Multiobjective Design Optimization of a PEMFC System Using Trust-Region Adaptive Differential Evolution


Yağcı D., Turgut O. E., Turgut M. S., Genceli H.

Fuel Cells, cilt.26, sa.6, 2026 (SCI-Expanded, Scopus)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 26 Sayı: 6
  • Basım Tarihi: 2026
  • Doi Numarası: 10.1002/fuce.70179
  • Dergi Adı: Fuel Cells
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source, Chemical Abstracts Core, Compendex, Greenfile, INSPEC, Engineering Source (EBSCO), Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest)
  • Anahtar Kelimeler: Differential Evolution, multicriteria decision making, proton exchange membrane fuel cell, radiator-fan parasitic load, system efficiency, thermal management, trust-region adaptation
  • İstanbul Gelişim Üniversitesi Adresli: Evet

Özet

This work couples a closure-verified, physics-based 50 kW automotive proton exchange membrane fuel cell model with a novel adaptive optimizer, Trust-region Inspired Differential Evolution with Coordinated Expansion Control (TIDECEC), to address the net electrical power versus system efficiency trade-off. The system energy, heat, and anode-water balances are strictly enforced over the seven-dimensional decision space, with the cell-voltage model calibrated to production-vehicle data. The proposed optimizer improves the L-SHADE framework by adding an individual-dependent geometric trust radius with asymmetric contraction–expansion dynamics, history-driven local-to-global mutation switching, and a targeted bubble restart. The algorithm is tested on the CEC 2014 and CEC 2017 benchmarks at 30 and 50 dimensions, and it achieves a lower mean Friedman rank than 10 state-of-the-art Differential Evolution variants. The multiobjective fuel cell design problem, solved with TIDECEC and benchmarked against NSGA-II, NSGA-III, and MOEA/D, yields a convex Pareto front from 2.80 kW at 56.94% efficiency to 42.47 kW at 29.63% efficiency. The adopted air-supply model charges compressor work only above ambient pressure. The optimizer therefore selects near-ambient cathode pressure and high air stoichiometry. As a result, radiator-fan power accounts for 87% of the 6.92 kW peak parasitic power demand. However, multicriteria decision analysis and Monte Carlo uncertainty propagation show the maximum-power and entropy-weighted points are feasible in just 47% and 69% of samples, respectively, and otherwise violate the 368.15 K catalyst-layer thermal ceiling. Thus, equal-weight TOPSIS (39.38 kW, 34.27% efficiency) and fuzzy Bellman–Zadeh (24.44 kW, 44.38% efficiency) are proposed as robust, fully feasible design-stage targets for rated and part-load duties, respectively.