<div class="csl-bib-body">
<div class="csl-entry">Román, V., Valtiner, M., Asouti, V., & Vernes, A. (2026). AI-driven identification of key roughness-related descriptors for electrical contact resistance. <i>Materials & Design</i>, <i>272</i>, Article 117313. https://doi.org/10.1016/j.matdes.2026.117313</div>
</div>
-
dc.identifier.issn
0264-1275
-
dc.identifier.uri
http://hdl.handle.net/20.500.12708/232195
-
dc.description.abstract
This paper investigates the effect of surface roughness parameters on the electrical contact resistance (ECR) in fuel cells. Despite extensive efforts, the effect of roughness on ECR remains poorly understood, mainly because roughness is commonly only characterized by the arithmetic mean height. This work proposes a comprehensive set of parameters spanning amplitude, lateral spatial distribution, and summit density metrics to investigate their influence on ECR at various external pressures. Numerically generated rough surfaces are brought into contact with compliant counter-surfaces. Random Forest regression quantifies their pressure-dependent importance, and SHAP analysis reveals the direction and magnitude of their effects. Three pressure-dependent regimes, each dominated by a different descriptor, are identified. At the onset of contact (0.1 MPa), the solidity factor, which characterizes the balance between peaks and valleys, dominates. Between 0.1 and 2.2 MPa, the autocorrelation length dictates the dependence of ECR on pressure, while above 2.2 MPa, ECR saturates and is controlled by the summit density. These results provide a unified interpretation of previously conflicting studies and demonstrate the necessity of multi-descriptor roughness characterization to understand its impact on ECR.
en
dc.description.sponsorship
European Commission
-
dc.language.iso
en
-
dc.publisher
ELSEVIER SCI LTD
-
dc.relation.ispartof
Materials & Design
-
dc.rights.uri
http://creativecommons.org/licenses/by/4.0/
-
dc.subject
Electrical contact resistance
en
dc.subject
Machine learning
en
dc.subject
Proton exchange membrane fuel cells
en
dc.subject
Rough surfaces
en
dc.title
AI-driven identification of key roughness-related descriptors for electrical contact resistance