Modelling of a thermal cracking process involves establishing a set of kinetic reactions that transform the feed into products. In a typical feed, there exists a large number of real components. Thus, an exact feed composition is not known. The most common method to overcome this difficulty is to define pseudocomponents based on physical properties such as boiling point or molecular weight. In this work, we propose a set of model thermal cracking reactions based on nuclear magnetic resonance (NMR) data. NMR carbon type analysis improves the characterization of the feed and products by providing the contents of different carbon species (carbon bonds). This additional information enables a more descriptive and fundamental set of reactions to be developed for the model. The reactions chosen describe the types of carbon bonds that break and form in the feedstock under visbreaking conditions. The set of reactions is the centre of the pseudo-component model.