Influence of natural caves on hydraulic fracturing pressure curves: numerical modelling and ANNs

In carbonate fracture-cavity reservoirs, the resource of gas/oil mainly distributes in natural fractures and caves; therefore, it is critical to understand the property of the pre-existing caves. However, currently it is still lack of direct method to detect the hydraulic parameters of these natural caves in deep underground. This work develops a potential useful method to determine the hydraulic parameters of natural caves according to hydraulic fracturing pressure curves (HF pressure curves), which can be easily obtained on the ground surface. Firstly, the connection between hydraulic fractures and natural caves is modelled using TOUGH-AiFrac, and the experiences of HF pressure curves are recorded. Secondly, for pattern recognition, the experience of HF pressure curves is divided into three stages and four features are extracted. Thirdly, the implicit relationship between HF pressure curves and natural caves is generated using artificial neural networks (ANNs). The results show that the properties of natural caves are able to be predicted by HF pressure curves with low errors. © 2021, Saudi Society for Geosciences.