The Stress Sensitive Prediction Model Of Carbonate Reservoir In Shunbei Block Based On BP Neural Network

Carbonate reservoirs are rich in oil and gas resources,Shunbei Block is a carbonate fractured-cave oil reservoir with ultra deep oil and gas burial,which have shown rich oil and gas resource and have a remarkable economy benefits.Reservoir fractures and caves develop.During the actual drilling and production,the solid particles of the drilling fluid are easy to invade the reservoir.When the production differential pressure or mining speed is not reasonable,the shale migration may occur,blocking the pore throat,or the fracture may be closed,causing stress sensitive damage.Stress sensitive damage is one of the main damages of reservoir,which is the key technical issue for the efficient development of carbonate reservoirs.The conventional indoor evaluation method is limited by the fact that the reservoir of Shunbei oilfield is buried too deep,it is difficult to take out the core and preserve the fracture when preparing the plunger,and the poor representativeness of the core makes the experimental results random or accidental.Increasing the amount of experiments can improve the reliability of the results to a certain extent,but it will lead to problems such as longer research cycle and higher research cost.In order to study the stress sensitive damage of carbonate reservoir efficiently,the mathematical method is applied to predict the stress sensitive damage of carbonate reservoir considering the interaction of stress sensitive factors.Taking the damage degree coincidence rate and prediction error of the model prediction results as the evaluation criteria,the common mathematical methods for predicting the reservoir sensitivity were summarized and compared.The result showed that the neural network method is more accurate in predicting reservoir sensitivity.Taking the carbonate reservoir data of Yingshan formation in Shunbei block as the research object,the correlation between influencing factors and stress sensitive damage was analyzed,and the main influencing factors of stress sensitive of carbonate were determined.Based on the BP neural network theory,a stress sensitive prediction model of carbonate reservoir was established.The model was tested by the return test method.The result showed that the prediction coincidence rate of stress sensitive damage degree is 100%,and the error of prediction result is less than 10%.In order to test whether the neural network method is still the best method for predicting the stress-sensitive damage of carbonate reservoir,the stress-sensitive prediction models built by different mathematical methods were compared.The result showed that the prediction results of BP neural network model are the best,which is consistent with the optimization results of mathematical methods in literature research.In order to verify whether the BP neural network model is suitable for other carbonate reservoir,the stress sensitivity of carbonate reservoir in Yi Jianfang formation was predicted and compared with the results of laboratory experiments.The result showed that the average error between the predicted results and the experimental results is less than 10%,indicating that the BP neural network model is also suitable for predicting other carbonate reservoir Stress sensitive damage.