Abstract:In this work, a modeling and experimental study of a new thermoelectric cooler-thermoelectric generator (TEC-TEG) module is investigated. The studied module is composed of TEC, TEG and total system heatsink, all connected thermally in series. An input voltage (1-5 V) passes through the TEC where the electrons by means of Peltier effect entrain the heat from the upper side of the module to the lower one creating temperature difference; TEG plays the role of a partial heatsink for the TEC by transferring this waste heat to the total system heatsink and converting an amount of this heat into electricity by a phenomenon called Seebeck effect, of the thermoelectric modules. The performance of the TEG as partial heatsink of TEC at different input voltages is demonstrated theoretically using the modeling software COMSOL Multiphysics. Moreover, the experiment validates the simulation result which smooths the path for a new manufacturing thermoelectric cascade model for the cooling and the immediate electric power generation.

No abstract

There are two major types of substructure mode synthesis methods, i.e., the fixed-interface component mode synthesis and freeinterface component mode synthesis. There are two coupling methods, the interface degrees of freedom based coupling method and the interface force based coupling method, the former one is referred to as the primary assembly method, and the latter is referred to as the dual assembly method. However, the dual assembly method is theoretically shown to be unstable in this research, such reduced stiffness matrix is indefinite, this fatal weakness can be conquered by further interface reduction, and the interface compatibility is therefore rigorously enforced. Unfortunately, Craig's method leads to another numerical instability when inverting a submatrix of residual flexibility on the interface degrees of freedom, this problem is neglectable in small dimensional matrix problems, but it is prominent in large models when the number of interface degrees of freedom is large, this ill-conditioning problem may be circumvented by truncated singular value decomposition technique; here, a more efficient strategy is proposed, the substructure reduction is modified, this modification proves to be numerically stable, and it can be even more accurate than the prevailing Craig-Bampton method; the numerical examples validate the suggestion.

No abstract