Complex heterogeneous reservoirs penetrated by deviated wells pose significant challenges when modeling and interpreting packer-and probe-type formation-tester measurements. This paper introduces a streamline-based method specifically designed for near-wellbore fluid-flow modeling in vertical and deviated wells. The primary application of the method is for efficient simulation and automatic history matching of formation-tester measurements in the presence of invasion and variable petrophysical and geometrical properties. Based on the spatial distribution of pressure calculated with a previously validated near-wellbore finite-difference model (FDM), the method traces streamlines from the reservoir into fluid sampling probes. The calculated spatial distribution of pressure is input to the streamline-based method to propagate the spatial distribution of water saturation. Subsequently, the calculated spatial distribution of water saturation is input to the FDM to update the spatial distribution of pressure. Streamlines are then recalculated and this repeated pressure-saturation procedure continues until reaching the desired final simulation time.The combined FDM and streamline-based method utilize full-tensor permeability to couple wellbore-reservoir systems during fluid sampling in horizontal and deviated wells. It is shown that the spatial rendering of streamlines permits rapid assessment of the time evolution of fluid saturation near the sampling probe. We successfully test the streamline-based method in the dynamic simulation of fluid contamination into conventional and focused-type probes operating in vertical, horizontal, and deviated wells for the case of multiphase, immiscible fluid-flow in heterogeneous formations. The method implements successive streamline and FDM time steps adaptively to decrease computational time by a factor of four while achieving less than 5% relative difference in the simulation of fluid production measurements when compared to FDM results.For formation testing in deviated wells, the streamline-based method permits more rapid appraisal of anisotropy, bedboundary, and complex geometrical effects on fluid-sampling times than conventional 3D fluid-flow simulators. It also quantifies the sensitivity of measurements to variations of permeability, invaded zones, anisotropy, and well deviation. Our simulations indicate that fluid sampling in deviated wells subject to mud-filtrate contamination requires adequate positioning of the probe around the perimeter of the wellbore to secure the fasted cleanup possible.

Cotton Valley tight gas sands in the east Texas Basin of North America consist of very fine-grained, well-sorted quartz arenites and subarkoses that are overprinted by significant diagenetic processes. Stratigraphic variations in rock type control gas production. Hydraulic fracturing delivers economic gas production. Recently, horizontal wells over 4,000 ft long consisting of more than twelve hydraulic fracturing stages have been drilled. The good gas producing rock type reservoirs are usually less than 14 ft thick and exhibit strong diagenetic overprint. Low gas prices challenge the use of sophisticated geosteering logs. Using a very basic Gamma Ray (GR) log and very limited offset pilot well data, chasing such thin and variable sand bodies over a distance of 4,000 ft in a marginal marine sedimentary environment is daunting. Apart from this, horizontal wells that target thin layers present unique challenges to completion optimization. Quad-combo log data, including azimuthal density image data, was acquired using a Logging While Drilling (LWD) tool. Extensive log modelling was performed by combining the horizontal well logs with logs from its two offset vertical wells, and consistent interpretation was achieved. Log modelling has helped the post-drill well diagnosis in geosteering, completion design, and production performance. It has also supported formation evaluation. This paper will highlight an integrated study workflow in tight gas sands using open-hole and cased-hole data. It will demonstrate the geosteering challenges, explain the log modelling process, and display the formation evaluation results. Geomechanical study, together with hydraulic fracturing data and production log data will be used to confirm that good gas producing rock types are easier to fracture, and they contribute better to production.

The Wamsutter field is located in south-central Wyoming within the eastern part of the Greater Green River basin. The field covers 1,620 square miles and includes more than 8,500 wells; it produces approximately 450 million cubic ft of gas a day. The Almond formation, which accounts for approximately 85% of the produced gas, was deposited in shallow marine parasequences characterized by strandplain, deltaic, or barrier island complexes and their non-marine counterparts. The upper interval, which was produced extensively, contains the thickest, most laterally continuous and best rock quality. Most of the remaining field development potential remains in sections characterized by thinner and discontinuous facies. In addition to the larger stratigraphic uncertainty, poorer rock quality makes permeability interpretation in uncored wells quite challenging. We tackle reservoir uncertainty by taking gross depositional interpretation and superimposing gas effective permeability models that honor dynamic measurements. This paper describes our efforts to produce core-calibrated, log-derived, effective permeability models that match zonal contribution from production logging and initial production indices. The critical steps in the petrophysical workflow are related to the interpretation of the log-derived absolute permeability and core-derived gas relative permeability. Log-derived absolute permeability was addressed by evaluating alternative stochastic (neural network-based) and deterministic models on wireline logs. Core-derived gas relative permeability was tackled by modeling unsteady, pulse decay-type measurements along with capillary pressure tests by rock type. Alternative gas effective permeability models are integrated over perforated intervals (KegH) and compared to actual reservoir performance. The model exhibiting the best correlation with dynamic data is superimposed on gross depositional maps. We discuss the assumptions and limitations of the various measurements that make up petrophysical workflows in tight-gas sands as well as their contribution to the location of infill and step-out opportunities.

Summary Cotton Valley tight gas sands in the East Texas basin of North America consist of very-fine-grained, well-sorted quartz arenites and subarkoses that are overprinted by significant diagenetic processes. Stratigraphic variations in rock type control gas production. Hydraulic fracturing delivers economic gas production. We drilled horizontal wells more than 4,000 ft long consisting of more than 12 hydraulic-fracturing stages. The good gas-producing rock-type reservoirs are usually less than 14-ft thick and exhibit strong diagenetic overprint. Low gas prices challenge the use of sophisticated geosteering logs. By use of a very-basic gamma ray (GR) log and very-limited offset-pilot-well data, chasing such thin and variable sand bodies over a distance of 4,000 ft in a marginal marine sedimentary environment is daunting. Apart from this, horizontal wells that target thin layers present unique challenges to completion optimization. We acquired quad-combo log data, including azimuthal-density image data, by use of a logging-while-drilling (LWD) tool. We performed extensive log modeling by combining the horizontal well logs with logs from the two offset vertical wells, and achieved consistent interpretation. Log modeling has helped the post-drill well diagnosis in geosteering, completion design, and production performance. It has also supported formation evaluation. This study will highlight an integrated-study work flow in tight gas sands by use of openhole and cased-hole data. It will demonstrate the geosteering challenges, explain the log-modeling process, and display the formation-evaluation results. Geomechanical study, together with hydraulic-fracturing data and production-log data, will be used to confirm that good gas-producing rock types are easier to fracture, and they contribute better to production.

No abstract