Biomedical modelling · Model reduction

Thermoelectric generators for active implants

Compact thermal models for miniaturised devices that harvest temperature differences in human tissue to power medical implants.

Onkar Jadhav · Cheng Dong Yuan · Evgenii Rudnyi · Dennis Hohlfeld · Tamara Bechtold

MSc research · University of Rostock

Research outputs · 2017–2018

Abstract

Electrically active implants for regenerative therapies (e.g. regeneration of bone tissue or deep brain stimulation for the treatment of motion disorders) are gaining on importance within an aging population. The implants must be replaced during the course of the therapy. Their performance requirements vary from a few microwatts to a few milliwatts and will keep increasing with growing functionality. In order to extend the life of electrical active implants and thus avoid the expensive and risky operations for their replacement, a considerable amount of the implant’s energy requirement shall be covered by the conversion of mechanical or thermal body energy into electrical power. In this work, we present a multiphysical model of a miniaturized thermoelectric generator for electrically active implants, which uses temperature gradients in human tissue. Based on this model, we analyze the influence of the geometry and material parameters on the thermal and electrical properties and aim for an optimal transducer design. Furthermore, we use mathematical methods of model order reduction to create an accurate compact model that can be applied within a system simulation.

Introduction

Active medical implants commonly rely on batteries whose finite lifetime may require replacement surgery. The temperature gradient between the body core and skin offers a continuous energy source, but the performance of a miniaturised thermoelectric generator depends on tissue geometry, material properties, blood perfusion, ambient conditions and device design.

Goal
Evaluate body-heat energy harvesting for active implants without repeatedly solving a costly tissue-device model.
Key idea
Reduce a layered thermal tissue and thermoelectric-generator model while retaining nonlinear and parametric effects.
Takeaway
The compact models provide a pathway to efficient environmental, design and coupled device–circuit studies.

Methodology

The physical model represented a thermoelectric generator embedded in layered human tissue containing muscle, fat and skin. Heat transport included tissue conduction, boundary convection and temperature-dependent metabolic heat generation, allowing the available temperature difference across the generator to be estimated.

Several compact-modelling strategies were studied: snapshot-based linearisation of nonlinear heat inputs, parametric model reduction for ambient temperature and skin convection, and reduced representations suitable for repeated parameter studies and coupled device–circuit simulation.

Results and outcomes

Generator design and compact-model validation

The source paper reports a maximum electrical power of 94.5 μW for a thermocouple-leg cross-section of 275 × 275 μm². Its transient thermal model was reduced from 106,467 to 30 degrees of freedom, and the full and reduced temperature histories closely overlap at the TEG and skin surfaces.

Paper plot of electrical power and open-circuit voltage against thermocouple leg edge length
Figure 4. Maximum electrical power is obtained for a leg edge length of 275 μm, smaller dimensions cause an excessive electrical resistance; wider legs will lead to a drop in temperature difference. Source: Jadhav et al. (2017).
Paper plot comparing full and reduced thermal-model temperatures at the top and bottom TEG surfaces and the skin surface
Figure 5. Temperature result comparison between full and reduced model at top surface, bottom surface and skin surface of the TEG model. Source: Jadhav et al. (2017).

System-level power output

In the reported co-simulation, the load resistance was varied from 1 Ω to 100 Ω. The calculated TEG voltage was 84.72 mV, with the resistance sweep identifying the load region that maximised delivered power.

Paper plot of power dissipated in the electrical load as a function of logarithmic load resistance
Figure 8. Power dissipated in load resistance for different load resistance values. Source: Jadhav et al. (2017).

The paper’s results are presented here through its original figures and accompanying reported values; no additional results table has been constructed.

Technology and research setting

The work grew from MSc research at the University of Rostock and used finite-element thermal models, compact modelling, nonlinear and parametric model-order reduction, biomedical heat transfer and thermoelectric energy harvesting.

BibTeX

@inproceedings{jadhav2017design,
  author    = {Jadhav, Onkar Sandip and Yuan, Cheng Dong and Hohlfeld, Dennis and Bechtold, Tamara},
  title     = {Design of a thermoelectric generator for electrical active implants},
  booktitle = {MikroSystemTechnik 2017},
  year      = {2017},
  pages     = {402--405},
  publisher = {VDE Verlag},
  url       = {https://ieeexplore.ieee.org/document/8278689}
}

Research outputs