查看更多>>摘要：Data-driven surrogate models,which are trained by samples to replace time-consuming numerical simulations,have been widely used to solve production optimization problems in recent years.It is a challenging and meaningful subject to research advanced surrogate-model-based methods that can obtain superior optimization performance within a limited time budget.The key is to enhance the quality of each training sample,i.e.,the contribution of each sample to the overall optimization performance improvement,because the acquisition of each sample requires a numerical simulation that generally costs tens of minutes or even several hours.To obtain samples with enhanced quality,a novel approach named surrogate-reformulation-assisted multitasking knowledge transfer(SRAMT)for production optimization is proposed in this research.Multiple surrogate models,which can imitate the landscape of the initial production optimization problem,are constructed with diverse samples as reformulations of the target problem.These models reflect different landscape information of the same problem and thus can be regarded as multiple associated optimization instances,which just provide a solid foundation for the subsequent process.Then,an advanced optimization method,namely multitasking optimization(MTO),is leveraged to find optimal solutions for these surrogates.MTO can handle several optimization instances simultaneously and boost the performance of each one by transferring useful knowledge among tasks.Besides,in the absence of prior knowledge about the target production optimization task,as in most situations,an approach is proposed to determine the frequency of knowledge transfer adaptively based on the similarity between surrogates to improve efficiency and stability.To verify the effect,four 100-dimensional benchmark functions and two reservoir models are tested on the method and the results are in comparison with tiiose obtained by differential evolution(DE)algorithm and three other surrogate-model-based methods.The results show that the proposed method can achieve optimal well controls which can get the highest net present value(NPV)for target production optimization problems and have superior convergence speed.
查看更多>>摘要：Simulation of near-well flow around a single well in reservoirs is normally studied using radial logarithmic grids.A promising method for more complicated,multi-well cases is a combined approach of using a Cartesian and a radial grid.In this paper,a new method of overlapping grids is introduced in which a radial grid is overlapped on a coarse Cartesian grid in the area around a well.Connectivity between the two grids is established using an appropriate boundary condition.Five different algorithms of grids connectivity with internal boundary conditions are presented in which the data are transferred between the grids by interpolation pressure and saturations in each grid.This method is implemented in the open-source Matlab Reservoir Simulation Toolbox(MRST)using the modified black-oil model.The proposed approach is validated by simulation of a reference fine radial grid model on a gas-condensate reservoir.Comparing the results with the reference model indicates that the proposed Double-Scale method using a 21 x 21 coarse Cartesian grid is as accurate as a 1001 x 1001 simple Cartesian fine grid model.Then,a double-layer numerical problem with two production wells was simulated and sensitivity analysis was conducted to investigate the effect of critical condensate saturation and vertical permeability.Finally,the second SPE comparative solution project was simulated to compare the performance of the proposed method with the use of the perpendicular bisector(PEBI)grids around the well.Results showed that the accuracy of the PEBI grid and Double-Scale method are comparable.However,the proposed Double-Scale method is almost 2 times faster in this specific problem.
查看更多>>摘要：In this paper we present a novel approach for detecting early signs for the stuck events in drilling using Deep Learning.Specifically,we adapt neural network based unsupervised learning tool called Autoencodel for anticipating the'stuck'events during the drilling process.We build Autoencoders on Recurrent Neural Networks(RNNs)to model the normal drilling activity,thereby detecting the stuck incidents as anomaloul activity.We conduct experiments on the actual drilling data collected from 30 field wells operated by multiplJ drilling sources with diverse well profiles and demonstrate that our approach obtains promising results for thJ stuck sign detection.Furthermore towards explaining the trained model's prediction,we present reconstruction analysis on the individual drilling parameters.
查看更多>>摘要：Natural gas plays a crucial role in sustaining the economic development.The production behavior of natural gas reservoir is significantly affected by the connecting aquifer.To investigate the aquifer and reservoir properties,the pressure buildup well test is conducted and the pressure transient data need to be analyzed.However,the pressure transient data analysis usually involves a procedure to build an analytical or numerical model to predict tue pressure transient dynamics.With the existence of many historical records on the well test data and advancement of machine learning method,a data-driven method is proposed here to automate the well test data analysis of water-drive gas reservoir.The dataset generated from the previous well test records is used as the training set for the machine learning method.To identify the water invasion mode,the random forest classification method is introduced,and the discrete linear segment slopes are extracted from the pressure derivative curve as the feature.To predict the pressure transient dynamics,the random forest regression method is proposed to construct a projection between the aquifer/reservoir properties and the pressure transient curves.Once the water invasion mode is determined and the data-driven model to predict the pressure transient dynamic is constructed,the ensemble Kalman filter is used to estimate the aquifer/reservoir properties from the well testing data.The results show that the random forest classification method can accurately identify the water invasion mode and the random forest regression based ensemble Kalman filter method can estimate the aquifer/reservoir properties accurately with the reduced uncertainty.
