(245i) Impact of Thermo-Physical Data on the Modelling of High Pressure Polymerizations

Nowadays modelling is a basic instrument for process optimization. Especially for technologies, which are conducted under exceptional process conditions, such as high pressure polymerizations, simulations are a welcome alternative to costly experiments in miniplants.

The synthesis of Low Density Polyethylene (LDPE) is an example for such an industrial process, which is typically operated between 100°C and 300°C and under pressures up to 3000 bar. The mathematical prediction of process parameters for the LDPE-Synthesis, for instance conversion, molecular weight distribution and branching, has been studied and refined since the early 1970s. By now the process simulation for LDPE synthesis poses a crucial tool for the LDPE-process development and optimization. But for the conduction of reliable and precise process simulations it is essential to have a very profound knowledge and understanding of the process itself, its thermodynamics and kinetics and of course the structure-properties relationship of the resulting polymer. Therefore a uniform set of parameters is essential, which can describe the polymerization reaction independently of reactor type and process. However this proves to be a highly demanding task as numerous parameters have to been known sufficiently precise within a complex reaction network and under the extreme process conditions. This is why variable parameters are still necessary in the current simulation models in order to obtain good agreement between calculation and plant data [1]. For tube reactors these parameters are e.g. the initiator efficiency and the thickness of the fouling layer inside the reactor.

The existence of these variable parameters as well as systematic discrepancies within the reactor models give reason for further investigations. In order to identify and overcome the knowledge gaps a systematic sensitivity analysis of the existing model concerning possible faulty parameters is the reasonable approach. Firstly the fundamentals, such as the thermo-physical properties, which serve as the basis of the polymerization model, should be revisited. For the simulation these are the

• density
• heat capacity
• viscosity and
• heat transfer coefficient

of ethylene, LDPE and the reaction mixture. These data have to be available in dependence on temperature, pressure and composition, respectively. For the calculation of the viscosity molecular weight and branching structure of the polymer can also be taken into account. Whereas the density, the heat capacity and the thermal conductivity primarily influence the heat balance of the model, the viscosity of the reaction mixture also influences the kinetics directly. Reaction steps that depend on polymer mobility, such as the termination reaction of the polymerization, exhibit a direct dependence on the viscosity even at low polymer concentrations.

The literature on these properties for the pure components under process conditions is scarcely available and mostly dates back to the 1950s and 60s. Furthermore it often contains extrapolations and was analyzed with outdated mathematical methods. Experimentally based data on mixture properties is virtually inexistent, thus ideal mixing rules are applied without questioning [2], [3]. Moreover it becomes evident that some of the authors obtain differing values for the thermo-physical properties at a given temperature and pressure. Thereby the discrepancies reach up to several percent.

In the next step the impact of these parameters on the process simulation of high pressure polymerizations was studied and evaluated by the example of an industrial tube reactor.

Effects could mainly be observed in the resulting temperature profile and the achieved conversion. However these effects were negligible for the heat transfer, but conspicuous for the heat capacity and striking for the viscosities. For the heat capacity a constant error was introduced and evaluated. For small errors the adjustment of the variable parameters makes it possible to diminish the deviation caused by the introduced error to a minimum. This proves the concept of the variable parameters and explains why they sometimes exhibit physically unrealistic values.

But LDPE also possesses special rheological properties due to its unique branching structure. Thus the correct prediction of the molecular mass distribution as well as the polymeric microstructure is of striking importance for this polymer as well. Therefore it is also interesting to examine whether the viscosity of the reaction mixture, which directly influences the polymerization kinetics, shows any influence on the calculated molecular mass distribution and polymeric microstructure.

[1] T. Herrmann, Dissertation, Darmstadt 2011.

[2] A. Buchelli, M. L. Call, A. L. Brown, A. Bird, S. Hearn, J. Hannon, Ind. Eng. Chem. Res. 2005, 44, 1474-1479.

[3] G. Luft, R. Steiner, Chemiker-Zeitung 1971, 95 (1), 11-15.