Calculation of Average Reservoir Pore Pressure Based on Surface Displacement Using Image-To-Image Convolutional Neural Network Model

Abstract

The average pore pressure during oil formation is an important parameter for measuring the energy required for the oil formation and the capacity of injection–production wells. In past studies, the average pore pressure has been derived mainly from pressure build-up test results. However, such tests are expensive and time-consuming. The surface displacement of an oilfield is the result of change in the formation pore pressure, but no method is available for calculating the formation pore pressure based on the surface displacement. Therefore, in this study, the vertical displacement of the Earth’s surface was used to calculate changes in reservoir pore pressure. We employed marker-stakes to measure ground displacement. We used an improved image-to-image convolutional neural network (CNN) that does not include pooling layers or full-connection layers and uses a new loss function. We used the forward evolution method to produce training samples with labels. The CNN completed self-training using these samples. Then, machine learning was used to invert the surface vertical displacement to change the pore pressure in the oil reservoir. The method was tested in a block of the Sazhong X development zone in the Daqing Oilfield in China. The results showed that the variation in the formation pore pressure was 83.12%, in accordance with the results of 20 groups of pressure build-up tests within the range of the marker-stake measurements. Thus, the proposed method is less expensive, and faster than existing methods.

An Analytical Method for Timely Predicting of Coal Seam Pressure during Gas Production for Undersaturated Coalbed Methane Reservoirs

“…These can be interpreted by measuring the bottom hole pressure buildup after a stable production for some time, namely well test methods, which are mainly divided into Horner method [21], MBH method [22,23], MDH method [24], Dietz method [25], Chen's method [26], and other methods [27][28][29]. In addition, the coal seam pressure can also be predicted based on dynamic production data using models, including material balance equations [30,31], energy conservation equations [32], and artificial intelligence models [33].…”

Shear wave velocity prediction based on deep neural network and theoretical rock physics modeling

“…Thus, RNNs are well suited for processing sequential data, and since logging data are connected indepth, RNNs and their variants long short-term memory (LSTM) networks and gated recurrent units (GRU) networks have been introduced into the S-wave velocity prediction (Mehrgini et al, 2017;Zhang et al, 2020) and other rock parameters (Yuan et al, 2022). Moreover, convolutional neural networks (CNNs) have tremendous advantages in feature extraction, thus the CNNs were widely developed and applied in many research fields (Yuan et al, 2018;Hu et al, 2020;Hu et al, 2021), and a combination of RNNs and CNNs for S-wave velocity prediction has been proposed recently (Wang et al, 2022;Zhang et al, 2022). However, the neural networkbased S-wave velocity prediction method has poor generalization and limited labels for establishing S-wave velocity prediction networks, which brings many difficulties to real applications.…”