Multidisciplinary Data Analysis Method And Application Research For Oilfield Development Evaluation

Oilfield development process is a complex non-linear dynamic human-nature interaction system.With the development of the reservoir,the distribution of oil,gas and water in oil reservoir,reservoir porosity,permeability and other rock physical properties,oil-bearing indicators have undergone complex dynamic changes.Compared with the initial state of the reservoir,the correlation between data distribution characteristics and reservoir attributes is worse,and it is difficult to accurately describe them with deterministic mechanism models and statistical formulas in the later stage.Along with the exploration and development of oilfields in China for many years,abundant multidisciplinary dynamic and static data have been accumulated.If using the methods of geostatistics and artificial intelligence information processing to study and establish a new method and model for evaluating the evolution law of geological characteristics in reservoir development through multidisciplinary data analysis,we can provide a new method for comprehensive evaluation of reservoir and current situation analysis of well group unit production.Based on the abundant data and knowledge accumulated in a certain oilfield development area,and facing the typical problems such as the evolution law of geological parameters,the change characteristics of development indexes and the evaluation of watered-out status of oil reservoirs,this paper carries out the prediction method research based on geostatistics and dynamic neural network based on the multidisciplinary data of Oilfield development.The main innovative work of this paper includes the following four aspects:1.The organization and application model of oilfield multidisciplinary data is established.The data collected in the research area include geology,testing,production performance,coring well analysis and other multi-disciplinary static and dynamic information and process data.The organization and application model of multidisciplinary data is established for the subject data analysis.2.Reservoir parameter prediction method based on geostatistics.According to the geological characteristics of the research block,the Universal Kriging predictionmethod isconstructedwhich is in line with the theory of reservoir sedimentation.Based on the core analysis data of coring wells,the plane distribution prediction and analysis of reservoir porosity,permeability and other physical parameters are realized.3.Development index prediction and reservoir watered-out status discrimination based on radial basis process neural network.Considering the complexity of underground reservoir property changes in oilfield development,based on production data and core analysis data of coring wells,a radial basis process neural network model and algorithm are constructed to realize dynamic prediction of development indexes of research blocks and identification of water flooding status of reservoirs.4.Development of application software system.Based on the above research results,an application software system is developed to realize the prediction and analysis of reservoir geological parameters and development indicators.Facing the practical research of reservoir development geology,based on oilfield data,this paper establishes a prediction model which conforms to the evolution law of reservoir geological parameters and the change characteristics of development indexes in information processing mechanism,develops an application software system,and achieves good results in actual data processing.