For the estimation of kinetic parameters of a complex reaction system from experimental data, a non-linear least square method has been applied so far to treat all of the data. However, the procedure sometimes brings about numerous troubles, and needs a long computation time with fruitless iteration.
In this paper, the authors present a new statistical treatment to elucidate the reaction scheme and to evaluate their kinetic parameters.
In this treatment the overall rates of conversion of a key reactant and of the production of an appropriate final product are first analyzed statistically and their lumped rate equations and kinetic parameters are determined.
Next, a more detailed reaction scheme is assumed ; the reaction paths between the reactant, the final products, and the remains (all the other products treated as one group), and the rate equations and kinetic parameters of the individual reaction paths are estimated. Then an intermediate product is separated from the remains and a more refined reaction scheme is established.
These procedures are repeated until all of the detailed reaction scheme of the system and the kinetic parameters are definitely fixed.
In this treatment, all the reaction rates are assumed to be of the Freundlich type for convenience in statistical procedure.
This statistical treatment is less time-consuming than the former method, and is more advantageous in discovering and removing the insignificant reaction paths and negligible dependences on components in rate equations.
The detailed steps and techniques are discussed and an application to the experimental data of
p-xylene oxidation is described as an example.
View full abstract