Operating conditions of thermal units for processing raw materials predetermine defects in refractory elements resulting in their gradual accumulation, which leads to a change in technical condition. A large number of defects, their development, and the achievement of critical values lead to difficulties in modeling the physical processes of changing the technical condition of refractory elements.This study has investigated the mechanism of the occurrence, development, and accumulation of defects in refractory elements, as well as the processes of cumulative accumulation of damages; a probability model of their degradation has been constructed. The model was built using Markov chains; it describes the sequences of change in the states of refractory element damage and the probability of transitions between these states. Based on the statistical data about a change in the state of damage, the model makes it possible to assess the probability of a defect reaching the critical condition following the predefined number of load cycles. A special feature of the model is the possibility of its application to individual defects, as well as to refractory elements on which defects occur and develop, as well as to assemblies where such refractory elements are installed.The main patterns of change in the technical condition of refractory elements of coke ovens have been established: the distribution of cracks of a certain length according to the number of coke oven output cycles; the probability of the occurrence of a crack of a critical length at a certain point during operation; the dependence of the probability of a refractory element failure on the predefined number of coke oven output cycles.Based on the modeling results, it has been proposed, in order to prevent the degradation of refractory elements, to strengthen the structure of the surface layer of the refractory element by cold gas-dynamic spraying, to arrange laying elements that would stop the evolution of defects, and to make up schedules of hot repairs based on the time when the defects may reach critical values, determined during modeling