Potassium is an essential ion for plant nutrition, usually supplied in the form of chlorides and sulfates. According to Brazilian availability and demand of agriculture fertilizers, the importation of compounds carrying this chemical element is mandatory in order to fulfill the huge demand for this nutrient. So initiatives looking for alternative sources of potassium become increasingly interesting and economically attractive. A potential route is associated with the sulfuric digestion of glauconite-bearing greensands and sequential unit operations in order to recover aluminum, iron, magnesium and potassium compounds. In the context of this chemical process, the potassium alum dodecahydrate appears as a relevant intermediate product that allows the selective recovery of potassium and aluminum through thermal decomposition followed by solubilization in water and filtration. Based on what was said, the present work investigates the kinetics of potassium alum dodecahydrate decomposition under nonreductive and reductive conditions, and a novel mathematical model is proposed to describe the weight loss during time. A stochastic approach approach, using particle swarm optimization method, is employed to estimate the unknown model parameters. The model predictions are validated by experimental data obtained through dynamic thermogravimetric analysis at different reaction atmospheres (inert and oxidant), and with the presence or not of reducing agent (metallurgical coke breeze). With the validated model parameters, it is possible to use them to monitor the mass compositions of all compounds present in the process as well as to use the model for future online monitoring since its simulation takes less than 1 s to simulate 20 min of decomposition thermal.