Multiphase Simulation Model for Validating the Estimate of Light Non-Aqueous Phase Liquids (LNAPL) Transmissivity Using Bail-Down Test

Abstract

Bail-down testing is commonly used to determine the LNAPL transmissivity, volume and initial recovery rate for LNAPL contamination in soil. In this study, a non-compositional three-phase LNAPL/water/air bail-down test model is simulated on the multiphase simulator $$\hbox {DuM}u^x$$ (DUNE for Multiphase, Component, Scale, Physics flow and transport in porous media). Field observation data validate the model. Then, different scenarios were simulated to compare the estimated transmissivity using the modified Bouwer and Rice (Huntley in Groundwater 38:46–52, 2000) approach for different soil properties and fluid viscosity. The results show that (1) estimates are reliable for subsurface layers of variable permeability ranges and yield accurate results for low-permeable soil; (2) estimates are sensitive to the Van Genuchten $$\alpha $$ . Results show poor estimation using low $$\alpha $$ values. Employing the modified Bouwer and Rice approach for bail-down test interpretation at sites where the capillary fringe is thick is not recommended; (3) estimates are sensitive to the initial LNAPL thickness $$h_{n}$$ . Results show that interpretation is not possible if the initial thickness of the contamination is below 1 m for selected parameters; (4) estimates were flawed for low-viscosity fluid and improved as the viscosity of fluid increased; and (5) for scenarios where contamination exists in multilayer soil, the modified Bouwer and Rice approach yields reliable estimates. However, this study tested only one arrangement of this type, wherein the bottom layer consisted of coarse soil.