Summary Wireline formation testers (WFTs) collect fluid samples for pressure-volume-temperature analysis through a probe set against the borehole wall. Filtrate contamination is reduced prior to sampling by either pumping the mixture of filtrate and reservoir fluid from the formation to the borehole or flowing the mixture into one or more WFT chambers. The cleanup is monitored at the surface. The time to reach the level of acceptable contamination (LAC) depends on the depth of invasion, pumpout rate, and various fluid and rock properties. Generalized guidelines predict time to first oil based on simple volumetrics but do not predict the rate of cleanup. Excessive cleanup time increases costs and the risk of differential sticking of the tool/cable. In some cases it may not be practical to attempt the operation as the LAC may take too long to achieve. A numerical simulator was used to investigate the characteristics of the contamination level versus time curve and to define the variables governing cleanup. The model was validated using data from five wells from two fields with differing rock and fluid properties. One hundred and fifty simulation runs were made with different invasion depths, flow rates, and rock and fluid properties. An equation was developed for field use that estimates filtrate contamination fw as a function of cleanout time t. An alternative approach for WFT sampling is also suggested, using not one but two probes. With this method, both cleanup time and the final level of filtrate contamination can be substantially reduced. Introduction Formation fluid is drawn through the probe and pumped into the wellbore for periods ranging from a few minutes to several hours or more. In this way, formations that were previously too contaminated with mud filtrate to yield useful samples can now be sampled by pumping out fluid until the level of contamination has dropped to an acceptable level. This can provide samples for pressure-volume-temperature analysis from a number of zones before the well is completed. Drill stem or production tests whose main objective was to collect formation fluid samples may no longer be required as a result. The contamination level is monitored continuously until it has reached the level of acceptable contamination (LAC), at which time a set of samples can be taken. The results of a study of near wellbore fluid flow during wireline formation tester (WFT) cleanup prior to sampling are presented. The objective was to predict cleanup time. In this way, it becomes possible to ascertain whether an operation is feasible, in terms of the time to reach LAC, the likely time/cost of the operation, and the consequential associated risks of stuck tools. The proportion of filtrate fw in the formation fluid pumped can be approximated by a function of the following form:1 f w = 1 − • [ 1 − ( t 0 / t ) n ] , where t0 and n are functions of radius of invasion, flow rate, and rock and fluid properties. Geometry of Flow The invaded zone can be visualized as a cylinder, with a hollow center representing the wellbore, and its outer wall representing the end of the invaded zone. Zero fluid movement through the mudcake is assumed. As filtrate is removed from around the probe, it is replaced by fluid in an elliptical flow regime, the geometry of which is dependent upon rock permeability anisotropy. The outer wall of the cylinder, which is the filtrate/oil interface, is also drawn in by the shrinking filtrate body toward the probe leading to a filtrate saturation distribution around the wellbore similar to the shape of an hourglass. The simulator results confirm this. Fig. 1 shows the radial section adjacent to the probe and Fig. 2 shows the section opposite the probe. Water filtrate (cooler colors) can be seen feeding in from above and below, while oil forms a cone directed toward the probe. Both figures show the development of the hourglass filtrate saturation distribution centered upon the probe. Assumptions This study is limited to sampling in vertical wells set in horizontal beds. The rock is homogeneous, and anisotropic. The mud is water based, and the filtrate is brine. The filtrate and the oil are immiscible, and the degree of contamination during cleanup is represented by the watercut. The rock is water wet. Corey relative permeability curves were used in the simulation runs used to test the effect of various parameters on cleanup time. The appropriate field relative permeability data were used when validating the model against the modular formation dynamics tester (MDT)*2 data sets. Appendix A describes the model setup. A Representative Invaded Zone The invaded zone was simulated by injecting water for a duration and rate that gave a radius of invasion obtained from resistivity logs. Pseudoised relative permeability curves were used to create a shock front. Watercut Prediction Five sets of MDT field data were used to validate the model. The optical fluid analyzer (OFA)*2 monitors the relative proportions of oil and water in the flowline with an accuracy of 5%. The observed water fraction during cleanout was compared to that predicted from the simulator. The watercut vs. time can be approximated to a particular function of time. Two points on the watercut curve have been selected as reference points: t0 the time when first oil starts to flow into the probe, and t1 the time when the watercut drops to 10%. Figs. 3 through 8 compare the observed water cut from the OFA data to the simulator results. Matching the Model with Field Data Wells X1, X3, and Y1 are from a field characterized by heavy viscous oil and high permeability rock. The field relative permeability data and fractional flow curves were used to generate the pseudoized relative permeability curves input to the simulations.

