Refractory linings perform an essential role in most of the metallurgical processes, acting as thermal energy barriers and ensuring structural integrity. The refractory selection and lining design must consider the main wear mechanisms, the interaction between the refractory and liquid metals or slags, the costs, and the environmental impacts associated with its production and application. Maintaining the best performance of the materials in use requires good operational practices, the quality of the refractory products during installation, the quality and control of the application practices, and the continuous tracking of each step. In the case of refractory castables, the evaluation of the production and application aspects of the products was carried out by collecting samples during the maintenance/repairing of the linings along with the overhaul of an alumina calciner and comparing the fundamental properties of the castables with the reference values previously obtained during the selecting/development steps. The analysis pointed out which were the expected properties of the refractories, and the differences with reference indicators, indicating potential areas of the calciner that could undergo earlier repair or anticipated failure. The variability of the properties with application features, such as the quantity of water during mixing, time of mixing, temperature, etc, were detected. In conclusion, the study validated and showed the importance of tracking the performance of refractories, fostering the systematization of the methodology for higher productivity and lower costs.

MgO-C refractories are zonally installed in the heating units of steel and iron production. In these conditions, refractory materials are exposed to multiple corrosive factors. One of the main factors which determine the wear rate of MgO-C refractories is interaction with aggressive liquid slags of variable basicity.
In this work, we show the influence of the physicochemical parameters of a few types of MgO-C materials on their corrosion resistance under influence of different temperatures and slag basicity. The designed and industrially produced MgO-C products were characterized by different ratios of the used raw materials and variable forming pressure.
20 laboratory corrosion tests were conducted with the use of an induction furnace. The assessment of corrosion resistance of MgO-C refractories was conducted using a linear method. The input variables used in models were as follows: apparent porosity and bulk density of materials before and after coking, decarburization rate of MgO-C materials after laboratory tests at 900℃ and 1100℃, the chemical composition of designed materials, temperature of the corrosion test and slag basicity. The output variable was the wear rate of the selected MgO-C products after corrosion.
Applying this approach, we developed predictive models which allow for the estimation of the wear rate of MgO-C materials depending on the selected input variables. Predictive models were developed with the use of machine learning techniques: MARSplines (Multivariate Adaptive Regression Splines) and techniques based on decisive trees algorithms.

Masonry refractory linings are commonly employed in Industrial vessels to protect against extreme working environments. Mortarless refractory masonry is frequently used in the working lining of steel ladles to contain the molten steel and limit heat losses. During operation, such masonry lining undergoes high thermo-mechanical loads, primarily due to the complex thermal and mechanical boundary conditions. Therefore, large-scale experiments are required to characterise the behaviour of such structures. This work presents a novel approach for the experimental characterisation of refractory linings using a laboratory scaled pilot steel ladle. This 3D pilot model is a scaled representation of a ladle that includes all the linings as an industrial ladle. The pilot ladle was tested under transient thermal loads, and its behaviour was obtained through thermocouples, strain gauges and DIC (digital image correlation). The experimental observations describe a complex and nonlinear global behaviour of the pilot ladle due to thermal gradient, creep of the refractories and damage. Finally, the results obtained from these experiments were used to validate numerical models using a 3D meso-modelling approach. The numerical simulations revealed a good agreement with the experimental outcome that can further be employed to investigate and design an industrial steel ladle.