Estimation of reservoir properties using an integration of the capacitance-resistance model and tracer testing

Several tools and techniques exist to understand distributions of reservoir properties, which is one of the keys to successful reservoir management. The capacitance-resistance model (CRM) is an analytical tool to estimate connectivity between producer-injector pairs from historical rates and (when available) pressure data in waterfloods. Tracer testing is another common method to obtain reservoir information from the amount of tracer produced. Because the CRM is a physically based, simple input-output model, its combination with tracer testing can provide a variety of reservoir information. Nonetheless, the current CRM is limited to immiscible displacements, which do not have dissipative effects as do tracers. The objectives of this study are to extend the CRM to tracer testing and use this approach to quantitatively define reservoir characteristics.
To enable the CRM application to tracer flow, we incorporated a combination of two tracer models, based on miscible displacement theory, into the CRM. The tracer models allow us to calculate and match produced tracer concentration by using nonlinear regression; as a result, reservoir properties are estimated. This study used three tracer models: (1) a dispersion-only model developed from solutions to the convection-diffusion equation, (2) a Koval model developed from the Koval theory (1963), which predicts the performance of a miscible displacement by using a single factor and (3) combinations of the two by using dynamic upscaling technique. To incorporate the tracer models into the CRM, two methods, serial fitting and simultaneous fitting, were used. Both methods were applied to tracer data from 10 injectors and 10 producers of the Lawrence field. According to the results, interwell connectivity obtained from the CRM is in good agreement with the observed peak tracer concentrations. Going beyond peak concentrations, all tracer models are able to give a good fit in most of the cases. The reservoir properties estimated by each tracer model (drainage pore volume, dispersion coefficient, and Koval factor) were compared and analyzed. Results suggest that the combined model, which provides the best fit, can better represent a tracer flow than the other two models alone. We also found that the simultaneous fitting method gives the best fit to total producer rate data and tracer data. Simultaneous fitting mitigates the non-uniqueness of the fits, leading to an improvement of tracer matching. Consequently, the simultaneous fitting is an appropriate method providing results that are more accurate. An application of CRM to tracer testing serves as an alternative method to analyze and interpret tracer data. Furthermore, when both are available the synergy between CRM and tracer analysis provides insight into reservoir features. In addition to interwell connectivity and drainage volume, integration of the CRM and tracer analysis can estimate dispersion and Koval factors, which cannot be obtained from each method alone.