Paper presented to the Faraday Society Informal Discussion on Computer Modelling of Complex Reaction Systems, Thornton Research Centre, Chester
S F Bush, ICI Central Instrument Research Lab.
Summary
The chlorinolysis and cracking reactions which occur in the system C2Cl4 – CCl4 – Cl2 are important in several processes of industrial importance. The paper presents a mechanistic model of the system, including surface parameters, for which experimental data has been obtained at one atmosphere over the temperature range 300-700 oC. The experimental results have been obtained in three different types of apparatus, batch, pulse-flow and continuous-stirred. It is shown that the model is in agreement with results from all three types of apparatus which in itself affords some confirmation of its general applicability. A critical comparison of the three methods is given.
An important factor in the behaviour of the reaction system is the nature of the surface. At temperatures below 500 oC the main initiation and termination reactions are found to occur at the surface, so that in batch or pulse flow systems the results of changes in the activity of the surface during an experiment cannot be distinguished from the progress of the reaction in the vapour-phase. In the stirred-flow system, the variation of the surface activity with time for given reactant compositions can be accurately monitored. When a steady-state is achieved the rate of reaction is then obtained as a function of a single composition, pressure and temperature. By comparison a sequence of points is usually needed to define reaction rates in plug flow, pulse or batch systems.
For these reasons it is concluded the stirred flow system is usually to be preferred in cases where, for gases, (a) reaction times between 0.05 and 20 seconds are to be examined and (b) steady flows of the reactants can be maintained and measured. It is to be noted that the reaction time range quoted and the condition of infinite aging at any given reactant composition are directly applicable to most vapour-phase processes of industrial interest.
The reaction model is assembled partly from reaction steps which have been described previously in the literature, and partly from reaction steps which appear relevant on theoretical grounds and which have been found necessary to describe the experimental results. A distinction is drawn between changes in the structure of the model, i.e. the addition or subtraction of reaction steps, and changes in the parameter settings, chiefly rate constant values. To facilitate the exploration of different model structures, use has been made of a generalised model program developed for the purpose. The program requires as data, specification of a reaction mechanism in standard chemical form. It will then derive the corresponding differential or algebraic equations and solve for compositions.
A major problem in constructing models composed of many reactions, the rates of which are known only approximately or not all, is the danger of obtaining a model fit to the experimental data which has little predictive value, and which is a poor representation of chemical reality. The procedure described in the paper greatly reduces this danger by starting from the minimum structure which will reproduce in bare outline the known facts of the system. Only when parameter changes have been shown to be incapable of giving accurate predictions is an extension of the chemical structure permitted. Thus chemical knowledge, numerical exploration and experimental data are used in an iterative fashion to converge on to a realistic model of a complex system. The construction of the model in this way is further illustrated by the calculation of the formation rate of by-product hexachlorobutadiene at high temperatures.
See also the section on Chemical Rate Processes.