Optmization of simulink model parameters using algorithm written in matlab script

I am doing an optimization of a dynamic system of multi source- multi user water supply system using simulink. The time series inputs are rainfall, evaporation , user demands etc. I am using gain blocks (x1..xn) to determine the fraction of water from each source to each user. I would like the optimizer to run the simulink model for each hour (simulation time step) and optimize the parameters (x1..xn) so that the energy use associated with supply is minimum (objective function). Before going to the next step, the function should update the state variables (like water storages in dams and tanks etc which is determined by x values) in simulink and use them as initial inputs for the next optim step.

Currently i am running it but the supply is a lot more than demand (constraint is total supply= total demand), so i don't think correct optimiz is taking place. Could you suggest a solution for this problem?

Sudeep