Abstract: Aleatory and epistemic uncertainties exist in the previous analyses of construction scheduling for a concrete gravity dam due to the randomness of its construction environment and technical operation and the incomplete cognition of its cable operating parameters and repair time. Previous studies lack a comprehensive consideration of these two types of uncertainties. The evidence theory has the advantage of solving aleatory and epistemic uncertainty problems comprehensively; but in its traditional applications, the vertex method, usually used for calculating the extreme values of monotonicity problems, is inapplicable to construction schedule simulations or other non-monotonic problems, or to consideration of the relevance of construction parameters. This paper presents a new uncertainty analysis method of construction scheduling for concrete gravity dams based on a modified evidence theory. First, it uses a genetic algorithm to modify the extreme value calculating procedure and whereby searches the extreme value on an evidence body. Second, it adopts the Copula function to quantify the correlation between construction parameters and modifies the joint-probability calculating method by using this correlation coefficient as a difference. Then, it transforms the construction parameters with heterogeneous uncertainties into evidence variables, and calculates the belief and plausibility degree to quantify the uncertainty of construction schedule. Finally, this method is applied to an analysis of construction scheduling for a concrete gravity dam, and its consistency and superiority in uncertainty analysis is verified by comparing with the aleatory uncertainty theory.

Key words: concrete gravity dam, aleatory and epistemic uncertainty, modified evidence theory, genetic algorithm, correlation between construction parameters, construction schedule simulation