Research On Parallel Calculation And Visualization For Heat Transfer Of Slab/Mould In Continuous Casting

The mould is the core component of continuous caster for primary cooling and initial shell forming of liquid steel.During the processing of steel continuous casting,heat flux between mould copper plate and slab has significant effects on the slab's quality.Finding out the characteristic of mould heat transfer and the solidification of slab exactly in real-time is very important for forecasting the stick breakout and slab's quality on-line.The "mould expert system" being used provides some preliminary visualization function such as t "thermal imaging" technology,which can only give the temperature nephogram of mould copper plate by using simply interpolation algorithm rather than some important information like distribution of mould temperature,heat flux and solidified thickness of slab on-line.The mould visualization has not realized yet in fact.Lacking of suitable mathematical model combining the on-line monitoring with heat transfer simulation and the conflict between "accuracy" of numerical simulation and "real-time property" of on-line monitoring are the restrictions of development and application of mould visualization and intellectualization technology.Therefore the real-time computing and visualization of mould heat transfer based on temperature measurement are mainly studied in this paper by using the slab/mould as research object.In this paper,a mathematical model of inverse problem for heat transfer between slab and mould is established based on the temperature measurement of mould in slab continuous casting.Heat flux of measuring point and its distribution are determined by reducing the difference-value between measured values and calculated ones of thermocouples in mould copper plate by using the iterative method,which decreases the deviation of the heat transfer calculation which is using the experiential formula as convenient conditions.The information of mould heat transfer and solidification of strand which can reflect actual working conditions is provided.But inverse-problem simulation makes computing time increase greatly,which can not meet the need for on-line monitoring.A computer cluster is built by using normal PC and switch.The parallel programming platform is designed by using MPI and Visual C++ 6.0.The serial algorithm for numerical simulation of heat transfer between slab and mould is parallelized according to the heat transfer characteristic of slab/mould,and then the parallel computing program is generated and tested on the cluster.The calculated results indicate:under the experiment conditions,the computing time can be reduced deeply to 3.9s when selecting four computing nodes and 300×10 grids on our cluster,the speedup and parallel efficiency can be 3.47 and 86.75%,which provides a possibility for heat transfer simulation of mould on-line.The visualization environment is developed by using the OpenGL technology and Visual C++ 6.0.A plotting mode called "method of descending dimension" is designed through the linear relationship between physical values and colors.The mainly realizing way of "method of descending dimension" is drawing six 2-D faces of the model in turn and connecting them. The visualization software for mould heat transfer and solidification of strand is generated finally,multiple physical fields like mould temperature,heat flux and thickness of slab are showed through 2-D and 3-D colored nephogram.The visualization technology is a convenient learning method for the "black box" of mould heat transfer under high temperature and harsh working conditions,and also provides an effective way for adjusting the continuous casting production on-line.