Empirical feasibility analysis of thermoelectric energy harvesting in thermally limited compact mobile computers

Khan M. A. A. , Muhtaroglu A.

JOURNAL OF RENEWABLE AND SUSTAINABLE ENERGY, vol.6, no.1, 2014 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 6 Issue: 1
  • Publication Date: 2014
  • Doi Number: 10.1063/1.4865577


Thermoelectric (TE) generation technology was experimentally established in previous research by our group as a viable technique for energy scavenging in a large notebook computer with no significant impact to system performance. The computer under investigation was designed to have additional thermal headroom, with Central Processing Unit (CPU) temperature significantly below its maximum limit under maximum workload conditions. Yet the question remained on if and how such scavenging could be done in small, thermally limited systems, which increasingly represent a larger portion of the contemporary microelectronic products. This paper thus empirically demonstrates the feasibility of thermoelectric energy scavenging in a compact mobile system, where CPU temperature readily reaches the maximum limit as the workload activity is increased. A detailed Finite Element model is presented first for what-if studies. The simulation results from the model are then correlated with the experimental thermal characterization data from a small notebook computer. "Hotspots" as well as the plausible locations for TE integration are identified in the system through the thermal simulations, and are validated by integrating the TE module to the target system. TE power generation density has been measured as 4.27 mW/cm(3) under maximum workload conditions with no impact to system performance, as measured indirectly through cooling fan speed, CPU, and integrated graphics temperatures. For a well-characterized off-the-shelf TE component of size 6.05mm x 6.05mm x 2.09mm, the maximum generated power was 410.5 mu W, 3.5 times more than the corresponding value measured previously in the large notebook system under the same workload. Harvested power is expected to scale with the system workload activity, and the extension of the current solution to the similar opportunistic locations within the system. (C) 2014 AIP Publishing LLC.