A Monte Carlo-Based Adaptive Reduced Order Model for Gamma Density Measurement

Due to the frequently changing environments associated to thru-casing density measurement, the detector response varies largely. This creates the challenge to simulate all environmental variations to acquire detector responses, which is computational expensive and time consuming. To address this, a reduced order model (denoted as ROM) is proposed in this work to quickly obtain detector responses under changing parameters. The ROM is constructed by identifying correlations between detectors and environmental parameters, these correlations are then incorporated into a matrix which is solved using weighted least square method. During this process, a weighting factor is assigned to each environmental parameter for model optimization, this enables the ROM's flexibility in accommodating to rapidly changing environments and therefore improves its prediction accuracy. With its high adaptiveness and significant computational time saving ability, the proposed ROM provides an effective approach to quickly estimate logging feasibility in unknown environments before the job is executed