Foam is widely used in fractured reservoirs. However, few studies on the flow characteristics of foam fluid in fractures have been presented, and the flow mechanism of foam in complex fracture networks remains unclear. In this study, a variety of fracture models are used to systematically evaluate the flow characteristics of foam in fractures. First, based on the variable-thickness fracture and parallel fracture models, the variations in foam flow resistance and velocity are explored. Then, the foam flow path and sweep efficiency are evaluated with complex fracture network models. The results show that the foam flow resistance increases with increasing foam quality. At a higher foam quality (90%−92%), the pressure drop peaks and then decreases sharply as the foam quality increases. When the foam quality ranges from 50% to 90%, the foam volume increases with increasing foam quality, and the bubbles have larger diameters in thicker fractures. When foam flows in parallel fractures with different thicknesses, it preferentially flows in thick fractures (100 μm), and gas trapping occurs in the thin fractures. When the foam flows in a complex fracture network, the pressure drop increases with increasing foam quality and flow rate, and the foam quality corresponding to the maximum pressure drop is independent of the flow rate. In the vertical intersecting fracture network model, the range of flowing foam is the most extensive when the foam quality is 80%−90%. In the irregular fracture network model, when the foam quality reaches 92%, the volumetric sweep efficiency reaches a maximum of 86.97%. These findings reflect that it is necessary to consider fractures when foam flows in fractured/vuggy reservoirs and that reasonable predictions can be made with experimental results.

Summary We are interested in the development of surrogate models for the prediction of field saturations using a fully convolutional encoder/decoder network based on the dense convolutional network (DenseNet; Huang et al. 2017), similar to the approaches used for image/image-regression tasks in deep learning. In the surrogate model, the encoder network automatically extracts the multiscale features from the raw input data, and the decoder network then uses these data to recover the input image resolution at the output of the model. The input of multiple influencing factors is considered to make our surrogate model more consistent with the physical laws, which has achieved good results in the prediction of output fields in our experiments. Various reservoir parameters including the static reservoir properties (i.e., permeability field) and dynamic reservoir properties (i.e., well placement) are used as input features, and the water-saturation distributions in different periods are taken as the output. Compared with traditional numerical reservoir simulation, which has a high computational cost and is time consuming, not only does it present the same precision, but it costs less time. At the same time, it can also be used for production optimization and history matching.

No abstract

After a period of exploitation of carbonate fractured-vuggy reservoir, the water content of oil well keeps increasing, and some wells appear violent water flooding in a short period of time, which leads to the rapid decrease of production. The injection of particle plugging agent is one of the main measures to improve the oil recovery. Foam fluid has the characteristics of low density, high viscosity and low moisture filtration, which can effectively carry plugging particles to the target plugging layers. A stable foam system with high temperature and salinity is optimized and the rheological properties of the foam are studied in this research. The results show that the foam can maintain high viscosity and carry particles effectively even at a higher shear rate.

No abstract