Modeling of coal liquefaction kinetics based on reactions in continuous mixtures. Part II: Comparison with experiments on catalyzed and uncatalyzed liquefaction of coals of different rank

Abstract

Experimental kinetic data for three Australian coals are compared with predictions from the mathematical model developed in Part I. For these coals, as well as data reported for North American coals, the model is found to show good agreement, using the characteristic molecular weight as the only parameter. The effects of reaction time and temperature are coupled via a severity index that arises in the model as a dimensionless reaction time. The model is also shown to be applicable when catalysts are present. The product distribution for a particular coal seems to be a function only of the conversion even in the presence of catalysts, as long as the chemical reaction controls the rate of liquefaction. SCOPEConventional coal liquefaction kinetic modeling is based on the grouping together of products into several "lumps," usually determined by the solubility properties of the various species present. The reaction networks for such a lumped-product system depend clearly on the particular coal and it is not possible to generalize such models. In Part I, a model based on the concept of reactions in a continuous mixture was proposed and a methodology of applying such a model was developed.In this paper, the product distributions obtained upon liquefaction of three Australian coals-a low-rank Victorian brown coal and two subbituminous coals-are examined and compared with model predictions using the model developed in Part I and the lumping functions proposed therein. The kinetic data were obtained from two different types of reactors, a tubular flow reactor and a batch microautoclave reactor. All experiments were conducted under conditions where there was excess availability of hydrogen from donor solvent (tetralin). In addition, other reported kinetic data for North American coals were examined. The effect of catalyst addition (mostly iron-based), is also examined and the applicability of the model is evaluated. CONCLUSIONS AND SIGNIFICANCEThe model proposed in Part I is found adequate in predicting the product distribution for all coals examined. The only parameter varied between different coals is the characteristic molecular weight for coal.The molecular weight for coal broadly depends upon the rank, and increases as the rank increases.The model provides a basis for coupling the effects of reaction time and temperature using the concept of severity index. The implication for liquefaction of coal is that there is no optimum temperature for maximizing Correspondence concerning this paper should be addressed to T. Sridhar. the yield of a particular product as long as the solvent is capable of supplying all the hydrogen requirements. J . 9. Agnew is presently at the Department of Chemical Engineering, University of Adelaide, Adelaide, 5001. Australia. 1288August 1986 Vol. 32, No. 8 AIChE JournalThe product distribution is uniquely determined by reaction severity. This implies that the activation energy is similar for all reactions in the network. The first-order rate constant for the formation of tetrahydrofuran (THF) s...

Fundamental studies of retrograde reactions in direct liquefaction

“…We decided to simulate some literatu_'edata which included both time resolution and mixing. Good agreement was obtained with the data of Prasad et al (1986) for two Australian coals which are similar to Illinois No. 6 and with the data of Foster et al (1985) on Liddell coal (3imilar to Pittsburgh No.…”

Fundamental studies of retrograde reactions in direct liquefaction. Final report, September 20, 1988--November 20, 1990

“…lt was decided to simulate some literature data which included both time resolution and mixing. Good agreement was obtained with the data of Prasad et al (1986) for two Australi._ncoals which are similar to Illinois No. 6.…”

Effect of Petroleum Feedstock and Reaction Conditions on the Structure of Coal-Petroleum Co-Cokes and Heat-Treated Products

“…A reaction severity index (RSI) was used to compare product yield distributions on common basis, and it serves to couple the effect of reaction temperature and time into a single variable to examine the effect on co-coke yield. , The RSI was calculated using the following expression: RSI = 2 × reaction time (h) + (temperature (°C) – 450 °C). This is the approach that was proposed and used by Petrakis et al Figure shows the variation in the co-coke yield as a function of RSI.…”