Abstract: The goal of the OURAGAN project is to provide access of meta-computing resources to Scilab users. We present here an approach that consists, given a Scilab script, in scheduling and executing this script on an heterogeneous cluster of machines. One of the most effective scheduling technique is called clustering which consists in grouping tasks on virtual processors (clusters) and then mapping clusters onto real processors. In this paper, we study and apply the clustering technique for heterogeneous systems. We present a clustering algorithm called triplet, study its performance and compare it to the HEFT algorithm. We show that triplet has good characteristics and outperforms HEFT in most of the cases.