Flux coupling analysis is a method to identify blocked and coupled reactions in a metabolic network at steady state. We present a new approach to flux coupling analysis, which uses a minimum set of generators of the steady state flux cone. Our method does not require to reconfigure the network by splitting reversible reactions into forward and backward reactions. By distinguishing different types of reactions (irreversible, pseudo-irreversible, fully reversible), we show that reaction coupling relationships can only hold between certain reaction types. Based on this mathematical analysis, we propose a new algorithm for flux coupling analysis.