Proceedings, TOUGH2 Workshop '95, p. 181-186

Lawrence Berkeley Laboratory, Report LBL-37200, Berkeley, California, March 20-22, 1995
## Using simulation-optimization techniques

to improve multiphase aquifer remediation

*Stefan Finsterle and Karsten Pruess*

Lawrence Berkeley National Laboratory, Earth Sciences Division

University of California, Berkeley, CA 94720

**Abstract.**
The T2VOC computer model for simulating the transport of organic chemical contaminants in
non-isothermal multiphase systems has been coupled to the iTOUGH2 code which solves
parameter optimization problems. This allows one to use linear programming and simulated
annealing techniques to solve groundwater management problems, i.e. the optimization of
operations for multiphase aquifer remediation. A cost function has to be defined, containing
the actual and hypothetical expenses of a cleanup operation which depend - directly or
indirectly - on the state variables calculated by T2VOC. Subsequently, the code iteratively
determines a remediation strategy (e.g. pumping schedule) which minimizes, for instance,
pumping and energy costs, the time for cleanup, and residual contamination. We discuss an
illustrative sample problem to discuss potential applications of the code. The study shows
that the techniques developed for estimating model parameters can be successfully applied to
the solution of remediation management problems. The resulting optimum pumping scheme
depends, however, on the formulation of the remediation goals and the relative weighting
between individual terms of the cost function.