(575c) 3-D Compositional Streamline Reservoir Simulator – a Parallel Implementation

A large number of grid blocks are required to capture the viscous/gravity fingers and accurately predict the breakthrough time and oil production rate in field-scale compositional simulations. Excessive numerical dispersion, grid orientation effects and small time step size (limited by the CFL condition) limit the use of finite-difference reservoir simulators for field-scale compositional simulations. To attain high computational speed, a streamline-based compositional simulator is developed with the ability to take large time steps. 3-D transport equations are decoupled into series of 1-D equations along the streamlines and compositions are updated using a 1-D solver in terms of time of flight. Compositional streamline module is integrated with an existing finite-difference simulator where the pressure solution and velocity field are obtained.

In this work a parallelized version of this compositional streamline simulator is developed to run large scale simulations using multiple processors. Finite-difference calculations including pressure solution are done on a single processor. Streamline module is parallelized by distributing the streamlines among different processors allowing parallel streamline solution. Computations along any streamline are independent of other streamlines and no communication is required. The parallel code is written for the streamline part of the simulator which includes mapping from irregular streamline grid to regular grid, flux calculations, 1-D solver, mappings from regular to original streamline grid and mappings to underlying finite difference grid. Flux calculation is computationally expensive primarily due to flash calculations that are performed to distribute components in the hydrocarbon phases. Simultaneous solution of this time consuming step results in reduction of cpu time.

The parallel streamline simulator is validated for many benchmark problems. Results are in good agreement with the sequential streamline simulator and other commercial finite-difference simulators. Simulations are conducted by increasing the number of grid blocks, number of components, and streamlines and its effect on the cpu time is studied. This parallel compositional streamline simulator is also tested for scalability.