Comparing Sensitivity Analyses: Different Results from Deterministic and Probabilistic Risk Assessment. M. E. Hensle and P. B. Ryan, Emory University, Atlanta, GA; and T. W. Simon, EPA, Atlanta, GA
Sensitivity analyses of deterministic and probabilistic risk assessments were completed on Volatile Organic Compound (VOCs) concentrations obtained from the Twinns Inn field investigation study. The sensitivity analyses were developed to identify the ranking of exposure factors with respect to risk estimates. A shower model was created to simulate inhalation and ingestion of the VOCs. Deterministic and probabilistic assessments were applied to the data. The deterministic sensitivity rankings varied when produced using both Reasonable Maximum Exposure and Central Tendency Exposure estimates. The exposure factors that varied between the measures accounted for variation in the sensitivity rankings. The probabilistic sensitivity rankings were inconsistent when produced using Relative Partial Sum of Squares and Spearman Rank Correlation Coefficients. Differences in how these two measures deal with outliers account for the variations in the sensitivity rankings. The ability to vary the exposure factor truncation bounds in probabilistic Monte Carlo analyses makes comparisons of the sensitivity analyses between risk assessments difficult. However, the ability to set truncation bounds with the aid of site information allows the probabilistic risk assessment to provide a clearer picture of risk than that developed using the deterministic risk assessment. The ease of ability to manipulate the exposure factor truncation bounds can result in inappropriate risk estimates. Therefore, deterministic risk assessments need to be produced in conjunction with probabilistic risk assessments.
This paper was developed by the authors in their private capacities.
Go to . . .
1999 SRA Table of Contents
1999 SRA Author Index
Main Abstracts Menu Page
RiskWorld Home Page