Devyanshi Patel
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Devyanshi Patel has a total of 3 co-authors. Here are the top co-authors by number of shared publications:
- Bimal Das
School of Engineering and Applied Science Ahmedabad University Ahmedabad Gujarat India
1 shared publication - Dharmesh Rizwani
Center of Excellence for Composite, Ahmedabad Textile Industry's Research Association Ahmedabad Gujarat India
1 shared publication - Partha Bairi
Polymer Science Unit, Indian Association for the Cultivation of Science, Kolkata-700 032, India
1 shared publication
Latest Publications
- 5 November 2026
- Published by Wiley in Polymer Composites
Fatigue Life Characterization of Graphene‐Enhanced GFRP Composites by Experiment and Physics‐Guided Neural Network Modeling Using Nano‐Material‐Modified Composites
Improving the fatigue life of advanced composites and accurately predicting their fatigue life are crucial because composites are increasingly used in applications involving repeated loading, including automotive chassis, wind turbine blades, aircraft structures, and bridges. Graphene‐modified glass fiber‐reinforced polymer (GFRP) composites were fabricated using the vacuum resin infusion method. The fatigue life of graphene‐modified GFRP laminates was improved by about three to five times by modifying the epoxy resin with 0.5% graphene, compared to GFRP without graphene. Moreover, the fatigue strength coefficient and ultimate tensile strength of graphene‐modified GFRP were enhanced by about 38% and 27% compared to neat GFRP. Fatigue life prediction in composites, particularly nanomaterial‐reinforced systems, is challenging due to pronounced anisotropy and progressive damage evolution governed by reinforcement type and configuration. Moreover, the limited availability of fatigue data for nanomaterial‐reinforced composites presents challenges for conventional data‐driven prediction. Therefore, a novel physics‐guided neural network (PGNN) framework was proposed under small sample conditions for accurate fatigue‐life prediction of graphene‐modified GFRP composites. The PGNN exhibited superior predictive performance relative to the purely data‐driven model, achieving an R 2 of 0.9804 compared with 0.8900, while the RMSE decreased from 0.3449 to 0.1453 and the MAE from 0.0860 to 0.0747 on the log‐transformed scale.
- 5 November 2026
- Published by Wiley in Polymer Composites
Fatigue Life Characterization of Graphene‐Enhanced GFRP Composites by Experiment and Physics‐Guided Neural Network Modeling Using Nano‐Material‐Modified Composites
Improving the fatigue life of advanced composites and accurately predicting their fatigue life are crucial because composites are increasingly used in applications involving repeated loading, including automotive chassis, wind turbine blades, aircraft structures, and bridges. Graphene‐modified glass fiber‐reinforced polymer (GFRP) composites were fabricated using the vacuum resin infusion method. The fatigue life of graphene‐modified GFRP laminates was improved by about three to five times by modifying the epoxy resin with 0.5% graphene, compared to GFRP without graphene. Moreover, the fatigue strength coefficient and ultimate tensile strength of graphene‐modified GFRP were enhanced by about 38% and 27% compared to neat GFRP. Fatigue life prediction in composites, particularly nanomaterial‐reinforced systems, is challenging due to pronounced anisotropy and progressive damage evolution governed by reinforcement type and configuration. Moreover, the limited availability of fatigue data for nanomaterial‐reinforced composites presents challenges for conventional data‐driven prediction. Therefore, a novel physics‐guided neural network (PGNN) framework was proposed under small sample conditions for accurate fatigue‐life prediction of graphene‐modified GFRP composites. The PGNN exhibited superior predictive performance relative to the purely data‐driven model, achieving an R 2 of 0.9804 compared with 0.8900, while the RMSE decreased from 0.3449 to 0.1453 and the MAE from 0.0860 to 0.0747 on the log‐transformed scale.