查看更多>>摘要：Accurately evaluating fracture conductivity is key to optimizing the shale fracturing process.However,the influence of fracture morphology on conductivity remains unknown due to the non-replicability of rough samples in most laboratory-based studies.Thus,accurately duplicating rock samples with identical surface morphology and mechanical properties in the laboratory is a major concern for future fracture conductivity studies.Herein,a new method of reproducing rock samples for conductivity testing was proposed.Essentially,a laser scanner was used to obtain split rough fracture surface three-dimensional point cloud data for reverse reconstruction,which were subsequently combined with 3D engraving technology.This method successfully produced a significani number of conductivity test samples with uniform surface morphology and similar mechanical properties that can be used in experimental research on shale fracture conductivity.Comparison of the split shale sample with that created using the proposed method showed an area tortuosity error of only ±0.35%,a comparable elevation distribution frequency,and a high degree of similarity between the surfaces.Furthermore,due to the influence of the fracture surface roughness,rock samples with a proppant concentration below a certain"critical proppant concentration"may exhibit higher conductivity than samples with higher proppant concentrations under low closure pressure.Moreover,the"critical proppant concentration"was positively correlated with large-scale waviness,but did not show obvious correlation with small-scale unevenness.
查看更多>>摘要：Oil and gas exploration activities often face great challenges due to the nonlinear behavior of the reservoir's physical properties,which is commonly defined as"heterogeneity".Currently,well log data analysis techniques are a novel approach to unravel such nonlinear behavior,because well log data incorporates considerable geological information that determines reservoir property.However,the current complexity analysis techniques face two challenges:1)the spatiotemporal multiscale nature of complex geological systems and 2)the superposition of the trends in the geophysical well log data on the analysis results.To fill the research gap,we propose an empirical mode decomposition-refined composite multiscale dispersion entropy analysis(EMD-RCMDEA)to eradicate trends and obtain the complexity results with spatiotemporal characteristics.The proposed method produces more accurate results,and its effectiveness,stability,and efficiency are also verified by the simulation signals and the gamma-ray(GR)well log signals.Compared to previous refined composite multiscale entropy analysis(RCMSEA),the EMD-RCMDEA enhances the stability by 69.3% and efficiency by 53.5%.Additionally,using the GR well log data for reservoirs,this method is also applied to explore the heterogeneity of strata with diverse depositional environments and different composite patterns and acquire the following results.1)The EMD-RCMDEA values of the GR well log data are positively correlated with the heterogeneity of the strata.2)The reservoir developed in a delta-front depositional environment has the strongest heterogeneity.3)The heterogeneity of the composite patterns is much stronger than that of the single heterogeneity patterns.4)Among the heterogeneities of the composite patterns,the pattern consisting of different fades is stronger than that for single facies.
查看更多>>摘要：The binomial deliverability equation is one of the essential parameters that control the fluid production/injection rate from wells.The presence of caves and fractures introduces many difficulties related to the determination of the binomial deliverability equation of fractured caved carbonate reservoirs.In this work,a novel method for determining the binomial deliverability equation of fractured caved carbonate reservoirs is proposed for the first time.First,the detailed procedure of our method is illustrated.More specifically,the superposition principle is used to derive the expressions of(Ap/Q).The pressure buildup data and type curve matching are employed to induce the dimensionless bottom-hole pressure solution.Sequentially,the binomial deliverability equation can be obtained through the least square method.Then,we employ our method to a simple case and compare the calculation results with the literature data to calibrate the accuracy of our method.Next,we present the applications of our method to two example wells from the Shunbei Oilfield,western China.The results show that our method is reliable.Moreover,the results also indicate that our method is more suitable for fractured caved carbonate reservoirs than the deliverability test.Meanwhile,through the calculations of massive field data and theoretical analysis,we conclude three constraints on the input parameters.They are as follows:(i)the flow rate of each production system must be incremental;(ii)the production time must satisfy tp1 = tP2-tp1 = tp3-tP2 = tp4-tp3;and(iii)the flow rate must satisfy Q2 = 2Qi,Q3 = 3Q1,Q4 = 4Q1.Finally,we prove the necessity of these constraints.
