A new soft computing‐based approach to predict oil production rate for vapour extraction (VAPEX) process in heavy oil reservoirs

Abstract

There are vast resources of heavy oil and bitumen reservoirs in the Western Canadian Basin. For many of them up to 95 % of reserves still remain in place, and by considering the increase in future energy demand these abundant resources can be considered as potential sources for future years. Recently, solvent‐based heavy oil recovery methods such as vapour extraction (VAPEX) have gained attention due to the potential environmental and economic assets over thermal processes. Due to the complexity of the mechanisms associated with the solvent injection process (i.e. diffusion and gravity drainage processes), such models are incapable of accurately predicting the production rate during the VAPEX process. In this study, the artificial neural networks (ANN) technique is utilized to tackle the limitations that analytical methods encounter while predicting the complex relationships, where there is uncertainty, imprecision, and partial truth. Hence, in the first phase of the research a comprehensive experimental study in two large‐scale, visual rectangular VAPEX models was carried out by utilizing various injection solvents. Based on an extensive literature review and experimental results, the drainage height, solvent type, permeability, porosity, and heavy oil viscosity were considered as the inputs of the model to predict the heavy oil production rate as the output of the model. After trying different training scenarios, it was found that the back‐propagation learning algorithm can be successfully used to predict the ultimate recovery factor after implementing the VAPEX method in the heavy oil system of interest.

Experimental methods in chemical engineering: Artificial neural networks–ANNs

“…Eng. [5][6][7][8][9][32][33][34] ), ANNs are unboundedwe apply them across all specialties including polymerization, oil production, battery heating, modelling, control of industrial plants, and catalysis. Most of this research applies commercial software packages like the MATLAB Deep Learning Toolbox.…”

“…This analysis merely demonstrates the impact inputs have on outputs rather than identifying the underlying mechanism, but it is an improvement over a strictly black box representation. Feed‐forward ANNs also predicted heavy oil production rates . Moradi and Parvizian estimated yield, selectivity, and conversion of synthesis gas to dimethyl ether based on temperature, pressure and /CO molar ratio in the feed with a feed‐forward ANN.…”

“…Assenhaimer et al [32] Aging of metalworking fluid emulsion Torabi [33] Oil production rate for vapour extraction process [34] Epoxidation of waste soybean cooking oil [38] Optimize fed-batch streptokinase fermentation Control Elman, feed-forward, and auto-associative networks…”

“…Feed-forward ANNs also predicted heavy oil production rates. [33] Moradi and Parvizian [36] estimated yield, selectivity, and conversion of synthesis gas to dimethyl ether based on temperature, pressure and H 2 /CO molar ratio in the feed with a feed-forward ANN. They compared multiple BP algorithms, including Levenberg-Marquardt, scaled conjugate gradient and gradient descent with momentum training methods.…”

The prediction of shale gas well production rate based on grey system theory dynamic model GM(1, N)

“…The prediction of shale gas production rates and volumes is an important part of oilfield development (Elmabrouk et al 2014). Whether the production of shale gas wells can be predicted effectively in the future, based on the historical data, is related to the real-time adjustment of the shale gas well working schedule, thus playing the role of an assistant during decision-making (Mohammadpoor and Torabi 2018).…”