A method based on nonlinear programming for determining the optimal op
eration of general water distribution systems containing multiple sour
ces and reservoirs is presented. The problem is formulated and solved
so that, given the forecasted demands for the coming 24 hours, the ini
tial and final conditions in the reservoirs, the unit and maximum dema
nd electricity charge, and the constraints in the hydraulic properties
of all system components, an optimized pumping schedule is found. An
optimization algorithm which employs the generalized reduced gradient
method and the nonlinear sensitivity analysis has been developed for a
basic scheduling problem in which only unit charges are considered. T
he maximum demand charge, which is weighted by varying degrees from da
y to day, is incorporated into the scheduling problem. The algorithm u
ses a feasible initial solution as the starting solution and iterates
so that all the interim solutions are feasible.