SIMULATION AND EXPERIMENTS OF POLYETHYLENE PYROLYSIS IN A FLUIDIZED BED PROCESS
4. Kinetic simulation
Using temperatures below 600 °C for the pyrolysis of polyethylene, nearly no secondary side reactions happen. Under these conditions it is possible by kinetic constants and fl u-idized gas simulation to simulate the molar mass distribution of the formed waxy prod-ucts.
Figure 1: Benzene formation by the pyrolysis of PE in dependence of temperature and fl ow gas rate.
Benzene concentration in wt%; throughput rate of PE = 14,1 kg/h.
To verify the simulated data, test runs were carried out by 510 °C in the laboratory fl uid-ized bed plant. The waxy products were analyzed by Field Desorption Mass Spectrometry (Fig. 2).
Each signal group is characterized by six peaks for the α-ω-dienes, α-alkenes, and al-kanes and their isomers containing one 13C-isotop. It could be shown that the high-boil-ing wax fraction contains hydrocarbon products with chain lengths between 15 and 70 carbon atoms.
Table 1: Products in wt% from the pyrolysis of PE in a fl uidized bed under different conditions.
Temperature (°C) 530 700 740
Fluidized gas N2 steam circulated pyrolysis gas
Products (wt%) wax olefi ns aromatics
Hydrogene 0,1 0,5 1,2
Methane 0,8 11 20
Ethene 2,0 31 19
Propene 1,8 14 4,3
Butene/Butadiene 1,1 8,4 1,5
Other gases 0,7 8,1 5,0
C5-C20 Aliphatics 16 3,7 3,5
Bencene 0,1 9,4 24
Toluene - 3,2 6
Styrene 0,1 1,8 1,6
Other aromatics 0,1 6,6 12
Waxy products 76,0 0,1
-Soot 1,2 2,2 1,9
Figure 2: Field desorptions mass spectrum of pyrolysis products of PE in a laboratory fl uidized bed by 500 °C; molecules with a 13C-atom are marked with *
The simulation of the PE pyrolysis in the fl uidized bed reactor is based on several partial models: A mechanistical reaction model, based on the Rice- Kossiakoff mechanism [11]
describes the chemical pyrolysis reactions.
Pi → Pk + Pi-k
with i = 2,3,4…n and k = 1,2,3…i-1
The fl uidized bed reactor is modelled according to the two-phase model of Werther [12].
Additionally to these two models, a suitable rate law for the evaporation of low-molecular hydrocarbons was established in this study. The solution of the models required differ-ent approaches: To solve the reaction model there is usually set up a differdiffer-ential equation system that has to be integrated for the time t. Instead of that, in this investigation the stochastic procedure of Gillespie [13,14] (Monte Carlo procedure) has been used. The original procedure was supplemented by routines for the handling of the molar mass dis-tribution.
For the solution of the reactor, model a fi nite difference method was applied to integrate the differential equation system for the fl uid mechanics sof the fl uidized bed reactor as sit was developed by Bruhns [15] for a fl uidized bed drying process. With slight modifi ca-tions, this algorithm could be applied to PE pyrolysis in the fl uidized bed. The evapora-tion of the volatile hydrocarbons in this way generally accepted model concepevapora-tions were combined, partially extended and by simulation verifi ed with measured data.
The most important result of this research is the realization that the molar mass distribu-tion of the products is primarily controlled by the evaporadistribu-tion of low-molecular species to approx. C100.
In the reversal conclusion, the molar mass distribution is an important measure for the determination of the evaporation rate. Without a suitable evaporation mode, the mecha-nistical modelling of polyethylene pyrolysis fails. In this context it became evident that a rate law of fi rst order with kEvap.(i) as the evaporation rate constant is
(Ni: number of polymers with the chain length i NG: number of all polymer chains) unsuitable for the description of the evaporation kinetics, because this way the composi-tion of the evaporacomposi-tion stream directly depends on NG, the number of simulated mol-ecules in the liquid phase.
