An Integration of the Restructured MELCOR for the MIDAS Computer Code

The developmental need for a localized severe accident analysis code is on the rise. KAERI is developing a severe accident code called MIDAS, which is based on MELCOR. In order to develop the localized code (MIDAS) which simulates a severe accident in a nuclear power plant, the existing data structure is reconstructed for all the packages in MELCOR, which uses pointer variables for data transfer between the packages. During this process, new features in FORTRAN90 such as a dynamic allocation are used for an improved data saving and transferring method. Hence the readability, maintainability and portability of the MIDAS code have been enhanced. After the package-wise restructuring, the newly converted packages are integrated together. Depending on the data usage in the package, two types of packages can be defined: some use their own data within the package (let’s call them independent packages) and the others share their data with other packages (dependent packages). For the independent packages, the integration process is simple to link the already converted packages together. That is, the package-wise structuring does not require further conversion of variables for the integration process. For the dependent packages, extra conversion is necessary to link them together. As the package-wise restructuring converts only the corresponding package’s variables, other variables defined from other packages are not touched and remain as it is. These variables are to be converted into the new types of variables simultaneously as well as the main variables in the corresponding package. Then these dependent packages are ready for integration. In order to check whether the integration process is working well, the results from the integrated version are verified against the package-wise restructured results. Steady state runs and station blackout sequences are tested and the major variables are found to be the same each other. In order to verify the results, the integrated results were compared with the restructured results for each package. Some sequences were calculated such as a steady state and SBO (Station Blackout) accident. The major variables were the same as well as the graph trends. Through out the integrating process, the base was constructed for a code improvement and an addition of new models. The integrating process proposed in this paper will be extended to the T/H and F/P packages for the MIDAS development program.