The implementation of well testing models for dynamic characterization of Naturally Fractured Vuggy Reservoirs (NFVR) commonly involves inverse modeling. Thus, this work introduces new results about dynamic description of NFVR by using a triple porosity-dual permeability model. The values of the inter-porosity flow parameters, storativity ratios, permeability ratio between fractures and vugs, wellbore storage coefficient, skin and total permeability, are automatically identified in Laplace Space by using a gradient Newton-type global optimization method named The Tunneling. This method offers more reliable results in comparison to the local optimization methods commonly used in parameters estimations. Applications: The precision and computational efficiency of the Tunneling Method are demonstrated through the reliable results obtained from the automatic analysis of synthetic examples, where a very comprehensive and large set of synthetic noisy data was used. In all cases, the Tunneling Method obtained optimal solutions. Likewise, the analysis of several field well test data cases recorded in NFVR are shown to confirm the benefits above mentioned. Results, Observations, and Conclusions: The parameters determination in the problem here addressed is non-unique due to the ill-posedness problem forcing the local optimization methods to stop in non-optimal solutions. Likewise, local methods require many realizations until converge into an optimal solution. Thus, the superiority of the Tunneling Method for overcoming these challenges is demonstrated through the reliable analysis of several synthetic and field data cases. Performing the optimization in the Laplace space makes the procedure very efficient in computing time. To moderate or eliminate the non-uniqueness problem, information about storativity ratios obtained from core or well log analysis, can be used through the optimization process. Significance of Subject Matter: Currently, the well test data analysis is the only available procedure to estimate the dynamic parameters which characterize NFVRs. Likewise, the approach introduced in this work allows obtaining more reliable estimations of these parameters than local optimization methods.

This content is only available via PDF.
You can access this article if you purchase or spend a download.