This paper employs valency descriptors to investigate their predictive potential for determining thermodynamic properties of benzenoid hydrocarbons. It is customary to choose the heat capacity (Cp) and the entropy (So) to be the representatives of thermodynamic properties. First we put forward a theoretical (resp. computer-based computational) technique for hexagonal systems (resp. general chemical graphs) for computing valency descriptors. The theoretical technique is employed in the comparative testing of valency descriptors for measuring Cp and So of lower benzenoid hydrocarbons (BHs). Our statistical experiments showcase some unexpected outcomes as the best descriptors for correlating physcochemical properties perform rather poorly for the case of thermodynamic properties. Moreover, some under-researched descriptors such as the general sum-connectivity and the general Randić descriptors perform exceptionally well. The well-performing valency descriptors are computed for some families of silicon carbide nanotubes which could be utilized to correlate their thermodynamic properties.