(continued from previous page)

Accordingly, the user must realize that some physical phenomena may be expected to be reasonably represented in the results. Others may be utterly unreliable as a result of modeling choices. For example:

*Liquids in the form of liquid fogs or mists with a toxic gas in solution are often involved in an accidental oil or gas releases.

*None of the current atmospheric dispersion models even attempts to deal with atmospheric dispersion of the liquids or gases in that context, although some efforts at including ordinary fogs (water vapor) exist, accounting for condensation and evaporation, and perhaps the dynamics of such fogs.

*Needless to say, the complexity added in the modeling of turbulent, 2-phase, multi-component flows is even less advanced than that for pure gases.

For the purposes of health hazard assessment, it is useful to observe that turbulent processes involve mixing at various scales, from large-scale eddies formed in the flow, to mixing on a molecular level in the neighborhood of the interfaces between the larger eddies. This picture of turbulent flow should make clear that there are variations in time at any point attributable to the motions of the eddies of various sizes, irrespective of whether or not the source flows are steady in time.

Any actual exposure of a person to H2S gas from an accidental release can be expected to vary in time. Some atmospheric dispersion models account for those time variations, and some suppress such motions in favor of obtaining time-averaged results, employing various simplified models for the transport phenomena attributable to the turbulent motions.

One of the points to be made in this context is that time-averaged data can conceal time variations in the exposure process which might be significant enough to kill a person, even though the average concentration level might be below acceptable levels, if those levels are too high. Some models do account for such time variations, but then average data at a particular point for the purposes of presenting the results. Time average results from any model tend to be more reliable than transient data, even for models which do account for the transients in the modeling process, though transient data may be more relevant to health hazard assessment.

Health Hazard Assessment

Any health hazard assessment will require a medical judgment, and a judgment based on atmospheric dispersion. Given the state of the art in both, the best alternative appears to be the selection of a fixed exposure limit for public health and the use of a pure gas dispersion model, both certified by professionals in their fields.

The selection of a fixed legal benchmark for long-term exposure to H2S gas in the atmosphere does not account for effects of single or cumulative short term exposures to the gas at higher concentrations. And it does not account at all for exposures to liquid solutions containing H2S.

That choice, then, ignores phenomena which are real and for which there are, at the moment, inadequate medical data upon which to base a public health standard. That choice is not only at odds with the physical nature of exposures of the public to H2S, it also requires a judgment about averaging for the mathematical models of the atmospheric dispersion of the gas, at least for those models which describe the actual turbulence in the flow more faithfully. As noted, time averages can be more reliable than time varying data, but there remains the question of averaging for purpose of health hazard assessment, and for the purposes of representation of the physics of the flow.

Recommendation:

Despite the obvious shortcomings it seems prudent to select a public health hazard assessment for which there is a measure of collective agreement among public health professionals, within the current state of the art. For H2S that is a fixed maximum exposure limit for the public, consistent with values recommended by a number of states. There is no such value established by the federal government, comparable with its OSHA standard for healthy male laborers.

Such a fixed limit can be and has been employed for legal purposes, leaving ample legal precedent for such a use. This choice is a matter of pragmatic and legal convenience and provides no guarantee of protection for all members of the public, though it should suffice to provide ample protection for many. The foregoing outlined the nature of health hazard assessment, depending as it does on both a professional public health judgment and on a professional engineering atmospheric dispersion model. While both are flawed, they represent the best the state of the art can offer. Both can be upgraded as new research results are obtained. Moreover, there is a precedent for the utilization of such a legal process in the siting of systems handling hazardous materials, such as the nuclear industry.