The design of a modular tensegrity grid requires the determination of its selfstress states, before choosing an appropriate combination defining the system's initial stresses. However, the computation of the vectorial basis associated with selfstress states generally produces results that are difficult to exploit. We therefore propose two different strategies to identify and localize selfstress states in a modular tensegrity grid more pertinently. The first is based on a heuristic approach that exploits the system's structural composition of modularity and regularity. The second is numerical and aims at redefining the vectors of the basis in a more convenient and useful way. Two methods based on transformations of the vectorial basis of selfstress states have been developed for a minimal number of involved components. Finally, we suggest a selfstress state classification based on the number of components and their localization as well as on their mechanical behavior.