Could you teach me to settle the interface treatment in multi-block flow problem. I use nonstaggered grid and SIMPLE algorithms.

(1). There is no difference between the single block and the multi-block method. You simply define each block with boundary cells extending into the neighboring block, where the information (variables) from the neighboring block (under different indexing system) can be stored under the current indexing system as boundary conditions. (2). In this way, each block will be self contained and can be solved under its indexing system. So, this overlapping of computational domain along the boundary will automatically transfer the information from block to block (between different indexing systems associated with each block). (3). If the grid is not continuous across the boundary, and if the cell or grid type also changes across the boundary, special treatment is required there. (4). In general, the multi-block method is used to solve the storage (memory) problem with the structured mesh. The idea is similar to the yellow sticker pad, where there is adhesive material on one edge so that you can paste on the regular size page along the border. The yellow sticker is one block, the regular size page is another block, and there is a common overlapping region.

There was some articles on this topic of Lilek, Peric et al in the J. Numerical Heat Transfer