Segregated modeling of continuous heat treatment furnaces

Abstract

Heat treatment processes have a major impact on the mechanical and structural properties of the end products. Accurate control of the material temperatures during the heating and cooling regimes is very crucial for the quality of a given production. However, especially in continuous heat treatment furnaces the products inside the furnace are rarely in thermal equilibrium with the furnace and monitoring the air temperature inside the furnace provides a very indirect information about the solid temperatures of the products. In this study, the solid temperatures of the products inside a continuous glassware annealing furnace model is solved numerically. The continuous furnace model is divided into heating and cooling sections each filled with rows of goblets and they are treated separately using a segregated modeling approach. In this approach, the convective heat transfer inside the furnace is modelled using a steady-state convection solver in stationary frame of reference. The transient heat conduction inside the moving goblets is calculated using a separate transient heat conduction solver in moving frame of reference. Thermal radiation exchange between the surfaces is treated using a new backward Monte Carlo based surface-to-surface radiation model and the calculated radiative heat fluxes are added as heat flux boundary conditions on the goblet outer walls. Similarly, the convective heat fluxes calculated with the convection solver are also imposed as heat flux boundary conditions. This iterative solution approach showed a fast convergence behavior requiring only 4 iterations to converge both for the heating and cooling sections of the furnace. The overall computational cost of the simulation is measured as 10 h and 20 h for the heating and cooling sections, respectively. Among all the three heat transfer modes, convection is found to be by far computationally the most expensive, followed by the thermal radiation and conduction being the computationally least expensive one. Overall, the approach enables to conduct high fidelity analysis of the heat treatment processes with acceptable computational cost.