Improved Global Sensitivity Analysis for Critical Factor Analysis with Intercorrelated Inputs: Power Plant Cost Example

Abstract

Given the importance of early conceptual cost estimating, it is important for the estimators to identify the most critical factors at the early phase of a project that drive the final cost. This paper presents a sensitivity analysis–based critical factor analysis method (SA-CFA) to evaluate and rank the cost impacts of individual project parameters that are known in the proposal development phase. Two challenges are identified and tackled with the proposed method: intercorrelation among project parameters and the joint impacts of multiple project parameters on the final cost, or interaction terms. Specifically, an independent component analysis method is proposed to convert intercorrelated project parameters into independent parameters, and the Sobol’s global sensitivity analysis is used to calculate total sensitivity indices that capture the joint impacts of interaction terms. The proposed SA-CFA method was tested in a numerical simulation example and a real power plant case study. The results confirmed the effectiveness of the proposed method in identifying cost-driven factors.