A deep reinforcement learning(DRL)based approach for well-testing interpretation to evaluate reservoir parameters

Parameter inversions in oil/gas reservoirs based on well test interpretations are of great significance in oil/gas industry.Automatic well test interpretations based on artificial intelligence are the most promising to solve the problem of non-unique solution.In this work,a new deep reinforcement learning(DRL)based approach is proposed for automatic curve matching for well test interpretation,by using the double deep Q-network(DDQN).The DDQN algorithms are applied to train agents for automatic parameter tuning in three conventional well-testing models.In addition,to alleviate the dimensional disaster problem of parameter space,an asynchronous parameter adjustment strategy is used to train the agent.Finally,field applications are carried out by using the new DRL approaches.Results show that step number required for the DDQN to complete the curve matching is the least among,when comparing the naive deep Q-network(naive DQN)and deep Q-network(DQN).We also show that DDQN can improve the robustness of curve matching in comparison with supervised machine learning algorithms.Using DDQN algorithm to perform 100 curve matching tests on three traditional well test models,the results show that the mean relative error of the parameters is 7.58%for the homogeneous model,10.66%for the radial composite model,and 12.79%for the dual porosity model.In the actual field application,it is found that a good curve fitting can be obtained with only 30 steps of parameter adjustment.

Assessment on Evaluating Parameters of Rice Core Collections Constructed by Genotypic Values and Molecular Marker Information

Eleven evaluating parameters for rice core collection were assessed based on genotypic values and molecular marke＇ information. Monte Carlo simulation combined with mixed linear model was used to eliminate the interference from environment in order to draw more reliable results. The coincidence rate of range （CR） was the optimal parameter. Mean Simpson index （MD）, mean Shannon-Weaver index of genetic diversity （M1） and mean polymorphism information content （MPIC） were important evaluating parameters. The variable rate of coefficient of variation （VR） could act as an important reference parameter for evaluating the variation degree of core collection. Percentage of polymorphic loci （p） could be used as a determination parameter for the size of core collection. Mean difference percentage （MD） was a determination parameter for the reliability judgment of core collection. The effective evaluating parameters for core collection selected in the research could be used as criteria for sampling percentage in different plant germplasm populations.

Log identification method in complex sand reservoirs

In order to obtain effective parameters for complex sand reservoirs,a log evaluation method for relevant reservoir parameters is established based on an analysis in the gas-bearing sandstone with high porosity and low permeability,low porosity and permeability and on various characteristics of log responses to reservoir lithologies and physical properties in the Neopleozoic sand reservoir of the Ordos basin.This log evaluation method covers the Cook method that is used to evaluate the porosity and oiliness in high porosity and low permeability reservoirs and another method in which the mineral content,derived from geochemical logs,is used to identify formation lithologies.Some areas have high calcium and low silt content,not uniformly distributed,the results of which show up in the complex formation lithologies and conventional log responses with great deviation.The reliability of the method is verified by comparison with conventional log data and core analyses.The calculation results coincide with the core analytical data and gas tests,which indicate that this log evaluation method is available,provides novel ideas for study of similar complex reservoir lithologies and has some reference value.

Parameter Characterisation of Asphalt Mixtures for Visco-Elastoplastic Analysis

Asphalt mixtures exhibit strong viscous properties under repetitive loads. This phenomenon can be simulated in creep and recovery tests. By applying the visco elastoplastic model proposed, data recorded in the tests are interpreted. It is emphasised that applicability of the visco elastoplastic model depends on the proper method of characterising parameters involved in the constitutive equations. Since two or more strain components of elasticity and viscosity coexist during the loading or unloading, a measuring system of two loggers is introduced to decouple these components. Test data are analysed in three steps as elastic, visco elastic and visco plastic evaluations, using the records collected at the moment of unloading, during the periods of recovery and creep respectively. Factors that may influence the accuracy of data analysis are also discussed.

