Author:

Abstract:

Water of sufficient quantity and quality is indispensable for domestic and industrial use, irrigated agriculture, hydropower generation and ecosystem functioning. River systems are a major source of water to meet these demands. Excess surface water resulting into floods is harmful though for properties and infrastructures and may put human lives at risk. Since surface water availability is variable in space (e.g. related to latitude and orography) and time (seasonal) it often does not meet the spatio-temporal use requirements. To overcome this discrepancy reservoirs have been or are planned to be constructed on many large and smaller river systems throughout the globe. Management of these reservoirs with a view to allocate the available water fairly and to prevent floods requires time series data on inflow, demands and losses. In this contribution, this water allocation problem is considered as a discrete spatio-temporal Network Flow Optimization Problem (NFOP) solved by Linear Programming (LP) techniques. We apply NFO-LP to a virtual river system with two reservoirs, and compare different scenarios in which reservoir capacity, inflow, demands and losses vary. In a next step we intend to upgrade the LP-model to a Mixed Integer Linear Programming (MILP)-model in order to determine optimal locations of reservoirs.