Modeling and measurement of micromixing effects on bulk copolymerization in a stirred tank reactor
[Thesis]
M. Atiqullah
E. B. Nauman
Rensselaer Polytechnic Institute
1988
184
Ph.D.
Rensselaer Polytechnic Institute
1988
Micromixing problem is particularly well known in bulk copolymerization. This affects the product qualities. Therefore, in the present work, micromixing effects on styrene-methyl methacrylate, free-radical, bulk copolymerization were studied using a stirred tank reactor. Copolymer composition distributions have long been proposed for measuring micromixing effects. However, the necessary analytical techniques have been found difficult and ambiguous. The reason is interference between molecular weight, sequence length and composition distributions. Hence, first a technique for measuring copolymer composition distribution was developed. A high conversion, methyl-methacrylate copolymer was prepared in batch at 60C. Its composition distribution was measured using gradient elution, thin layer chromatography. The measured distribution qualitatively agreed with that calculated from conversion versus composition data of Dionisio and O'Driscoll (1979). Mixing experiments were conducted by copolymerizing these monomers in a laboratory scale reactor. Agitation was provided by a four-blade, 45, axial flow turbine. The reactor was run at 70C under various feed rates and stirrer speeds. Premixed feeds and feeds which were unmixed with respect to the monomers or the initiators were evaluated. Agitation Reynolds numbers were varied from 2 to 164. The measured composition distributions of the coploymers were sharp and close to the statistical composition distributions. The average compositions, calcuated from the measured distributions, agreed well with those measured by H-FT-NMR spectroscopy. The individual reaction rate of styrene and methyl methacrylate was modeled using the literature data. Based on this, the lamellar stretch model was applied to unpremixed-feed cases. The macroscopic composition distributions predicted by this model failed to conform to the measured distributions. Therefore, a new model, which has been named the Atiqullah-Nauman model, was developed. The composition distributions calculated by the new model satisfactorily matched the measured composition distributions. Unlike the lamellar stretch model, the new model considers the effect of backmixing. It can treat equal and unequal feed rates; premixed and unpremixed-feed situations. Using the Atiqullah-Nauman model, the segregation effect was quantified by segregation number. From this an upper limit of the initial striation thickness was calculated. This limit was approximately 1 mm for the present system.