In order to keep the composition of the evaporation stream independent from the sample extent, the evaporation coeffi cient DEvap. (i) was introduced, and the quotient of DEvap(i)
[ ] ∑ [ ]
and NG was used instead of the constant kEvap. ·f(i) was the fact that different functions f could be compared easily this way, so that the rate law takes the form:
With this empirical rate law, the composition of the evaporation stream becomes exclu-sively dependent on D0,Evap.. Different compositions of the evaporation stream could be simulated by the variation of D0,Evap. It became evident that values of D0,Evap. In the order of magnitude of 103 – 104 s-1 (Fig. 3) are best suitable to describe the experimental results.
D0,Evap. Values greater than 104 s-1 result in a too high a portion of high-boiling compo-nents in the evaporation stream, values smaller than 103s-1 result in a predominance of the portion of low-molecular components. In this context, the form of the function f (i) which differ particularly in the low-molecular range, cause only slightly different product compositions.
( )
i ,
i 0
· ·N dt
dN
G Evap
N i f
= D
Figure 3: Mass spectrum simulation of the pyrolysis products of PE by 510 °C with an evaporation coeffi cient for the waxy products of D0,Evap = 5 · 103 s-1.
Furthermore, the simulation shows that conversion in the gaseous phase is neglegibly small at temperatures of 500 °C. The reason is that the absolute concentrations in the gaseous phase are too low to cause a considerable propagation reaction. Therefore, under the – for a pyrolysis – mild conditions examined here primary cracking almost exclusively takes place in the melt and secondary cracking hardly occurs in the gaseous phase.
The portions of low-molecular products up to C11 which are higher than predicted by the simulation, are caused by non-statistical reactions in primary cracking which were ne-glected in the reaction model.
As far as it was examined, the simulated radical concentration in the polymer melt was nearly constant. This fact fortifi ed the quasi-stationary state assumption.
This study emphasized the enormous infl uence the evaporation has on the composition of volatile hydrocarbons formed by pyrolysis of polyethylene.
References
[1] Kaminsky, W. and Sinn, H., in J. Brandrup et al., eds., Recycling and Recovery of Plastics, Hanser, New York, 1996, p. 435.
[2] Arena, U., and Mastellone, M.L., Chemical Engineering Science 55: 2849 (2000).
[3] Bockhorn, H., Hentschel, J., Hornung, A., and Hornung, U., Chem. Eng. Sci. 54: 3043 (1999).
[4] Williams, E.A., and Williams, P., J. Anal. App. Pyrolysis 40-41: 347 (1997).
[5] Mastellone, M.L., Perugini, F., Ponte, M., and Arena, U., Polym. Degrad. Stab. 76:
479 (2002).
[6] Mastral, F.J., Esperanza, E., Berrulco, C., Juste, M., and Ceamanos, J., J. Anal. App.
Pyrolysis 70: 1 (2003).
[7] Luo, G., Suto, S., Yasu, S., and Kato, K., Polym. Degrad. Stab. 70: 97 (2000).
[8] Kaminsky, W., Schlesselmann, B., and Simon, C.M., Polym. Degrad. Stab. 53: 189 (1996).
[9] Kaminsky, W., Predel, M., and Sadiki, A., Polym. Degrad. Stab. 85: 1045 (2004).
[10] Kaminsky, W., and Rössler, H., Chemtech 2: 108 (1992).
[11] Kossiakoff, A., and Rice, F.O., J. Am. Chem. Soc. 65: 590 (1943).
[12] Werther, J., Chem. Eng. Sci. 35: 372 (1980).
[13] Gillespie, D.T., J. Phys. Chem. 81: 2340 (1977).
[14] Gillespie, D.T., J. Computational Phys. 22: 403 (1976).
[15] Bruhns, S., Thesis, Technical University Hamburg-Harburg, 2002.