查看更多>>摘要：The fractured-vuggy carbonate reservoir possesses the characteristics of a highly discrete distribution of fractures and vugs,a wide scale range,and the coexistence of free flow and porous flow.However,it also exhibits a high degree of filling media in vugs,including chemical sedimentation,carried fillings and collapsed fillings,which poses challenges for studying the mechanism of enhanced oil recovery(EOR)in filled fracture-vug reservoirs.In this paper,3D printing technology is utilized to construct heterogeneous,filled models.With a multiphase flow visualization platform,displacement experiments of different injection-production methods,including water injection,gas injection and the co-injection of water and gas,are conducted to elucidate the mechanisms of EOR.It is found that the lower is the gas injection rate or the higher is the water injection rate,the higher is the recovery factor for a horizontal filled vug.For a vertical filled vug,gravity exerts a greater influence on the gas swept area,but inertial forces dominate fluid flow when the water injection log Ca exceeds-2.68.In addition,in a displacement experiment of a simultaneous injection of gas and water,the gas-water mixing zone appears until log Ca ≥-4.38(gas)and log Ca ≥-2.68(water).Then,the synergistic gas-water effect displaces the oil phase in the filled medium with a higher recovery rate.Therefore,the displacement of gas and water co-injection is preferred over sequential injections,and the injection velocity of gas and water should be compatible.
查看更多>>摘要：Countercurrent spontaneous imbibition(SI)is an important flow mechanism to recover oil in water-wet porous media through capillarity.Experimental imbibition tests conducted in core samples show that oil recovery as a function of time occurs in a characteristic S-shaped curve,which describes the infinite-acting and the boundary-dominated regime.Although several dimensionless time groups have been proposed to model countercurrent SI in porous media,they fail to properly scale the results,causing inaccurate estimates of oil recovery.In this work,we present a new approach,using a hybrid solution,to model countercurrent SI for oil-water systems dominated by capillary forces.The infinite-acting regime is modeled by an early-time solution using a new dimensionless time group,which enables the correct scaling of imbibition results onto a universal single curve.To model the boundary-dominated regime,we introduce a late-time solution derived as a function of a characteristic distribution of saturation to estimate the flow behavior after the imbibing fluid reaches the no-flow boundary.The novelty of the model is that it enables the accurate estimation of fluid imbibition under the boundary-dominated regime,a flow condition critical to evaluate the true potential of countercurrent SI driven by capillarity for oil recovery in porous media.We verified the model against numerical simulations under a wide range of flow conditions relevant in water-oil systems,and used experimental data reported in the literature to validate our model.The solution presented provides an accurate approach to model countercurrent SI,which could be extended for dynamic conditions and the modeling of the flow of fluids in fractured media.
查看更多>>摘要：Low-salinity waterflooding is considered to be an advanced oil recovery method,but the main mechanism behind the increase in oil recovery factor is still under investigation.Fines migration has been recognized to be one of the mechanisms responsible for the increase in oil production.In this case,the increase in oil recovery happens as a result of increased sweep efficiency and consequent reduction of residual oil.The release of particles due to reduction of electrostatic forces results in flux diversion to previously unswept areas in the reservoir,increasing oil recovery.This paper provides a better understanding of the effects of formation damage during low-salinity water injection in Berea sandstones and proposes a methodology to forecast its effects on water relative permeability.We have performed one-and two-phase flow coreflood experiments with varying salinity of the injected solution.The effluent was analysed for salinity,pH and particle concentration.The laboratory data was treated using a mathematical model that accounts for permeability impairment during low-salinity waterflooding due to fines detachment,migration and capture in small pore throats.The parameters of the model were used to predict the impact of formation damage on the water relative permeability during two-phase flow.The mathematical model matched the laboratory data for pressure drop and particle concentration in the outlet with great accuracy.A methodology to predict the impact of formation damage in two-phase flow using one-phase flow data was proposed.A good agreement between the predicted and the measured water relative permeabilities was obtained.The forecasted water relative permeability curve was able to reproduce the non-monotonic behaviour of water relative permeability due to low-salinity water injection.
NSTL