We develop a simplified one-dimensional numerical model that simulates the performance of thermoelectric generators (TEG). The model is based on the energy and electrical potential field equations. The Seebeck coefficient, thermal conductivity, electrical resistivity and Thomson coefficient of the TEG material are used to predict the harvested power. Bismuth-telluride is used as semiconductors materials of the TEG, which is the most commonly used material by industry. Experiments on three TEG modules were performed to validate the numerical model. A comparison with predicted levels of harvested energy based on the TEG specifications is also performed. The results show differences between the experimental and numerical values on one hand and the predicted ones on the other hand. The reason for these differences are discussed. A procedure to estimate the sensitivity of the harvested power to different inputs and TEG parameters is detailed. In the second part of the dissertation, we integrate a thermoelectric generator with an organic storage device. The performance of the integrated system for different values of load resistances and temperature gradients is determined. Finally, we demonstrate that power generated from a TEG is related to the flow rate in a pipe and can, thus, be used as a flow meter. Particularly, a dimensionless relation between the TEG's peak power and Reynolds number is determined.Ph. D.