A multivariable nonlinear control strategy which accounts for unmeasured st
ate variables and input constraints is developed for the free-radical polym
erization of methyl methacrylate in a continuous stirred tank reactor. Mono
mer concentration and reactor temperature are controlled using a technique
which combines nonlinear input-output decoupling and linear model predictiv
e control. Input constraints are handled explicitly by applying linear mode
l predictive control to the constrained linear system obtained after feedba
ck linearization and constraint transformation. Unmeasured initiator and so
lvent concentrations are accounted for by treating the live polymer concent
ration as an unknown parameter which is estimated on-line. The performance
of the control strategy is compared to other nonlinear control techniques t
hrough closed-loop simulations.