Global parameter sensitivity analysis using Sobol and extended Fourier amplitude sensitivity test methods for associated gas injection in the shale oil reservoir

After volume fracturing of shale oil reservoirs, rapid oil rate decline and poor development performance are observed. Associated gas huff-and-puff has emerged as a promising method to enhance shale oil recovery. However, the influence patterns of key parameters on huff-and-puff performance remain poorly understood and lack systematic quantitative analysis. To address this, three categories (reservoir parameters, fracturing parameters, and operational parameters) with a total of 11 parameters were selected. Radial basis function neural network models were constructed under different parameter configurations, and global sensitivity analyses were performed using Sobol and extended Fourier Amplitude Sensitivity Test (EFAST) methods to quantify parameter influences on cumulative oil production and huff-and-puff oil production. Orthogonal experimental design was also applied to production prediction in a field case. Results indicate: (1) Sobol and EFAST methods provide consistent sensitivity analysis results. For cumulative oil production, oil saturation, effective thickness, fracturing scale, and permeability have the most significant direct effects, with strong interactions. For huff-and-puff oil production, oil saturation, injection timing, and effective thickness dominate with strong parameter interactions. (2) Under certain reservoir and fracturing conditions, gas injection volume and injection timing are the most critical parameters affecting huff-and-puff oil production, with notable interactions among the four operational parameters. (3) Favorable associated gas huff-and-puff production can be achieved with high injection volumes and rates, a relatively long soaking time, and moderate gas injection timing, resulting in an increased oil production by 31.2%. These findings can serve as a reference for optimizing strategies for developing shale oil reservoirs.