Fracture network modeling conditioned to pressure transient and tracer test dynamic data

Abstract

We previously showed that the validation of fracture network models generated using discrete fracture network (DFN) approach through multi-well pressure transient data had certain limitations. In this paper, we tested other approaches used for modeling fracture networks: single porosity (SP) and dual porosity/permeability (DK), which are more common models for dynamic simulation than the DFN models. Multi-well pressure transient and tracer tests were used to validate the fracture network models generated as DFN and converted to grid based SP and DK models using static data. A sensitivity analysis was conducted to clarify the effects of fracture permeability, tracer dispersion, matrix–fracture interaction and other terms on tracer performance. Finally, a comparative analysis was performed to critically assess the reliability of different fracture network representations. The data obtained from the Midale field, Canada, was used in this exercise.Even though single or multi-well pressure transient tests are frequently used as an input for fracture network (or permeability map) generation and for the hydraulic calibration of discrete fracture network models, such tests provide limited information about the topology of fracture networks, i.e. connectivity and orientation. Tracer testing was suggested as a richer source of information about reservoir heterogeneity, especially in capturing fracture trends and orientation. The DK model was capable of satisfactorily modeling the pressure interference test in the validation process. However, it did not demonstrate the same accuracy for the tracer test simulation even though only the DK model was able to provide some meaningful match to the tracer tests. The major disagreement between the real and simulated tracer test response was the network connectivity. The limitations of this modeling approach were discussed in terms of representing the connectivity of the fracture network accurately.