Aquifer parameter identification by the best chromosome clone plus younger generation chromosome prepotency genetic algorithm

This paper developed an improved combinatorial method called the best chromosome clone plus younger generation chromosome prepotency genetic algorithm （BCC-YGCP-GA） to evaluate aquifer parameters. This method is based on a decimal simple genetic algorithm （SGA）. A synthetic example for unsteady-state flow in a two-dimensional, inhomogeneous, confined aquifer containing three hydraulically distinct zones, is used to develop data to test the model. The simulation utilizes SGA and BCC-YGCP-GA coupled to the finite element method to identify the mean zonal hydraulic conductivities, and storage coefficients of the three-compartment model. For this geometrically simple model, used as a prototype of more complex systems, the SGA does not reach convergence within 100 generations. Conversely, the convergence rate of the BCC-YGCD-GA model is very fast. The objective function value calculated by BCC-YGCD-GA is reduced to 1/1 O00th of the starting value within 100 generations, and the hydraulic conductivity and storage of three zones are within a few percent of the “true” values of the ideal model, highlighting the power of the method for aquifer parameterization.

Characteristics of China's Automobile Recall Actions

For the purpose of enhancing automobile safety and the effectiveness of recalls,increasing manufacturers' quality control capabilities,and reducing defects,we investigated thestatus of recalls over the last five years.We have examined patterns and trends in motor vehiclesafety recalls using databases from China,Europe and the United States.A recall rate parameterwas developed to evaluate the level of recall enforcement based on analyzing the reasonsfor recalls and manufactures' characteristics.Moreover,there were substantial differencesin the recall rates of China compared with those in other countries,indicating that Chinese manufacturersneed to take more active action in recall.

Study on Process Economics of Natural Resource Utilization

Finite supply of non-regenerative resources triggers a competition between economic entities or between areas, which requires the ways regarding their utilization with higher levels in science and the standards regarding their use with higher efficiency in economics. To solve a problem of process evaluation in science and of driving force in economics during a process design or a process run for natural resources utilization, a process evaluation parameter originated from natural gas hydrate preparation from a small scale to industrialization scale and the equation of the criterion dependent are introduced to evaluate a variety of processes of natural resources utilization. The analyses indicate that the parameter is relevant to internal undeveloped resources amount change with a stable mass composition in a virtual black box model and external variable market with an implication of process efficiency in economy or of process efficiency of resource utilization and that the parameter has similar features of the thermodynamic state functions. Moreover, the equation of the criterion provided is a difference between the value of the process evaluation parameter at the final state and the value of the process evaluation parameter at the initial state in an actual process, which can be used to determine the direction of development and to judge the size of the driving force in an actual process or an economical run. The provided examples and correlative mathematical description can guide how an identification for the undeveloped resources and a real-time adjustment of dynamic production for the developing resource are done and how decisions regarding resource exploitation, the venture forecasting of capital utilization and updating technology are made. The parameter used itself and the equation of the derived criterion can help by playing a predictive role for selecting the optimal use processes and for designing new process of the natural resource utilization or capital use, and by playing a practical role for adjusting factual production status and for improving the actual process of the utilization of resource or capital in an economic society. Finally, those closed resource systems having accumulation or depletion of the resources or a variable mass such as a decomposition system, a fission system and a biological reproduction system will become possible future research objectives under the guide of this work.

Investigation of non-uniform leading-edge tubercles in compressor cascade:Based on multi-objective optimization and data mining

