Sensitivity analysis of a problem solution allows quantitative understanding of how the solution depends on the various parameters contained in a model. [125] Sensitivity analysis is often an invaluable tool in interpreting the results of stirred reactor [126] and flame experiments [127], [34], [128].

In Ansys Chemkin, we consider the first-order sensitivity coefficients of the gas temperature, species fractions, and, where appropriate, the bulk-phase growth rates, with respect to the reaction rate coefficients. The computationally efficient sensitivity analysis methods exploit the fact that the differential equations describing the sensitivity coefficients are linear, regardless of any non-linearities in the model problem itself. Rate-of-production analysis provides complementary information on the direct contributions of individual reactions to species net production rates.

Sensitivity analysis results are normalized and included in the XML Solution File (for example, XMLdata.zip) for post-processing. In addition, results may be printed to the diagnostic output file. Here we describe the methods used for calculating sensitivity coefficients, for both steady-state and transient solutions, and for normalizing the resulting quantities.