Pressurization Risk Assessment of CO2 Reservoirs Utilizing Design of Experiments and Response Surface Methods

Abstract:

Monitoring of pressure buildup can provide explicit information on reservoir integrity and is an appealing tool, however pressure variation is dependent on a variety of factors causing high uncertainty in pressure predictions. This work evaluated pressurization of a reservoir system in the presence of leakage pathways as well as exploring the effects of compartmentalization of the reservoir utilizing design of experiments (Definitive Screening, Box Behnken, Central Composite, and Latin Hypercube designs) and response surface methods. Two models were developed, 1) an idealized injection scenario in order to evaluate the performance of multiple designs, and 2) a complex injection scenario implementing the best performing design to investigate pressurization of the reservoir system. A holistic evaluation of scenario 1, determined that the Central Composite design would be used for the complex injection scenario. The complex scenario evaluated 5 risk factors: reservoir, seal, leakage pathway and fault permeabilities, and horizontal position of the pathway. A total of 60 response surface models (RSM) were developed for the complex scenario with an average R² of 0.95 and a NRMSE of 0.067. Sensitivity to the input factors was dynamic through space and time; at the earliest time (0.05 years) the reservoir permeability was dominant, and for later times (>0.5 years) the fault permeability became dominant for all locations. The RSM’s were then used to conduct a Monte Carlo Analysis to further analyze pressurization risks, identifying the P10, P50, P90 values. This identified the in zone (lower) P90 values as 2.16, 1.77, and 1.53 MPa and above zone values of 1.35, 1.23, 1.09 MPa for monitoring locations 1, 2, and 3, respectively. In summary, the design of experiments and response surface methods allowed for an efficient sensitivity and uncertainty analysis to be conducted permitting a complete evaluation of the pressurization across the entire parameter space.