(93g) Combined Use of MILP Optimizer and Process Simulator to Evaluate Control of Water Treatment Tanks

It is desired to develop a simulator which combines a dynamic process simulator model with a MILP optimizer. This will allow evaluation of the use of the MILP optimizer for setpoint determination on a virtual plant in the dynamic process simulator. Using a custom interface will also allow for operator training.

A mixed integer linear programming (MILP) optimizer was used to calculate the setpoint for controls on a water treatment plant. The system consists of a series of large storage tanks at different elevation, with pipeline connecting the tunnels. It is important to minimize process disturbance (change of level in the various tanks) to maintain proper turbidity of the water. A MILP optimizer has been developed to calculate setpoint for the system, based on predicted future demand scenarios.

In order to evaluate the MILP optimizer, a process model was created in a rigorous process simulator (CHEMCAD). This dynamic model considers hydraulic balances on pressure and flow.

The MILP optimizer was connected to the Process simulator in a custom user interface.

The custom user interface allows:

1 _ operator training using the dynamic process model as a virtual plant

2. determination of optimum operation of plant by use o f the simulation and separate MILP optimizer