The Development of Methodologies and a Novel Test Facility for the Characterisation of Thermoelectric Generators

Abstract

<p>With the rising prices of energy and the harmful environmental effects many of conventional energy generation techniques the world is pushing for new, cleaner, more efficient and more environmental renewable energy sources. Thermoelectric generators are one of the potential solutions to these problems of unclean and expensive energy. Thermoelectric generators are solid state devices that convert thermal energy into useful electrical energy.</p> <p>Over the last ten years the progress in materials science have led to advancements in thermoelectrics. However as of yet no standardised method of testing thermoelectric generators has been established and as such data provided for thermoelectric generators is regarded as questionable. This thesis deals with two commercial thermoelectric generator models, TEG1 12610-5.1 AND TEG1B 12610-5.1, and quantifies the deviation of the manufacturer’s specifications to what is experimentally achieved by the generators as 147% and 22% respectively. The variance of the outputs between thermoelectric generators was measured by comparing the maximum power output for the models in question over a sample size of four, it was found to be as much as 20% and 8% respectively.</p> <p>A full characterisation of the thermoelectric generators is performed on the two generator models to obtain the data as to their power output and thermal conductivity for the purpose of design of a waste energy harvesting device. The full characterisation was also used to validate the testing apparatus as a device capable for the use as a standardised method of characterising the performance of thermoelectric generation modules.</p> <p>A mechanistic model is created using the experimental characterisation data. This mechanistic model has the ability to accurately predict the voltage and current output of the thermoelectric generator models under any given temperatures and electrical loading condition with a minimum R-squared value of 0.94. The thermal conductivity is also found to be predictable using an established equation modified with an empirical constant.</p>