Thermoelectric (TE) energy harvesting is a promising perspective to use waste heat. Due to the low efficiency of thermoelectric materials many analytical and numerical optimization studies have been developed. To be validated, an optimization must necessarily be linked to the experience. There are a lot of results on thermoelectric generators (TEG) based on experiments or model validations. Nevertheless, the validated models concern most of the time one TE module but rarely an entire system. Moreover, these models of complete system mainly concern the optimization of fluid flow rates or of heat exchangers. Our choice is to optimize the number of these modules in a whole system point of view. A numerical model using a software for numerical computation, based on multi-physics equations such as heat transfer, fluid mechanics and thermoelectricity was developed to predict both thermal and electrical powers of TEG. This paper aims to present the experimental validation of this model and shows interesting experimental results on the location of the TE modules. In parallel, an experimental set-up was built to compare and validate this model. This set-up is composed of a thermal loop with a hot gas source, a cold fluid, a hot fin exchanger, a cold tubular exchanger and thermoelectric modules. The number and the place of these modules can be changed to study different configurations. A specific maximum power point tracker DC/DC converter charging a battery is added in order to study the electrical power produced by the TEG. The analysis of the influence of the number of thermoelectric modules and influence of electric currents on the produced electrical power was investigated. Different operating points of hot inlet gas airflow rate and of cold inlet source temperature were tested. Both experimental and numerical results show the necessity to optimize the position, the number of TE modules and the electrical currents. (C) 2015 Elsevier Ltd. All rights reserved.