Access Official Technical PDF

Share Summary:LinkedInEmail

Thermoelectric waste heat recovery offers industrial operators a way to convert some otherwise unused heat into electricity. A study of a hybrid thermoelectric generator (TEG) examined this approach using heat from combusted syngas produced from de-oiled Pongamia pinnata seed cake. Its results highlight how heat transfer, cold-side cooling and performance forecasting affect the practical assessment of small-scale power generation.

The work combined laboratory testing with an artificial neural network (ANN) model. Experiments covered hot-side temperatures from 200°C to 250°C, while the model was used to estimate performance at higher temperature differences. The findings are relevant to engineers evaluating thermal energy recovery, though the higher-temperature results are predictions rather than experimental measurements.

From Pongamia seed-cake waste to usable heat

After oil is extracted from Pongamia seeds, the remaining de-oiled cake presents a disposal challenge because it is not suitable for animal feed or fertilizer. The study investigates gasifying this biomass in a fluidized bed reactor at 700°C to produce syngas, a mixture that includes hydrogen and carbon monoxide, along with other gases. Combusting the gas provides a heat source that can be applied to a thermoelectric system.

For the laboratory tests, an electrical heater reproduced the heat input rather than supplying heat from syngas combustion directly. The heater warmed Therminol-55 synthetic oil held in a hollow copper block on the TEG’s hot side. The oil served as the heat-transfer fluid, while the research team tested two ways of removing heat from the cold side: forced water cooling and natural-convection air cooling.

How the hybrid thermoelectric generator was tested

The tested module combined bismuth telluride (Bi2Te3) and lead telluride (PbTe) semiconductor materials. A temperature difference across the module produces an electrical output. To measure its behavior, the researchers varied the external load resistance and recorded voltage and current, allowing them to determine output power. They also assessed open-circuit voltage and the matched load resistance—the load value corresponding to the generator’s internal resistance and maximum power output.

The hot-side temperature was varied between 200°C and 250°C. For both cooling approaches, the coolant or ambient cold-side temperature was maintained at about 30°C. In the water-cooled configuration, water flowed through a copper heat exchanger at 0.5 litres per minute. The air-cooled configuration used a finned copper block and natural convection.

Cooling method influences electrical output

At a hot-side temperature of 250°C, the reported maximum output was 6.51 W with water cooling and 5.06 W with air cooling. The cold-side temperature was approximately 50°C in the water-cooled test and 70°C in the air-cooled test. In this setup, forced convection transferred heat away more effectively, maintaining a lower cold-side temperature and a larger temperature difference across the TEG.

This comparison illustrates a design consideration for energy recovery: generator output depends not only on the heat source, but also on how effectively the cold side rejects heat. Cooling-system choice should therefore be evaluated alongside available temperature, installation constraints and the intended electrical load. The reported figures describe this particular module and test arrangement; they should not be treated as guaranteed output for other systems.

The generator’s output also depends on electrical loading. Maximum power occurs around the matched load resistance, so a system intended to use the recovered electricity needs an appropriate electrical interface and load strategy. The study’s measurements across a range of resistances helped identify this operating point for each test condition.

Using ANN models to estimate TEG performance

Experimental testing can be difficult to interpret because measured performance is affected by factors such as heat loss, insulation, thermal contact resistance and temperature measurement. The researchers used a feedforward, backpropagation ANN trained on sample experimental data to estimate open-circuit voltage, maximum power and matched load resistance.

For the tested range, ANN estimates were close to experimental results. The model was then used to project performance at larger temperature differences, including up to 320°C for water cooling and 300°C for air cooling. These projections extend beyond the laboratory test range. They can support preliminary performance assessment, but they are not substitutes for validation under the operating conditions of a proposed installation.

Interpreting the cost estimate

The paper reports a simplified electricity cost estimate of $0.0018/kWh for the biomass gasification–thermoelectric system. This figure is based on the study’s stated assumptions, including treating the de-oiled seed cake as a zero-cost fuel because it is a waste material. It should be interpreted as a study-specific estimate, not a general cost benchmark. Project economics would depend on the actual cost of feedstock handling, gasification equipment, the TEG system and other site-specific requirements.

Practical implications for energy recovery

The study demonstrates a pathway for combining biomass waste management with thermoelectric power generation: gasify the seed cake, combust the resulting syngas, and use the heat to drive a hybrid TEG. For industrial automation and energy teams, its most useful lessons are that cold-side heat rejection materially affects output, electrical loading must be matched to the generator, and ANN forecasts should be distinguished from measured results.

Thermoelectric waste heat recovery may be worth assessing where a suitable heat source and cooling arrangement are available. Before adopting the concept, engineers should validate thermal conditions, electrical performance and economic assumptions at the intended site. The reported experiment provides a technical basis for that evaluation, while also showing why modeled performance beyond tested conditions requires further verification.

Access Official Technical PDF

About The Author