Corner stall receives noticeable attention in the aeroengine field as an important phenomenon in highly-load compressors.Non-uniform leading-edge tubercles,as an effective method to delay stall,are introduced into the compressor.In this paper,the shape of leading-edge tubercles was controlled by a third-order Fourier function.To judge corner stall,a more precise stall indicator for compressor cascade with flow control methods was defined.Besides,the total kinetic energy of the secondary flow at large incidence was adopted as a parameter for stall evaluation to save computing resources.The results of multiobjective optimization reveal that the loss coefficient exhibited negligible variation at design incidence,while the total kinetic energy of secondary flow showed a significant reduction at large incidence,resulting in a substantial increase in stall incidence.In the optimal profiling cases,the stall incidencewas delayed from 7.9°to 11.6°.The major purpose of the research is to provide proper design guidelines for nonuniformleading-edge tubercles and uncover the flow controlmechanisms of leading-edge profiling.Hence,the geometric features that meet different optimization objectives were extracted through geometric analysis near the Pareto Front and through Self-OrganizingMap(SOM)dataminingmethods in the optimization database.Besides,flow field analysis reveals the flow control mechanism of leading-edge tubercles.The convex-concave-convex structure at the 0%-70%blade height region can form two branches of leading-edge vortex pairs that are opposite in the rotation direction to the passage vortex.The two branches of leading-edge vortex pairs mixed with the leading-edge separation vortex to form two stronger mixed vortices,which can effectively suppress the development of passage vortex and delay stall incidence.

Heat Transfer Enhancement of Supercritical Nitrogen Flowing Downward in a Small Vertical Tube:Evaluation of System Parameter Effects

In this paper,the heat transfer enhancement(HTE)of supercritical nitrogen flowing downward in a vertical small tube(diameter 2 mm)is studied using the commercial software CFX of Ansys16.1,to provide theoretical guidance on the design of high-performance heat transfer systems.An effective numerical simulation method,which employs the SSG Reynolds stress model with enhanced wall treatment,is applied to study the heat transfer of supercritical nitrogen under typical working conditions.The objective is to evaluate the effect of the main parameters taking into account the buoyancy and flow acceleration effects.Simulation results are compared with results calculated from three well-known empirical correlations and the applicability of empirical correlation is discussed in detail.It is discovered that the Watts and Chou correlation accurately fits the simulation results of supercritical nitrogen and the Dittus-Boelter and Jackson correlations can only be used for high-pressure conditions.The HTE of supercritical nitrogen is closely related to the laminar sub-layer and buffer layer of a boundary layer.The buoyancy effect on the HTE should be considered at low mass flux conditions,and thermal acceleration can be completely ignored for the cases studied.The special HTE featured by the increment in heat transfer coefficient with increasing heat flux is discovered at low pressure,and simulation results proved that this HTE is caused by the combined actions of buoyancy as well as significant variations in specific heat and viscosity.

Parameter evaluation for soil erosion estimation on small watersheds using SWAT model

This research was undertaken for the evaluation of soil erosion using the semi-distributed basin scale SWAT model for four subcatchments of the Dhrabi River Catchment(DRC),which is located in the Pothwar Plateau region.Two subcatchments(catchment-25 and-31)are characterized by gullies while the other two(catchment-27 and-32)are managed with terraced landuse system.The performance of the model was satisfactory with coefficient of determination(R^(2))=0.67 to 0.91 and Nash-Sutcliffe efficiency(ENS)=0.54 to 0.85 for both surface runoff and sediment yield during the calibration(2009-2010)and validation(2011)periods.The PUSLE factor was found to be the most sensitive parameter during model calibration.It was observed that all of the rainfall-runoff events occurred during the monsoon season(June to September).The estimated annual sediment loss ranged from 2.6 t/hm^(2) to 31.1 t/hm^(2) over the duration of the simulation period for the non-terraced catchments,in response to annual precipitation amounts that were between 194.8 mm to 579.3 mm.In contrast,the predicted annual sediment levels for the terraced catchments ranged from 0.52 t/hm^(2) to 10.10 t/hm^(2) due to similar precipitation amounts.The terraced catchments resulted in 4 to 5 times lower sediment yield as compared to non-terraced catchments.The results suggest that there is a huge potential for terraces to reduce soil erosion in the DRC specifically and Pothwar area generally,which have proven to be an efficient approach to establishing soil and water conservation structures in this region.