(37e) Predicting the Combustion Properties of Hydrofluorocarbon (HFC) Refrigerants Using the Automated Reaction Mechanism Generator (RMG)

The halocarbons commonly used today as refrigerants and flame suppressants have high global warming potentials (GWPs), motivating their replacement with more environmentally friendly alternatives. Proposed replacement Hydrofluorocarbon (HFC) refrigerants have lower GWPs due to their shorter atmospheric lifetimes. However, the chemical properties that make these low-GWP HFCs break-up in the atmosphere also promote higher flammability, raising safety concerns. To mitigate this flammability hazard, blends of eco-friendly HFCs are currently being explored [1]. Predicting the combustion properties of new HFC refrigerants and blends over a wide range of conditions will facilitate their innovation, development, and implementation as the next-generation of refrigerant working fluids. To predict the safety-critical properties of next-generation compounds, computational modeling of halocarbon combustion will be vital. However, predicting these properties is challenging because detailed kinetic models often feature thousands of reactions between hundreds of intermediate species with scarce data. Our goal is to enable the open-source Reaction Mechanism Generator (RMG) software [2] to automatically construct detailed kinetic models of halocarbon combustion. Recently, we extended RMG to model Chlorinated Hydrocarbon Combustion and constructed a methyl chloride combustion mechanism [3]. In this work, we expand RMG to model HFC combustion and build a kinetic model of refrigerant R32 (CH2F2).

To teach RMG fluorine chemistry several steps were taken, as summarized here. First, to enable RMG to identify Fluorine, we added Fluorine atom types to RMG’s Molecule class and functional group definitions. Then, to equip RMG to make educated predictions of the thermochemical parameters of fluorinated compounds, we added Benson Group additivity values for closed shell molecules and Hydrogen Bond Increment groups for fluorinated radicals. We also compiled a thermochemistry library of known fluorinated species from reliable literature sources and our own quantum chemistry calculations. To enable RMG to automatically generate fluorinated combustion reactions, we added Fluorine to existing reactions templates (H Abstraction and R Recombination) and added new reaction families specific to Fluorine chemistry (F_Abstraction and F2/HF insertion into a double bond). To improve RMG’s rate estimates for these new reaction families, we calculated kinetic parameters for several reactions using AutoTST [4], an automated transition state theory calculator. With these additions, we tested RMG’s ability to automatically generate accurate detailed kinetic models for HFC combustion by comparing the predicted laminar flame speeds of an RMG-built CH2F2 and air combustion mechanism to experimental data [5] and a published literature model [6].