In the context of code coupling, efficient data redistribution is a crucial issue to reach high-performances. However, most of the works in this area have limited their studies to simple objects with regular distribution patterns, such as multidimensional arrays distributed in a block-cyclic fashion. In order to deal with modern code-coupling applications (e.g. multiphysics simulations, computational steering environments), it becomes necessary to define a description model for more complex objects than multidimensional arrays (e.g. structured grids, particle sets or unstructured meshes) and to propose new redistribution algorithms based upon this model. In this paper, we present two approaches for data redistribution: the spatial approach and the placement approach. Finally, these algorithms are implemented in the RedGRID library and are validated by experimental results. MOTS-CL