Modeling free-product recovery systems for petroleum hydrocarbon liquids within a heterogeneous aquifer
The American Petroleum Institute (API) Publications Numbers 4682 (1999) and 4729 (2003) provide an overview of recovery technologies for petroleum hydrocarbon liquids that are released to the subsurface environment and accumulate near the water table. API 4729 is a supplement to API 4682, and it describes four (Excel spreadsheet) models that may be used to characterize the subsurface distribution and mobility of liquid hydrocarbons (light nonaqueous phase liquids, LNAPL). These models may also be used to calculate the potential recovery rate and time using single and dual pumping wells, as well as vacuum enhanced wells. However, these spreadsheet models are unsuitable for some specific field conditions. In this thesis project, those spreadsheet models were rewritten in Fortran language using Lahey Fortran 90/95 compiler. Soil heterogeneity is an issue that increases uncertainty in groundwater modeling. Heterogeneity is defined as horizontal stratification of homogeneous soil layers. One can simulate up to a three-layer aquifer with new model. To quantify the amount of recoverable LNAPL, a new term, recoverable LNAPL specific volume" was introduced to exclude the residuals from the total LNAPL recovery volume calculation. A model for a special case where there is a vertical hydraulic gradient (VHG) was included in the formulation. When a fine grain zone overlies a permeable stratum that is hydraulically connected to a discharge zone with lower hydraulic head, a strong downward vertical hydraulic gradient may occur within the fine-grain zone. A possible scenario in this environment is that there can be little of no LNAPL in the fine grain zone under equilibrium conditions, with LNAPL accumulation beneath this FGZ.. Technologies for removal of liquid- and vapor-phase LNAPL were reviewed and formulated within the model. For vapor-phase volatile organic compounds (VOCs), soil vapor extraction (SVE) can be used effectively with other clean up technologies to complete restoration of contaminated sites. The model predictions for a specific site condition were compared with the field data. The result showed that the model predictions for the recovery volume and the recovery rate of LNAPL fit well with the field data.