You are here:
AUTOMATIC CALIBRATION OF A STOCHASTIC-LAGRANGIAN TRANSPORT MODEL (SLAM)
Citation:
Enfield*, C G. AND F. Espinoza. AUTOMATIC CALIBRATION OF A STOCHASTIC-LAGRANGIAN TRANSPORT MODEL (SLAM). Presented at AUTOMATIC CALIBRATION OF A STOCHASTIC-LAGRANGIAN TRANSPORT MODEL (SLAM), Cincinnati, OH, November 05, 2004.
Impact/Purpose:
To inform the public.
Description:
Numerical models are a useful tool in evaluating and designing NAPL remediation systems. Traditional constitutive finite difference and finite element models are complex and expensive to apply. For this reason, this paper presents the application of a simplified stochastic-Lagrangian model (SLAM) for advective dominated transport. The model is calibrated using tracers (partitioning and non-partitioning). For the application of this model, the hydraulic properties are defined as a random variable and the NAPL structure by four different configurations; uniform, lognormal, uniform correlated to the hydraulic properties and lognormal correlated to the hydraulic properties. The problem is formulated in two ways: coupled, evaluating hydraulic and NAPL properties simultaneously; and uncoupled, evaluating hydraulic properties in first place (with non-partitioning tracer data) and then the NAPL properties (with partitioning tracer data).The parameters of the model are evaluated by means of four variations of the simple genetic algorithm (SGA). Three of the algorithms solve the coupled formulation and the last one solves the uncoupled formulation. The model was applied to evaluate the properties for three real cases. The results show that the four algorithms solve the three problems with the same level of quality, but with different effort. The reductions in effort with respect to the first algorithm (SGAI) are 50%, 80% and 90%.