The value of considering autocorrelation between inflows in the stochasticplanning of water resource systems

Reliability programming formulations offer a family of explicit stochastic models for planning the operation of complex water resources systems. These models use cumulative probability distributions of the sum of inflows to c haracterize their variability in the planning period. Applicability of thes e models for a variety of problems has been limited, mainly due to the assu mption of independence between inflows in different time periods that leads to the derivation of conservative operating policies. This paper presents three new approaches to overcome this limitation. The performance of the pr oposed approaches is demonstrated through comparison of the operating polic ies derived from these approaches and the Independent Approach. Operational planning of the Manitoba Hydro energy generation system, a predominantly h ydro-based utility company in Manitoba (Canada), is used as the case study.