Fuel sorption into polymers: Experimental and machine learning studies

In the automotive industry, introduction of alternative fuels in market or even consideration new fluids such as lubricants requires continuous efforts research and development to predict evaluate impacts on materials (e.g., polymers) contact with these fluids. We address here compatibility between polymers by means both experimental modelling techniques. Three were considered: a nitrile butadiene rubber (NBR), fluorinated elastomer (FKM) fluorosilicon (FVMQ), series hydrocarbons mixtures formulated study swelling polymers. The samples has been investigated terms weight not volume variations measure this former is assumed be more accurate. Multi-gene genetic programming (MGGP) was applied data obtained order derive models predict: (i) maximum value mass gain (ΔM) (ii) sorption kinetics, i.e. time evolution ΔM. Predicted values are excellent agreement (with R2 greater than 0.99), have demonstrated their predictive capabilities when external (not considered during training procedure). Combining experiments modelling, proposed work, leads accurate which drastically reduce necessary quantify polymeric fluid candidates compared experiments.