An experimental study on the on-line optimizin
g control of a sample free radical bulk polymerization system, namely, methyl methacrylate (MMA), is carried out in a rheometer–reactor assembly. Two initiator loadin
gs and three cases involvin
g external disturbances (power failure) are studied. The disturbances are assumed to be of two kinds: one that leads to a sudden increase in the temperature of the reaction mass (coolin
g water pump failure) over the planned temperature history,
N68NKN-1&_mathId=mml157&_user=10&_cdi=5229&_rdoc=15&_acct=C000050221&_version=1&_userid=10&md5=821746cc7a1542ce18baedaa30faf6d1"" title=""Click to view the MathML source"">T(t), and one leadin
g to a sudden drop in the temperature (heater failure). The temperature and the viscosity,
η, histories are used to describe the ‘state’ (conversion,
xm, and wei
ght-avera
ge molecular wei
ght,
Mw) of the polymerizin
g mass. The polymerization is first carried out under an
off-line computed optimal temperature history,
Top(t), obtained usin
g the adapted jumpin
g gene version of the elitist
genetic al
gorithm (GA-II-aJG). A planned disturbance is introduced after the start of polymerization and
continues for a pre-specified duration. A new optimal temperature history,
Treop(t), is calculated
on-line (in about 3 min of real time) usin
g GA-II-aJG. This is implemented as soon as the disturbance is rectified. Experimental values of
xm(t),
Mw(t) and
η(t) are also measured. These are observed to be in
good a
greement with model predictions for all the cases. It is found that the information on the viscosity of the reaction mass can be used effectively for on-line optimizin
g control. This can help ‘save’ the batch (
give a product havin
g the desired values of the avera
ge molecular wei
ghts) optimally, in as short a reaction time as possible. The effect of re-tunin
g of the model parameters usin
g experimental data on the temperature,
Texp(t), and the viscosity,
η(t), is also demonstrated.