Twin-model method for ethanol detection in flex fuel vehicles

In this paper we propose to use volatility difference between gasoline and ethanol to detect the ethanol percent. The approach relies on a well-calibrated model of the wall-wetting effect, a condition satisfied by today´s engine controllers. Because of its complexity, we essentially treat the model as a black box, but with an important difference that the model can be duplicated. Thus, a pair of identical wall wetting (transient fuel) models is driven by different ethanol content and the difference in the resulting predicted air-fuel ratios is used to update the ethanol estimate. It turns out that this volatility-based estimate is very insensitive to sensor and injector biases and drifts. The approach is backed by stability analysis as well as by simulation results and post-processing experimental vehicle data.