specified in the report. I have not seen the final report on the Jack Rabbit I trails (and I suspect it is classified and thus I will never see it), but with only a total of ten releases split between chlorine and anhydrous ammonia, I would expect that the researchers would have varied the environmental situation as much as possible to obtain the type of results necessary to develop a robust model.

Without a robust discussion about the differences in the two models used to report the data in these two different editions of Pamphlet 74, it is hard to make a judgement as to which is the more appropriate model upon which to base regulatory and emergency response decisions. And make no mistake, with the wide discrepancy between the two sets of predictions provided in the two versions of the Chlorine Institute report, it is absolutely assured that that the manufacturers and users of chlorine will press for changes in safety and security regulations to reduce their costs of operations. And just as surely they will be vocally and vigorously opposed by any number of environmental activists.

Modeling Problems

I have discussed the modeling problems associated with chlorine dispersion modeling in a couple of earlier blogs. Back in March of 2010 I reported on the exposure estimates from Edition 4 of Pamphlet 74; saying:

“Looking at the charts on pages 24 and 25 of Pamphlet 74 it looks like anyone inadequately protected in the cloud for up to a couple of miles away from the catastrophic release from a full railcar is at serious risk of being killed by the cloud. Inadequate protection in the cloud at distances of up to 15 miles from the release could have very serious medical consequences.”

In my blog post on what we now know as the start of the Jack Rabbit Project I noted that:

“It seems that the chlorine cloud from the two most recent catastrophic releases [Graniteville, SC and Macdona, TX] of chlorine from rail cars did not come anywhere near following the dispersion model for the release. TSA is initiating a two phase study of chlorine gas dispersion to clarify these discrepancies.”

Based upon the observed chlorine dispersions seen in these two accidents, it would seem obvious that the model used in the 4^th Edition of Pamphlet 74 is more than a little exaggerative in it prediction of downwind areas of concern. And that, of course, was the reason for initiating the Jack Rabbit project in the first place. It would be interesting to see if anyone has gone back and used the observed data from those two real-world incidents to see how well the new model used in Edition 6 fits the observed conditions.

Of course, I would be very surprised if any reasonable model could accurately predict the dispersion cloud in a real incident. There are just too many variables that would affect both the macro and micro dispersion effects of the cloud. I discussed some of these in an earlier blog post.

The Use of the Dispersion Model

Practically speaking we do not need for an absolutely accurate prediction of gas cloud dispersions patterns. What we need is something that will allow facility safety personnel or emergency response personnel (depending on where the release takes place) to develop an initial plan for responding to a catastrophic chlorine release. With that in mind it would seem to me that we would want a plan that errors on the side of a larger downwind hazard area rather than one that underestimates the area.

From the new model, if it is appropriately supported by the Jack Rabbit data (including the upcoming Jack Rabbit II later this year), we have been grossly overestimating the potential exposure to chlorine releases from industrial facilities and transportation incidents. If that is true we may have overburdened producers and users of chlorine in the protective measures that we have required them to take.

Having said that, the relatively small downwind hazard areas described in the latest version of Pamphlet 74 feel like they are too small. This may lead us to reduce the protective burden to the point where we put people needlessly at risk near chlorine facility storage areas. Before we take that risk it would be prudent to ensure that there was a solid peer review of the model construction and methodology.