Stochastic risk assessment of sites contaminated by hazardous wastes

Stochastic risk assessment models offer the potential for being more objective by explicitly considering variability. Such a model has been developed for a site contaminated with polychlorinated biphenyls (PCBs) by representing the parameters used in the risk assessment as probability distribution functions (pdf) rather than single values. The pdf for total risk calculated by the model is approximately lognormal, although the pdf of parameters in the model take on a variety of forms. A first-order approximation to the model provides good estimates for the high end of the distribution, which is of concern when conservative risk assessments are desirable. The first-order approximation provides good estimates even when the level of variation of the parameters is increased well above levels that are normally expected. A procedure was developed to apply the stochastic risk assessment model in a series of calculations to determine preliminary remediation goals for the site. In addition, a simplified technique was developed to calculate preliminary remediation goals using only results from simulating risk with initial site conditions.