Fuzzy Control of Semibatch Poly(ethylene terephthalate) Reactor Temperature based on Linguistic Rules

Research subject:

 It is not an easy task to get a suitable model of polymerization due to complex mechanism and kinetic of such processes. Polymerization temperature, as an intermediate variable between determining final polymer properties, is a good selection to be controlled. Fuzzy logic has ability to be applied to processes with unknown or less informed dynamics.

Research approach:

 In this research, control of semi batch poly(ethylene terephthalate) reactor temperature was studied. To do so, error and error variation were calculated using measured reactor temperature. Error and error variation were fuzzified using triangular membership functions. Five and three fuzzy sets were introduced to fuzzify error and error variation, respectively. Hence, fifteen rules were defined. Five fuzzy sets were defined to quantify these fifteen rules. Weight average defuzzification method was applied to calculate necessary heat to the reactor. Poly(ethylene terephthalate) was synthesized in a semi batch reactor based on a two steps method. It is possible to monitor temperature, pressure, rotation speed and mixing torque in this set up.

Produced water during esterification determines reaction advancement. In polycondensation step, mixing torque determines end of the process. Linguistic based fuzzy rules were applied to both steps. Reference temperatures were 230oC and 260oC, respectively. Reactor temperature was controlled with 1-2oC precision. Control logic was applied using C#.net real time programming.