Optimal Placement of Pressure Monitoring Wells for CO2 Leakage Detection based on a Kalman filter and a Multi-Objective Genetic Algorithm

Measurement of pressure changes in monitoring wells located in a formation overlying an injection formation can provide an early warning for CO2 or brine leakage. If this strategy is to be part of an overall monitoring framework, then questions about how many monitoring wells are needed to detect a leakage event, and where should these well be placed, need to be addressed. In this study we present a methodology that uses a combination of a Kalman filter, a physically-based analytical model that solves for pressure propagation across old/abandoned leaky wells in a multi-formation system, and a multi-objective genetic algorithm, to answer the questions of how many wells should be used and where should they be placed. The Kalman filter is used to explore the covariance reduction based on possible well positions. The physically-based model is used to simulate, in a Monte Carlo scheme, a wide range of possible leakage scenarios where the random variable is the permeability of the old/abandoned leaky wells. The multi-objective genetic algorithm employed in this work is the Non-dominated Sorting Genetic Algorithm (NSGA-II), which is used to optimize three objectives: (i)The reduction of the total variance of the pressure field, (ii) the reduction of the number of wells used to detect a leakage event, and (iii) the reduction of the detection of leakage events which are not "harmful". In this work a "harmful" leakage event refers to an event in which the pressure change in the monitoring formation is large enough to induce leakage into the deepest potable water formation. The methodology is applied to a synthetic case study, which serves to prove the applicability of the methods and to gather insights on the strengths and weaknesses of using pressure monitoring wells to detect a CO2 leakage event.