K. Katterbauer, A. Alhashboul, H. Chen, A. Yousef
Saudi Aramco,
Saudi Arabia
Keywords: CO2 injection performance, saline aquifers, artificial intelligence;
Summary:
CO2 injection for CO2 storage performance assessment represents an important consideration in determining carbon sequestration operational performance. Saline aquifers have been of considerable interest in order to permanently store the CO2 within the formation as the CO2 dissolves within the brine. The trapping of the CO2 is both in terms of physical trapping and geochemical trapping. Physical trapping arises from inability of the CO2 to permeate through the cap roc, while geochemical trapping is the dissolution of the CO2 within brine water. The interaction of the CO2 with rock formation and salinity may lead to significant challenges in the well performance determination. Pressure transient analysis represents an important factor for effective CO2 storage as well as CO2 geothermal technology, as it enables to analyze reservoir characteristics based on the pressure response data. Specifically, injection-falloff cycle pressure and rate transient response enables to determine permeability, distance to boundaries and injection performance. This enables to determine additionally mobility of the supercritical CO2 phase as well any well damage. In order to model the PTA response subject to temperature and rate dependent supercritical CO2 injection, a physics-based deep-learning model was developed to take into account temperature and rate effects into the pressure transient analysis. The deep learning model utilizes a time-series based adapted Long Short Term Memory Network for the estimation of the pressure response. We investigated the pressure response arising from temperatures effects based on the Ahuroa saline aquifer in New Zealand. The Ahuroa saline aquifer is part of the Ahuroa gas reservoir that is located within the Urenui formation and has been utilized for the storage of natural gas. The saline aquifer is intersected by the Ahuroa-3 that is utilized for CO2 injection. Several injection-fall off tests were simulated with different CO2 injection rates, and temperature dependent effects. The results indicate that the temperature effects may impact injection performance due to near well phase changes affecting the wellbore environment. Specifically, the explainable model indicates that while temperature effects impact the pressure, the impact is limited as compared to that of rates, porosity and the viscosity of the CO2. Determining temperature and pressure effects for injection falloff tests for CO2 injection into saline aquifers is critical in order to determine CO2 storage injectivity performance and overall CO2 storability within the formation. While saline aquifers are attractive for trapping CO2, the two-phase nature requires a solid analysis of the various effects that have been outlined by the analytical deep learning model. This enables to determine additionally mobility of the supercritical CO2 phase as well any well damage.