Community analysis of bifurcation maps of diluted hydrogen combustion in well stirred reactors

This article focuses on applying an automatic procedure for inspecting detailed kinetic mechanisms to predict key species/reactions playing a dominant role in the dynamic behaviour of regime transitions of combustion systems. We adopted a novel analysis method of kinetic mechanisms built by coupling the bifurcation analysis, which identifies the different regimes occurring with a change of the parameters, with algorithms typical of Artificial Intelligence, specifically the Community Analysis, which identifies within a large set of individuals (the chemical species) common behaviour patterns. Then we showed that this method is capable of automatically extracting meaningful information to identify the key species responsible for the dynamic behaviour and change of regimes observed in the solution maps of a combustion system. Application to a diluted hydrogen combustion system is illustrated as a first validation bench. It reveals that the recognition of state change is effective when based on proper indexes. On the full range of the parameter investigated, the inlet mixture temperature, when adopting an index correlated to the heat release rate, different community partitions were identified for each of the three regions with different dynamic behaviour. The most complex community partitions, with three communities, arise in the first two regions of lower temperature. The partition simplifies in the region of the higher parameter values where the system shows steady solutions without oscillations. The change in the link of atomic hydrogen, OH, and H2O2 with the remaining species is identified in the regime transitions.