Use this functionality to optimize any kinetics mechanisms containing either elementary or global reaction steps. Targets for optimization can be provided as a full mechanism using the reactor model. Alternatively, user data can also be used as targets. Targets can include any species concentration, such as CO and NO, and any observable quantity, such as ignition times or laminar flame speeds.
Optimization is performed for reaction rate coefficients, including Arrhenius parameters. For a pressure-dependent formulation, Arrhenius parameters for low- and high-pressure limits are considered for optimization. The first parameters in the Troe formulation are also considered. However, other types of formulations, including PLOG and CHEB (see Pressure-dependent Reactions in the Chemkin Theory Manual), are not optimized.
When optimizing a global mechanism as opposed to a detailed mechanism, as specified on the Operation Setup tab, all the reactions are considered for optimization. We do not recommend using this option for a mechanism involving more than 20 reactions as the compute cost to find an optimum set of parameters increases exponentially. For all mechanisms other than global mechanisms, reactions to be optimized are identified automatically based on rate and sensitivity analysis. This analysis is performed for the starting mechanism that is used in the initial Chemkin project.
Tip: The initial compute cost for these analyses can be fairly high. Therefore, we recommend using optimization for mechanisms involving 500 species or fewer.
Note that reactions involving small hydrocarbons can be preserved. To accomplish this, set the maximum number of carbon atoms in a species for those reactions that must be excluded from optimization.
Global optimization using the Genetic algorithm is performed first followed by local optimization using Powell’s Direction Set method. These are the same methods used in Surrogate Blend Optimization described in Surrogate Blend Optimization Methods. In general, the Genetic algorithm requires many function evaluations to find an optimum solution. On the Optimizer Setup tab, you can limit the number of function evaluations to restrict the computation time for this method. The best set of parameters is then used as a starting point for the local optimization. The Direction Set method typically is more efficient and requires fewer function evaluations than the Genetic algorithm.
Mechanism reduction with optimization can be performed in an automated way. The implementation of the optimization method for Mechanism Reduction is similar to that described for Global Mechanism Optimization (see Global Mechanism Optimization. Reactions to be optimized are chosen automatically based on rate and sensitivity analysis. Only Powell’s Direction Set method is used for the optimization. Ranges of uncertainty and other solver settings for optimization are obtained from the Preferences (from the menu, select , then ). When the target error from an optimized mechanism satisfies the criteria set in the Preferences (the default error is 25%), further optimization is aborted. Mechanism reduction then continues with the next method specified in the project setup.
Mechanism reduction with optimization can potentially lead to a significantly smaller kinetics mechanism. However, the resulting mechanism may significantly alter the rate parameters from the full mechanism. Therefore, the applicability range of the reduced plus optimized mechanism will be significantly limited for the range of conditions used for such operations.