Estimation of parameters of dual porosity reservoirs from well testing signals using adaptive neuro-fuzzy inference systems (ANFIS)

Dual porosity system refers to those porous media composed of two different structures i.e. fracture and matrix. These porous media is characterized by two special parametersnamely interporosity flow coefficient (λ), and storativity ratio (ω). Determination of these parameters is required for understanding of current and future performance of naturallyfractured reservoirs. Pressure transient signals (well testing) is a reliable tool for determination of parameters of dual porosity reservoirs. Indeed, it’s possible to estimate these parameters from pressure derivative (PD) graphs using an appropriate intelligent technique. In this study adaptive neuro-fuzzy inference systems (ANFIS) is used for estimation of λ and ω from PD graphs. Required pressure transient signals for dual porosity reservoirs have been generated using commercial PanSystem software, and then converted to the PD graphs. Theobtained PD graphs were entered to the ANFIS model by 26 successive points. It should be noted that an appropriate optimization algorithm and number of cluster are determinedby a trial and error procedure. Using this procedure, the ANFIS model with 9 and 10 clusters are found as the best model for prediction of storativity ratio and interporosity flow coefficient, respectively. Moreover, the hybrid optimization algorithm shows better performance during training of these ANFIS models. The proposed ANFIS models determined ω and λ of 312 PD signals with the average absolute relative deviation (AARD%) of 14.69 and 36.28, respectively.

Estimation of parameters of dual porosity reservoirs from well testing signals using adaptive neuro-fuzzy inference systems (ANFIS) Keywords:

Dual porosity , storativity ratio , interporosity flow coefficient , Pressure derivative , ANFIS