Abstract:
Retrospective pre-remedial modeling was conducted at two active remediation sites to evaluate the effectiveness of the LNAPL Distribution and Recovery Model (LDRM) in estimating LNAPL recovery. The two research sites contain petroleum hydrocarbon contamination of shallow subsurface soils and groundwater, which has resulted in LNAPL discharge to surface water receptors and subsequent regulatory action. Vacuum-enhanced recovery, also known as multiphase extraction, has been implemented at each site to recover free-phase LNAPL (i.e., free-product) and to prevent further discharge to adjacent canals. In addition to minimal success in preventing LNAPL discharge, recoverable LNAPL volume at both sites was overestimated by an approximate order-of-magnitude. The objective of this thesis is to determine if a relatively simple analytical modeling software program, readily available at no cost, can predict accurate vacuum-enhanced LNAPL recovery volumes based on commonly available site data.
The LDRM is a semi-analytical model based on steady-state radial flow and is used to estimate LNAPL distribution and recovery for several hydraulic remediation technologies. Input parameters include maximum LNAPL thickness, fluid properties, soil capillary and petrophysical characteristics, and recovery well performance. Maximum LNAPL thickness is corrected by estimating LNAPL hydrogeologic condition using diagnostic gauge plots and hydrostratigraphs. Additional inputs are determined through site-specific data and estimated based on empirically derived values for matching fluids and soils. In order to account for potential error, an input value range was established for each LDRM parameter and designed to simulate maximum and minimum LNAPL recovery. The results of the maximum and minimum LDRM simulations were compared with actual LNAPL recovered at each research site. The results for each of the four simulations were within an error factor of five of the actual recovered LNAPL. The results indicate that the LDRM can be utilized to prevent order-of-magnitude errors in vacuum-enhanced LNAPL recovery volumes with commonly available site data.
