Computer simulations were performed using the Dynamic Matrix Control (
DMC) algorithm for the control of a reverse osmosis (RO) desalination
pilot plant. The plant dynamics were modeled by the use of empirical t
ransfer functions developed at the Kuwait Institute for Scientific Res
earch. The two output variables considered were: production flow rate
and product water electrical conductivity as a measure of purity. With
the selection of proper tuning constants the DMC approach allows for
substantial improvement over proportional-integral (PI) control based
on the integral square error (ISE) performance criterion. A setpoint c
hange in production flow rate was considered in the simulations since
this type of change is the one most likely to occur in an RO plant. Fo
r the same magnitude setpoint change in flowrate, the ISE results for
the flowrate were comparable between the two controllers; however, the
ISE results for conductivity were substantially better using DMC cont
rol. The ISE for conductivity using the PI controller is 1688.99 while
for the DMC controller, with the proper tuning parameters the ISE may
be much less than 1.0. These results indicate that much more flexibil
ity in the operation of an RO plant is available with DMC control.