Abstract
A new, fast and simple numerical method is proposed for modelling data of thermogravimetric analysis under arbitrary temperature–time relationships. The algorithm searches for activation energies and rate constants by means of minimisation of the average square of deviation, Δ, between computed and experimental curves on a scale of the logarithm of reduced time that, in turn, is expressed as the integral of the Arrhenius exponential. The algorithm tests phenomenological relations by considering the process mechanism. Eighteen known models corresponding to different physical and chemical processes are included as the basic set in the algorithm. Sequential analysis of the 18 variants and arrangement of values of Δ1/2 in ascending order allow a selection of the best of models. The less Δ1/2 is, the nearer is the calculated activation energy to the correct value. Then one can detect that satisfactory models provide a good approximation of the original kinetic curves.