Benchmarking physics-based GPT and empirical Arps DCA for EUR and CO₂ storage capacity assessment of depleted gas reservoirs

Accurate well performance forecasting is crucial for optimizing hydrocarbon recovery and also for repurposing depleted reservoirs for CO₂ storage. This study compares the physics-based Gaussian Pressure Transient (GPT) method with the empirical Arps Decline Curve Analysis (DCA) method for history matching and predicting production/injection behavior of legacy gas wells in the UK North Sea. Using bootstrapping of historical production-rate data from the Kelvin, Mimas, and Tethys Fields, we evaluate each method’s effectiveness in capturing reservoir performance over the full well-life. Our findings show that both methods can accurately history-match historical production data. However, GPT outperforms the Arps method in predicting production from bootstrapped, short-term production data, demonstrating better resilience to noisy data and more accurate estimation of the ultimate recovery (EUR). Subsequently, CO₂ injectivity and storage capacity were evaluated for abandoned wells in lease Blocks 43 and 48. The Kelvin Field well (Block 43) showed the highest injectivity at 10.96 Mscf/day/psi, more than three times that of the Mimas and Tethys wells (2.04–3.00 Mscf/day/psi) in Block 48. The three wells collectively offer a storage capacity of 16.91 Bscf (∼1 Mt) over a certain injection-period duration. Wells in Block 48, have potential for continued CO₂ injection, suggesting opportunities for long-term storage optimization.