One of the main concerns in the aquifers adjacent to oil facilities is the leakage of LNAPLs. Since remediation processes costly and time consuming, so the first step in these systems is determining design goals. Often the most important goal of these systems is to maxi More
One of the main concerns in the aquifers adjacent to oil facilities is the leakage of LNAPLs. Since remediation processes costly and time consuming, so the first step in these systems is determining design goals. Often the most important goal of these systems is to maximize pollutant removal and minimize the cost. Identifying the thickness of LNAPL and its fluctuations can determine the type of recovery method and thus can be effective on the amount of removal and the cost of the implementation. In this study, three methods of gene expression programming (GEP), adaptive neuro-fuzzy inference system (ANFIS) and multivariate linear regression (MLR) were used to estimate and predict the LNAPL level. Input variables are groundwater level elevation and discharge rate of LNAPL and the output variable is the LNAPL level elevation. The results of the three models were analyzed by statistical parameters and it was determined that GEP technique has better results and could be used successfully in predicting LNAPL level fluctuations in recovery processes. Also, the GEP model provides an equation for predicting the LNAPL level that can be used in the field to predict the elevation of the LNAPL level.
Manuscript profile
