OCR Text |
Show 3.3.5. Grid generator if-Diss has a full automatic grid generator which utilizes the genetic algorithm optimization as well as conventional manual methods. The genetic algorithm (GA) is an engineering model simulating a biological evolution process. The process is expressed by iteration of four operations, selection of parents, crossover, mutation and natural selection. The algorithm is applied to the mesh division of each axis. The user has to specify the total number of numerical cells, the maximum ratio of the width of consecutive two cells and a few parameters for G A . Users can select conventional manual methods for grid generation as well. Manual methods include mesh division by specifying the number of subdivision and geometric progressions for each block, mouse handling, and typing. Whichever method is selected, generated grid configuration is shown on the display allowing the user to manually modify it. 4. TEST RUNS OF if-Diss Among various test runs conducted to validate if-Diss, two test runs will be presented here. One was a benchmark test that validated the output files pre-processed by if-Diss by solving an identical problem with three different solvers. The other validated the accuracy of models by comparing simulated result with experiment[3]. 4.1. CFD solvers As a benchmark test to validate the output files generated by if-Diss, a forge furnace was simulated with the three C F D solvers. The results were compared in terms of flow and temperature distributions and heat balance. Figure 6 shows the simulated forge furnace, equipped with five pairs of FDI (Fuel Direct Inject) regenerative burners[4] with the total fuel input of 1.7MW. Figure 7 shows simplified analytical model. Cylindrical shape of steel billets was deformed to a rectangular solid. Five burners along one of the side walls are assumed to be in the firing mode simultaneously while five burners along the other side walls in the flue mode. Input conditions are shown in Table 1. Identical input conditions and numerical grids for flow calculation were made for all the three solvers. The number of numerical cells was 20,286 (49x23><18). Radiation grids and the methods of calculating radiative heat transfer were not exactly the same because of difference in the solvers option. For differencing schemes, default schemes specific to the solvers were automatically employed. Table 2 shows numerical conditions for the solvers employed in the benchmark test. Figures 8 (a)-(c) show vector and temperature distribution on the cross section shown in Figure 7. Care must be taken that the number and positions of vectors were not identical, since different post-processors (drawing software) were used. Figures 8 (a)-(c) suggest that in general the distribution of flow pattern and temperature were very close among the results from three solvers. Table 3 shows heat distributions as percentage to the total thermal input. Total thermal input includes the fuel input and the heat of the preheating air. The results of the three solvers agree very well considering the difference in the methods for radiative heat transfer. The results demonstrated that if-Diss has successfully generated mutually appropriate input files for all the three solvers. 5 |