| Abstract |
Injection tracers are an important tool for geothermal reservoir engineers to monitor fluid return from injection to production wells and understand reservoir flow paths. Fast return times indicate a direct connection between injection and production sites, and field development strategies are modified to mitigate thermal breakthrough between the wells. Tracer data can also be very helpful in calibrating numerical models to provide greater confidence in model forecasts. Traditionally, tracers have been modelled as an additional mass component in the control volume method. However, this is problematic due to the discretized nature of reservoir models. Reservoir models often use grid blocks with greater than 100 m scale grid blocks and, when modelling tracers as additional mass components, numerical dispersion can strongly influence tracer simulation results. Numerical dispersion tends to smear out the tracer return; however, numerical dispersion will be dependent on grid block size, and it is not the same as actual hydrodynamic dispersion. Furthermore, modelling tracers as an additional mass component significantly increases simulation run times. Streamline modelling treats the tracer problem differently. In our work we model the tracer as an ensemble of virtual tracer particles which move through the reservoir along the velocity field established by the reservoir simulator. This allows the tracer to move as a sharply defined front, with the return curve shape reflecting different pathways through the reservoir and their relative importance. In this paper we describe the tracer particle methodology, present some validation models for the streamline approach, and compare the new approach with results from the control volume method. |