No abstract

A North Oman Field producing from two stacked Cretaceous reservoirs characterized by variation in inter-particle porosity along with variable vuggular and fractured secondary porosity system was studied. The objective was to build a reliable DPDP reservoir static model with scarcely available key data. An interdisciplinary approach utilizing available data, supplemented with analogs was used to implement a hierarchically linked reservoir characterization and modeling workflow for the purpose dynamic flow simulation studies. In the absence of core data, the NMR T2 distribution and derived permeability scaled to well tests mobility were correlated with borehole image features in a key well to define a rock typing scheme. The saturation height function was developed directly from the Sw and resistivity logs, by transforming and adjusting NMR T2 distribution to saturation height. In wells with only conventional logs, the SHF was used to back-calculate permeability within the transition zone. Electrical image logs in horizontal wells were used to build a high-resolution layering framework extrapolated inter wells to model highly conductive features (vugs and fractures). To address a relationship between secondary porosity selectively seen in thin dense layers, a BHI-based layering along horizontal wells was used to build the reservoir stratigraphic correlation to capture vertical flow barriers and high permeability vuggy layers. This approach used textural characteristics of rocks together with production data to capture mechanical stratigraphic boundaries and enabled fracture density estimation per mechanical layer. Use of hierarchical modeling workflow enabled the use of available BHI based rock texture, VCL from computed logs and acoustic impedance from inverted 3D seismic data to build 3D probability cubes of "mud-supported" and "grain-supported" rock textures. Conditioned to those 3D textural trend models, some seismic attributes were used as a guide to stochastically model the distribution of rock fabric based on the Lucia classification and the related inter-granular porosity. Subsequently the 3D distribution of Lucia-based Permeability and SW properties were also developed. Based on the assumption that fractures are developed within the perturbed stress field caused by the activity of the main pre-existing faults, a geomechanically-based process NFP workflow enabled us to build reservoir-scale fracture models. This workflow integrated seismic scale faults, and the distribution of fracture geometry and density from the wells coming from BHI logs, together with seismic discontinuity planes extracted from frequency-based filtering of seismic structural attributes. The tectonic model boundary conditions were estimated using 1D geomechanical models and analog data from neighboring fields. NFP-workflow generated fracture drivers; together with other fracture parameters, estimated from analog fields, neighboring outcrops and open literature, which were used to build a 3D multi-scale hybrid fracture model of the reservoir. The DPDP static reservoir model allowed dynamic history matching of the field with only global parameter adjustments, thus validating the property distribution from this static model.

In many oil wells, production is commingled from several layers. In such environments, understanding the properties of individual layers is essential to reservoir surveillance and production optimization. The inflow properties that typically require measuring are productivity index (PI), water cut, and static reservoir pressure. These measurements have traditionally been taken with wireline-conveyed production logging tools (PLT); however, in many wells and operating environments, completion and logistic considerations make running these tools difficult or even impossible. In those instances, an alternative is required. This paper presents a new procedure based on multirate testing in combination with distributed temperature sensors (DTS), an electric submersible pump (ESP) fitted with a gauge in the commingled fluid stream, and conventional surface testing. The additional test rates provide sufficient equations to resolve all the unknowns, whereas the DTS provides essential information such as which layers are producing and which are taking fluid, as well as a mass flowrate tool for measuring the flow rate of each layer. The procedure requires varying well flow rates over a range sufficient to ensure that all layers are producing, which in many cases requires an ESP to provide sufficient drawdown to overcome crossflow as well as a variable-speed drive to establish the test rates. The theoretical basis for the protocol is described, and an example is detailed to demonstrate the validity and robustness of the method for determining inflow properties. Finally, theoretical and practical guidelines are provided to demonstrate how the test procedure is affected by fiber-optic resolution, fluid properties, and the geothermal gradient. For wells equipped with forms of artificial lift such as ESPs, beam pumps, and progressive cavity pumps (PCPs), running PLTs is often not possible because of the completion obstruction. In such cases, this new DTS-enabled procedure has the potential of becoming the PLT substitute of the future.

As an increasing percentage of the world’s production comes from mature fields, there is a growing need for production enhancement techniques that are both rapid and easy to use for the practicing production engineers. For mature waterfloods, the ln(WOR) versus Np plot enables rapid well screening on the basis of incremental recovery factor, where WOR is the producing Water Oil Ratio and Np is the cumulative oil production. Published in-depth information on application of this tool is sparse. Yet, this is often the only tool available to the production engineer for evaluating development options, where a history-matched simulation model has not been maintained. In this paper, the theoretical basis for the use of the ln(WOR) versus Np is reviewed and studied, and is used to arrive at practical guidelines for interpreting production data. Its applicability as a forecasting tool to single-layered and multilayered clastic, waterflooded reservoirs of varying heterogeneity is demonstrated. Numerical simulation models then predict the behaviour of this plot for a wide range of heterogeneities. Production data is then analysed to show the applications of the theory for multilayered reservoirs. The ln(WOR) versus Np plots are analysed, and the impact of various factors is observed. The authors also demonstrate that, where applicable, this plot is the preferred decline curve for the following reasons: –Ln(WOR) versus Np does not require any pressure data; only surface well test production history is required.–It can be assumed that the ln(WOR) versus Np function is an approximate function of the reservoir only, and is decoupled from the outflow and facility constraints. This is especially useful when comparing artificial lift and drawdown strategies.–It is a decline curve model that provides a forecast of water cut, which is indispensable on waterflood projects.