Numerical precision of minimum residence time calculations for glass tanks Recalculation of round robin test 1 of technical committee 21 "modelling of glass melts" of the International Commission on Glass (ICG)
Résumé
Prediction of glass is a main objective of glass furnace modeling. Minimum residence time (MRT) counts among the crucial parameters for representing glass quality. Numerical simulation is an efficient tool for estimating this parameter. For a Round Robin a simple case had been chosen with a parallepiped-shaped glass tank to test the performance of numerical codes and their application in glass melting simulation. It was shown that the numerical discretization fixed by the participants is sufficient to represent the temperature and velocity fields. However, significant differences in the calculation of the MRT (6.2 to 7.8 h) between the different participants of the Round Robin were noticed. In this work it is shown that the grid resolution and the number of tracer particles have to be increased at certain locations to capture correctly the shortest path-time trajectory. The final value of 5.5 h for the MRT does not significantly change with further refinement of the grid and of trajectory calculation. The two principal methods known for the calculation of massless virtual particle trajectories are numerical tracing and flow-field stream lines. One established method to check the reliability of stream line calculations is the successive refinement of the numerical resolution. However, for large glass tanks, this method can be quite tedious or even unfeasible. Simplified methods are proposed to check the reliability of MRT calculation.