In the article, the temperature dependence of the viscosity of a complex inorganic substance - potassium carbonate was obtained and the proposed mathematical model was verified. Viscosity is considered as a chaosensitive property of a liquid inherent in it in motion and at rest. The mathematical model of viscosity was developed using the Boltzmann distribution and the concept of chaotic particles. On this basis, a hierarchical cluster-associate viscosity model is constructed, which takes into account not only the formation of primary clusters, but also secondary associates with respect to them, with the possibility of identifying the degree of cluster association. To adapt the cluster-associated model to experimental data, certain data processing techniques have been developed to identify unknown parameters of the model. The method of processing viscosity data using the entire set of three reference points allows you to determine the indicator of the degree of aggregation of associates. When processing the data on the viscosity of potassium carbonate, a high correlation coefficient was established calculated in comparison with reference values, which indicates the adequacy of the new relationship. This model makes it possible to predict the behavior of the viscosity of potassium carbonate in a higher temperature range. The degree of association of clusters with an increase in temperature decreases, corresponding to the dynamics of the destruction of associates and viscosity in general.