With powerful nonlinear mapping and fitting ability of neural network, a production predicting neural network model is constructed. In view of high error, easy to default and other characteristics of oil field production data or data fitting and prediction is not easy to converge, a method of extended training data set and improved mean square error loss function are presented to get remarkable results in oil production fitting.

Temperature pattern in circulation phase of SAGD process is studied. Two in-situ points, three-dimensional heat conduct and partial penetration of horizontal well into reservoirs, are considered. The model is processed by dimensionless treatment, Laplace transformation, separation of variables and Stehfest numerical inversion. The results are in accordance with simulations by CMG software which verifies accuracy of the model. It shows relationship between temperature and heating time and concludes that the smaller the radial distance, the less the time is needed. Calculation plate can be helpful to design heating time and amount of heat.

Fractal approach is introduced. Mathematical model of fractal triple medium fissure cave reservoir with low velocity non-Darcy effect is established. By means of Laplace transform and Stehfest numerical algorithm, bottom hole pressure solution is obtained. Seepage characteristics analysis and nonlinear parameter sensitivity analysis are conducted. Finally, combined with actual well data fractal model is verified. It shows that seepage process of fractal triple medium fissure cave reservoir can be divided into 6 seepage stages. They are presence of wellbore storage, transition flow, channeling of caves to fractures, pseudo radial flow, channeling of matrix to fractures and caves, total radial flow, respectively. Furthermore, effect of velocity non-Darcy effect on percolation law is gradually increasing with time. The higher starting pressure gradient, the higher pressure dynamic curves in total radial flow stage bend upward. Fractal coefficient affects whole percolation process. With increase of fractal coefficient, fracture tortuosity and flow resistance increase and whole pressure dynamic curves move up.

A visco-elasticity polymer flooding mathematical model with three phases and five components is established with mass conservation principle.IMPES method is used to simulate the polymer flooding considering tackffying effect of elasticity,action of decreasing residual oil and visco-elasticity of heavy crude.It indicates that visco-elasticity polymer flooding decreases saturation of residual oil and increases viscosity of flooding system effectively,and then enhances oil recovery.As polymer concentration increases, oil recovery rises qucikly,and then tends stable.An optimal value is about 2 000mg·L^-1.As injection rate grows.oil recovery increases and then declines.There exists an optimum value.Oil recovery increases rapidly with the increasing of injection volume,and then tends stable.The best volume is 0.6PV.