Rather than try to model fuel combustion based on the long list of molecules the fuel contains, the researchers found a shorthand: they now show that they can distill the complexity into a very short list of molecular subunits or functional groups most fuel molecules are made from. This radically simplified method for accurately simulating fuel combustion was developed by Abdul Gani Abdul Jameel under the guidance of Mani Sarathy and his team.

The project began with the hypothesis that the combustion behavior of each component of a fuel is dictated by the functional groups it comprises. To corroborate the theory, the team performed high-resolution nuclear magnetic resonance analysis in KAUST’s Core Labs to identify the main functional groups in a series of complex fuels. They then made simple surrogates for each fuel by selecting one or two molecules that contained the functional groups in the same balance as the real fuel.

Comparing key combustion parameters, such as ignition delay time and smoke point in the lab, the researchers confirmed the simple surrogates were faithful mimics of the real fuel. They showed a good surrogate needed to match the average molecular weight and contain the right proportions of just five key carbon-hydrogen functional groups: CH3, paraffinic CH2, paraffinic CH, naphthenic CH–CH2 and aromatic C–CH.

Traditional combustion modeling accurately captures the behavior of fuel mixtures by adding detailed chemical kinetics data for increasingly more of the components in the fuel, but the drawback is that the simulation becomes prohibitively slow to run. “We have shown that adding complexity to models is not necessary, as long as underlying features of simpler molecular parameters, the functional groups, are captured,” Sarathy says.

The team’s method for making simple fuel surrogates will directly improve the design of efficient new engines, explains Abdul Jameel. “Using a minimal number of components significantly reduces the time involved in developing chemical kinetic models and the computational expenses involved in simulating combustion in internal combustion engines,” he says.

But the team’s functional-group-based approach will reach far beyond surrogate formulation. “We are currently developing machine-learning-based models to predict the combustion properties of fuels based on their functional groups,” Abdul Jameel adds.