The sustainability of water resources depends on the dynamic interactions among the environmental, technological, and social characteristics of the water system and local population. These interactions can cause supply-demand imbalances at diverse temporal scales, and the response of consumers to water use regulations impacts future water availability. This research develops a dynamic modeling approach to simulate supply-demand dynamics using an agent-based modeling framework that couple models of consumers and utility managers with water system models. Households are represented as agents, and their water use behaviors are represented as rules. A water utility manager agent enacts water use restrictions, based on fluctuations in the reservoir water storage. Water balance in a reservoir is simulated, and multiple climate scenarios are used to test the sensitivity of water availability to changes in streamflow, precipitation, and temperature. The framework is applied to the water supply system in Raleigh, North Carolina to assess sustainability of drought management plans. Model accuracy is assessed using statistical metrics, and sustainability is calculated for a projected period as the satisfaction or deficit of meeting municipal demands. Multiple climate change scenarios are created by perturbing average monthly values of historical inflow, precipitation, and evapotranspiration data. Results demonstrate the use of the agent-based modeling approach to project the effectiveness of management policies and recommend drought policies for improving the sustainability of urban water resources.