This article presents an extensive series of laboratory tests carried out to experimentally validate a modified equivalent-circuit model for proton-exchange-membrane (PEM) fuel cells using electrochemical impedance spectroscopy (EIS) under varied operating conditions (pressure 0.5–1.5 bar; current 1–5 A; active area 49 cm2). To robustly capture the high-frequency (HF) inductive “hook”, a compact inductive term (LNET) that de-embeds wiring/instrumentation effects was introduced without altering mid/low-frequency parameters. Across eight operating points, the proposed model reduces the maximum modulus error from 12.3 % (reference) to <2.0 %, and the maximum phase error from 7.0 % to <1.5 % (e.g., at 1 bar–3 A), while preserving the expected trends of transfer/diffusion arcs with pressure and current. The fitted LNET is nearly condition-invariant with a mean of 24.6 μH, supporting its role as a bench/network de-embedding term. The approach improves the identification of the high-frequency resistance and yields more reliable parameter estimates for diagnostics. The results demonstrate that the model is accurate, robust to operating-point changes, and practical for rapid EIS-based analysis of PEMFCs.
Model enhancement and experimental validation of PEM fuel cell performance via electrochemical impedance spectroscopy
Palermo, Federico;Genovese, Matteo
;Corigliano, Orlando;Fragiacomo, Petronilla
2026-01-01
Abstract
This article presents an extensive series of laboratory tests carried out to experimentally validate a modified equivalent-circuit model for proton-exchange-membrane (PEM) fuel cells using electrochemical impedance spectroscopy (EIS) under varied operating conditions (pressure 0.5–1.5 bar; current 1–5 A; active area 49 cm2). To robustly capture the high-frequency (HF) inductive “hook”, a compact inductive term (LNET) that de-embeds wiring/instrumentation effects was introduced without altering mid/low-frequency parameters. Across eight operating points, the proposed model reduces the maximum modulus error from 12.3 % (reference) to <2.0 %, and the maximum phase error from 7.0 % to <1.5 % (e.g., at 1 bar–3 A), while preserving the expected trends of transfer/diffusion arcs with pressure and current. The fitted LNET is nearly condition-invariant with a mean of 24.6 μH, supporting its role as a bench/network de-embedding term. The approach improves the identification of the high-frequency resistance and yields more reliable parameter estimates for diagnostics. The results demonstrate that the model is accurate, robust to operating-point changes, and practical for rapid EIS-based analysis of PEMFCs.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


