Simulation Of LDPE Autoclave Reactor Based On Computational Fluid Dynamics

Low-density polyethylene (LDPE) autoclave reactor is a constantly stirred vessel with high height-diameter ratio and multiple impellers of different types. The flow behavior generated by the impellers in the reactor is related to the reaction process. The degree of mixng not only affects the reaction rate and conversions of the free-radical polymerization, but also affects the molecular weight and the degrees of long chain branches (LCB) and short chain branches (SCB). However, the flow characteristics, the coupling between flow process and chemical reactions, and the effect of mixing on the product properties, especially the effect of micromixing on the LCB and SCB have not been studied. In addition, it is difficult to study the LDPE process through the experiment in the laboratory because of the high temperature (200-300℃) and the pressure (2000-3000 atm). Therefore, this thesis employed the computational fluid dynamics (CFD) to study the macromixing and micromixing characteristics of the experimental LDPE autoclave reactor. The simulated results were compared with the experiment to validate the feasibility of the simulated methods. Furthermore, the reaction kinetics was coupled with the CFD model to study the macromixing, heat transfer and reaction process in the industrical scale LDPE autoclave reactor. The micromixing model was also coupled to study the effect of micromixing on the LCB/SCB. The reason why the LDPE producted by the autoclave reactor had more LCB was discussed.The main research achievements of this paper are as follows:(1) The macromixing characteristics of experimental LDPE autoclave reactor stirred with 32 impellers were studied by CFD. The simulated results were compared with the experimental data. The mixing cures showed an excellent agreement with the experimental results. The residence time distribution (RTD) showed a good agreement with the experimental results. The minimum error of mean RT (tm) was 2.78%, and the maximum was 10.85%. The minimum error of σ/tm was 3.12%, and the maximum was 17.68%. These results validated the feasibility of CFD as a tool to study the LDPE autoclave reactor. The macromixing characteristics were further analyzed. The results showed that the reactor could be divided into two reaction zones. In each reaction zone, the downward flows near the reactor wall and the upward flows near the shaft formed a big circulation loop. In each big circulation loop there were many small radial circulation loops. Finally, a more accurate macromixing model was proposed.(2) The reaction kinetics was coupled with the CFD model to study the macromixing, heat transfer and reaction process in the industrical scale LDPE autoclave reactor. The predicted temperature distribution showed an excellent agreement with the industrical data. The mean error was 0.74%, which validated the feasibility of CFD simulations. The temperature and concentration disributions (including initiator, monomer, radical, and polymer) were further discussed. The method of moments was used to predict the molecular weight distribution (MWD, NWD, and PDF). The viscosity of LDPE in the reactor was defined as a function of the reaction process. The distribution of LCB and SCB were studied and it was found that the effect of micromixing on the LCB/SCB reaction could be consided in simulation model.(3) The micromixing characteristics of experimental LDPE autoclave reactor stirred with 32 impellers were studied by CFD. The CFD model was coupled with the micromixing model to establish an innovative finite rate/eddy dissipation-engulfinent (FR/ED-E) model. The simulated results were compared with the experimental data. The minimum error was 0.79%, and the maximum was 8.57%. These validated the feasibility of the established model. The micromixing characteristics of the reactor were further analyzed. It was found that the micromxing characteristics in the reactor differed because of the location of acid injection points. The different acid injection points affected the product selectivity of a parallel competitive reaction system. The acid injection point al favored the formation of the fast reaction, but a2/a4 favored the formation of the slow reaction.(4) The reaction kinetics and micromixing model were coupled with the CFD to study the effects of micromixing on the LCB/SCB reactions in the industrical scale LDPE autoclave reactor. Because the effect of micromixing on the reaction process was considered, the predicted temperature distribution showed a more excellent agreement with the industrical data. The mean error reduced from 0.74% to 0.10%. The predicted distribution of LCB/SCB was also agreed with the industical data. The result showed that the effect of micromixing on the LCB/SCB reactions was very important.The effects of micromixing on the reaction rate of LCB/SCB in reactor were further analyzed. The LCB reaction was controlled by the kinetic. The SCB reaction was controlled by the turbulent mixing because of the micromixing characteristics in the autoclave reactor. It favored the formation of the SCB reaction at the beginning of each reaction zone, but the LCB reaction at the ending. The reaction rate of LCB increased more sharply than the SCB because of the worse micromixing resulted from the back mixing at the ending of each reaction zone. Therefore, the SCB and LCB reaction rates were of the same order of magnitude in the ending of each reaction zone, favoring the formation of LCB.