OPTIMAL SOIL VENTING DESIGN USING BAYESIAN DECISION-ANALYSIS

Remediation of hydrocarbon-contaminated sites can be costly and the de sign process becomes complex in the presence of parameter uncertainty. Classical decision theory related to remediation design requires the parameter uncertainties to be stipulated in terms of statistical estim ates based on site observations. In the absence of detailed data on pa rameter uncertainty, classical decision theory provides little contrib ution in designing a risk-based optimal design strategy. Bayesian deci sion theory, however, allows the use of subjective judgments of the de signer together with parameter uncertainty in the decision-making proc ess. In the present work, Bayesian theory is used in developing the th eoretical framework for risk-based design analysis for soil gas ventin g operations. Two utility functions, exponential and quadratic, were u sed in the analysis, identifying the cost, and risk associated with un der- or over-achieving the site stipulated remediation targets. Prepos terior analysis provided the information on estimated optimal venting time, optimal effective how rate and sample size for both utility func tions, and the applicability of each function. Hypothetical field-scal e simulations using normal and weathered gasoline were performed to de monstrate the applicability of Bayesian decision theory based analysis for practical scale problems.