Groundwater and soil contamination resulted from light nonaqueous phase liquids (LNAPLs) spills and leakage in petroleum industry is currently one of the major environmental concerns in the North America. Numerous site remediation technologies, generally classified as ex-situ and in-situ remediation techniques, have been developed and implemented to clean up the contaminated sites in the last two decades. One of the problems associated with ex-situ remediation is the cost of operation. In recent years, in-situ techniques have acquired popularity. However, the selection process of the desired techniques needs a large amount of knowledge. Insufficient expertise in the process may result in large inflation of expenses. In this study, petroleum waste management experts and Artifical Intelligence (AI) researchers worked together to develop an expert system (ES) for the management of petroleum contaminated sites. Various AI techniques were used to construct a useful tool for site remediaiton decision-making. This paper presents the knowledge engineering processes of knowledge acquisition, conceptual design, and system implementation in the project. The case studies have indicated that the expert system can generate cost-effective remediation alternatives to assist decision-makers.
Automation of engineering selection is important for tbe petroleum industries in which decision for a desired remediation technology at a contaminated site is critical for ensuring safety of the environment and the public. A variety of remediation methods/technologies are available. However, different contaminated sites have different characteristics depending on pollutants' properties, hydrological conditions, and a variety of physical (e.g. mass transfer between different phases), chemical (e.g. oxidation and reduction), and biological processes (e.g. aerobic biodegradation). Thus, the methods selected for different sites vary significantly. The decision for a suitable method at a given site often requires expertise on both remediation technologies and site hydrological conditions (Sims, 1992).
In general, soil and groundwater remediation techniques can be divided into two classes depending on whether the pollutant is directly removed/degraded in-place or not, i.e. in-situ or ex-situ. One of the main problems associated with ex-situ remediation is its high operating cost for activities like soil excavation and groundwater pumping. In recent years, in-situ techniques have become popular. However, with in-situ remdiation methods, knowledge on processes and factors controlling the results is lacking, which translates to much inflated expenses. Several mathematical models have been proposed to furnish representations as close as possible to reality of the effects of widely known remediation techniques. Some quantitative models have also been proposed for coupling multiphase flow and transport in a porous medium, with consideration of various remediation strategies such as water pumping, vapor and air venting, and steam injection. All of these techniques rely on human intervention for removing the contaminant. These techniques are fast, but costly. Moreover, most of them are too complex and not easily comprehensible for managers and engineers in industries and government. Therefore, a new approach is needed for developing useful, cost..effective, and user friendiy systems which can be readily adopted by industry and/or government to support decision-making on site remediation techniques.