Reliable models of known accuracy are indispensable for the application of model based techniques. In case of multicomponent mixtures such as crude oil or hydrocarbon fractions, suitable reduction techniques have to be applied to obtain models of appropriate complexity. A novel approach for the reduced modeling of multicomponent mixtures is presented, originating from the combination of continuous thermodynamics with a Wavelet Galerkin discretization technique. It is this combination that overcomes the deficiencies usually associated with continuous thermodynamics. The novel approach combines high efficiency with control of the local and global approximation error.