Reaction network reduction for distributed systems by model training in lumped reactors: Application to bifurcations...

Area of Science:

• Chemical Engineering
• Combustion Science
• Computational Fluid Dynamics

Background:

• Developing accurate reduced reaction mechanisms is crucial for simulating complex reacting flows.
• Distributed reacting flows present challenges due to intricate transport-chemistry coupling.
• Existing methods often struggle to efficiently capture these coupled phenomena.

Purpose of the Study:

• To present a novel methodology for deriving reduced reaction mechanisms for distributed reacting flows.
• To train these mechanisms using a lumped parameter system, specifically a continuous-stirred tank reactor.
• To ensure the derived mechanisms accurately represent transport-chemistry coupling in distributed systems.

Main Methods:

• Model training within a lumped parameter system (continuous-stirred tank reactor).
• Identification of relevant transport time scales and local composition vectors across operating conditions.
• Sensitivity and principal component analyses at bifurcation points within a defined parameter space (pressure-transport time scale-composition).

Main Results:

• A training box in the parameter space was identified for effective model training.
• The derived reduced chemistry from the lumped system effectively captured transport-chemistry coupling.
• The methodology was successfully applied and validated for hydrogen/air and methane/air ignition in premixed and diffusion flames.

Conclusions:

• The presented methodology provides an effective approach to derive reduced reaction mechanisms for distributed reacting flows.
• Training in a lumped parameter system offers a viable strategy for capturing complex transport-chemistry interactions.
• The validated application demonstrates the utility of this method for combustion simulations.