Analytic-Numeric Hybrid Modeling Approach for Interpreting Transient Tests in Horizontal Wells

Abstract

Uncertainties arise while interpreting transient tests in horizontal wells. This study presents a hybrid approach in that both analytic and numeric models are used to circumvent most of the problems associated with the conventional analytic interpretation. This novel approach entails multiple steps. Initially, reservoir parameters are estimated by identifying as many flow regimes as possible using the analytic method. Thereafter, a numeric simulation model is built using a coupled wellbore-reservoir simulator. Herein, we seek to match the wellbore flow profile data, collected by production logs (PL), by adjusting local permeability and skin. This PL-matched model then allows rate calculation during the transient-test periods, using measured bottomhole pressure as the inner-boundary condition. Finally, the computed rate is convolved with pressure to interpret the drawdown data for seeking reconciliation with buildup interpretation. We present two field examples to illustrate the application of the notion presented here. Complications arise while interpreting flow and buildup tests using conventional tools, because of complex well trajectories penetrating this heterogeneous carbonate reservoir. Results show that reservoir anisotropy (kv/kh) is a very sensitive parameter and is therefore well determined. The history-matched model was quite successful in predicting the water cut behavior over 18 months in one of the wells. Introduction Interpretation of transient tests in horizontal wells is fraught with uncertainties. Complexities of long wellbore trajectory, reservoir heterogeneity, wellbore storage, variable skin along the borehole, among others, contribute to the problems associated with conventional interpretation. In fact, most test responses rarely conform to those of the idealized analytical models. Both the literature and our own experiences support this notion. Consequently, many are hesitant to collect expensive data of dubious value. Despite advancement of analytic models1–7of various formulations, interpretation of transient pressure data in horizontal wells has been a daunting task. The interpretation problem is rooted in the lack of development of various flow patterns, expected for a given geometry. Experiences suggest that often only one flow regime can be identified with certainty, thereby rendering the system indeterminate. Let us explore some of the reasons for this difficulty before offering any solution. Insights into flow dynamics in long horizontal sections perhaps hold the key to our understanding of the complex system. In-situ measurements8,9have identified some features that arise from reservoir heterogeneity and well geometry. These conditions along the well length include, among others, non-uniform reservoir pressure, variable skin, and non-uniform flux precipitated by local water sumps and gas traps in an undulating well trajectory. Frictional and accelerational pressure drops along the borehole may add another dimension to this complex problem. Occurrence of any or all of these conditions inhibits development of recognizable diagnostic features. Consequently, the use of conventional analytic methods can be limited in many cases. Indeed, gathering of a string of nonanalyzable test data may prompt some not to conduct transient tests. Therefore, questions surface, what, if anything, can be done to alleviate the problem of having to deal with data of seemingly limited value. In our proposed approach, the use of PL data is required for interpreting flow and shut-in tests. We note that the PL data have furnished the contributing borehole length in conventional test interpretation10–12using analytic models. In our approach, we match the entire wellbore influx with the coupled simulator, thereby providing a more detailed picture of reservoir heterogeneity.

Equivalent Skin Factors for Nonuniformly Damaged Horizontal and Multilateral Wells

“…Several groups of researchers have investigated the impact of formation damage on horizontal well performance by using numerical models or commercial simulators. [21][22][23][24][25][26][27][28] In the numerical models, two different methods are used to quantify the skin factor; direct and indirect methods. In the direct method [21][22][23][24][25] , formation damage is generally mimicked by assigning a reduced permeability to the grid blocks surrounding the wellbore.…”

“…[21][22][23][24][25][26][27][28] Most recently, a numerical model to characterize the formation damage due to use of water-based-mud and natural cleanup during production has been proposed. [26][27][28] The numerical model presented by Ding and co-workers [26][27][28] is capable of predicting the horizontal well performance under the influence of unevenly distributed formation damage.…”

Lessons learned from Wireline Formation Testing with Single and Multiple Probes

“…Different PI values only change the probe pressure value (bottomhole flowing pressure) but not the calculated cell pressure value as discussed by Su 12 . Several researchers [13][14][15] have reported successful numerical modeling of transient pressure tests.…”