This numerical study deals with the design of experiments. It aims at optimizing thermocouple positions in order to reduce uncertainties in the estimation of thermophysical properties when an inverse method is applied. The 2D system under investigation is a squared sample of orthotropic material submitted to a constant heat flux on left and bottom edges. The temperature response is given by three sensors. The unknown parameters are the volumetric thermal capacity and the conductivities in the two principal directions. The experiment design is based on an original optimality criterion. Two stochastic algorithms are used to find its minimum and their efficiency is compared to a pure random search algorithm. To deal with the fact that the experiment design depends on the unknowns, a robust optimization approach is proposed. (C) 2012 Elsevier Ltd. All rights reserved.