The physical-informed neural network (PINN) model can greatly improve the ability to fit nonlinear data with the incorporation of prior knowledge, which endows traditional neural networks with interpretability. Considering the seepage law in the tight reservoir after hydraulic fracturing, a model based on PINN and two-dimensional seepage physical equations was proposed, which can effectively predict the flow field distribution of the tight reservoir after fracturing. Firstly, the dataset was obtained based on physical and numerical models of the tight reservoirs developed by volume fracturing. Furthermore, coupling the neural networks and the two-dimensional unsteady seepage equation, a PINN model was developed to predict the flow field distribution of the tight reservoir. Finally, a systematic study was performed concerning the noise corruption levels, training iterations, and training sample size that affect the prediction results of PINN models. Besides, a comparison between PINN and traditional deep neural networks (DNN) was presented. The results show that the DNN model was not only sensitive to noisy data but also more vulnerable to overfitting as the training iterations increase. In addition, the prediction accuracy cannot be guaranteed when the samples are inadequate (<500). In contrast, the PINN model was less affected by noise and training iterations and thus indicates greater stability. Moreover, the PINN model outperforms the DNN model in the case of inadequate samples attributing to prior knowledge. This study confirms that the adopted PINN model can provide algorithmic support for the accurate prediction of flow field distribution of the tight reservoirs.

South Ordos sandstone reservoir is mainly featured by tiny pore, which mainstream throat radius is around 50nm, high filtration resistance, resulting in low oil productivity and more obvious non-linear seepage characteristics. As of low formation pressure, well production is poor and declines dramatically, therefore primary recovery is hard to sustain effective development for the reservoir. The core problem of tight oil development focuses on the evaluation of tight matrix flowing capability and reservoir producing condition. In the paper, in Ordos typical tight oil basin, by means of microscopic flowing simulation, numerical simulation as well as lab experiments results, single-phase and oil-water two-phrase flowing mechanisms have been analyzed, revealing tight oil single phase percolating resistance and movable oil saturation, providing key evaluation parameters for effective reservoir division. For oil-water two-phase flowing, Jamin effect is so serious that water flooding is hard to displace the oil in micro-pores, accordingly relative permeability and displacement efficiency are calculated. Tight matrix-fracture coupling model recovery mechanism have been analyzed, effective producing radius and mechanism of matrix are defined in the condition of fracturing horizontal wells developing, according to which productivity percentage of Ordos tight oil between fracture and matrix have been determined. On basis of geology evaluation and reservoir engineering analysis, correlation of geological properties-well dynamic characteristics are set up, then influencing factors have been studied to identify tight oil producing conditions on depletion development at different oil price. As different classified fracture developed in the reservoir, water flooding producing condition has been studied, laying the foundation for study of effective development method and technical strategy. Our research indicates that Ordos tight matrix is of low productivity, with movable water saturation increasing, well productivity sharp decline. During production period, production ratio from fracture is only amounted to 6~14% of accumulation oil. Fully excavating the potential of matrix reserves is predominant to achieve effective development of tight oil. Owing to high start-up pressure gradient, as high as 0.1~0.2MPa/m, for water flooding development, well spacing should be reduced to 50m□ to set up pressure response without fracture developing. While in Ordos basin natural fracture is developed, water channeling is so heavy that accumulative oil is lower than depletion method. CO2 start-up pressure gradient is far smaller than that of water flooding with composite EOR mechanisms, expected to be an effective injection medium for tight oil. It is a critical period how so many shut-in wells could be revitalized under low oil price condition. Relying on research results, Ordos tight oil new development method target has been determined, promoting application research and pilot test on CO2-gelled fracturing fluid and effective injection fluid sustaining matrix displacing pressure in tight oil development.