Document IIF

Surrogate-guided coupled optimisation unit for thermal energy conversion.

Résumé

Coupled thermal energy conversion systems, combining solid-state or fluidic coolers with generators, can recover net positive electrical power from low-grade waste heat. Evaluating such coupled configurations requires matching interface temperatures and heat fluxes between devices while searching across material properties and operating conditions. TCCbuilder, a physics-based simulation engine, addresses this by modelling coupled device performance and sweeping candidate interface temperatures at discrete steps. However, TCCbuilder relies on exhaustive enumeration of all operating points, which scales poorly as the number of device configurations and material combinations increases, and cannot suggest new configurations beyond those already simulated. This paper presents Surrogate-guided Coupled Optimisation Unit for Thermal Energy Conversion (SCOUT), a device-agnostic optimisation framework that wraps TCCbuilder's models with Bayesian optimisation using Gaussian process surrogates. SCOUT replaces brute-force enumeration with an intelligent, machine learning model-guided search that identifies optimal coupled operating conditions using significantly fewer simulation evaluations. On a synthetic benchmark coupling an electrocaloric cooler with a pyroelectric generator, SCOUT recovered 99.5% of the exhaustive-search optimum (Pnet = 1.53 W) using only 35 evaluations compared to 5,625 for grid search, a 160-fold reduction in computational cost. Moreover, SCOUT generalises to any cooler-generator pairing supported by TCCbuilder, including thermoacoustic devices. The framework provides a scalable pathway for optimising coupled thermal energy conversion systems as the material database expands.

Documents disponibles

Format PDF

Pages : 10 p.

Disponible

Gratuit

Détails

  • Titre original : Surrogate-guided coupled optimisation unit for thermal energy conversion.
  • Identifiant de la fiche : 30035071
  • Langues : Anglais
  • Sujet : Technologie
  • Source : 11th IIR Conference on Solid-State Cooling, Heating and Energy Harvesting.
  • Date d'édition : 07/06/2026
  • DOI : http://dx.doi.org/10.18462/iir.thermag.2026.0022

Liens


Voir d'autres communications du même compte rendu (29)
Voir le compte rendu de la conférence