Sensitivity Analysis in Chemical Kinetics

Abstract

Complex mathematical models are increasingly being used as predictive tools and as aids for understanding the processes underlying observed chemical phenomena. The parameters appearing in these models, which may include rate constants, activation energies, thermodynamic constants, transport coefficients, initial conditions, and operating conditions, are seldom known to high precision. Thus, the predictions or conclusions of modeling endeavors are usually subject to uncertainty. Furthermore, regardless of uncertainty questions, there is always the overriding matter of which parameters control laboratory observations. Quantification of the role of the parameters in the model predictions is the traditional realm of sensitivity analysis. A significant amount of current research is directed at conceptualization and implementation of numerical techniques for determining parametric sensitivities for algebraic, differential, and partial differential equation models including those with stochastic character and nonconstant parameters. Recent studies have also served to extend the range of the conventional parametric analysis to address new questions, relevant to the process of model building and interpretation. This review attempts to present a full range of results and applications of sensitivity analysis, relevant to chemical kinetic modeling. We exclude a large class of literature that deals with system sensitivities from a control theory perspective (1). We further limit discussion of related subjects such as parameter identification (estimation of best parameter values for fitting data) and optimization, in which parametric sensitivities play only