Even with the continual advances made in bothcomputational algorithms and computer hardwareused in reservoir modeling studies, large-scalesimulation of fluid and heat flow in heterogeneousreservoirs remains a challenge. The problemcommonly arises from intensive computationalrequirement for detailed modeling investigations ofreal-world reservoirs. This paper presents theapplication of a massive parallel-computing versionof the TOUGH2 code developed for performinglarge-scale field simulations. As an applicationexample, the parallelized TOUGH2 code is applied todevelop a three-dimensional unsaturated-zonenumerical model simulating flow of moisture, gas,and heat in the unsaturated zone of Yucca Mountain,Nevada, a potential repository for high-levelradioactive waste. The modeling approach employsrefined spatial discretization to represent theheterogeneous fractured tuffs of the system, usingmore than a million 3-D gridblocks. The problem oftwo-phase flow and heat transfer within the modeldomain leads to a total of 3,226,566 linear equationsto be solved per Newton iteration. The simulation isconducted on a Cray T3E-900, a distributed-memorymassively parallel computer. Simulation resultsindicate that the parallel computing technique, asimplemented in the TOUGH2 code, is very efficient.The reliability and accuracy of the model results havebeen demonstrated by comparing them to those ofsmall-scale (coarse-grid) models. These comparisonsshow that simulation results obtained with the refinedgrid provide more detailed predictions of the futureflow conditions at the site, aiding in the assessmentof proposed repository performance.
展开▼