Digital Twin Assisted Intelligent Prediction of Polymer Composite Degradation Under Environmental Exposure

Notice

This is an unedited manuscript accepted for publication and provided as an Article in Press for early access at the author’s request. The article will undergo copyediting, typesetting, and galley proof review before final publication. Please be aware that errors may be identified during production that could affect the content. All legal disclaimers of the journal apply.

Year : 2026 | Volume : 14 | 04 | Page :
By

Nilesh V. Ingale,

Mahesh T. Dhande,

Pankaj Deshmukh,

Ritesh S. Fegade,

Ashwini L. Patil,

Bhausaheb Varpe,

Vithoba Tale,

Rupendra Nehete,

  1. Associate Professor, Department of Computer Science &Engineering (AIML), KCE Society’s College of Engineering and Management, Jalgaon, Maharashtra, India
  2. Assistant Professor, Department of Artificial Intelligence and Data Science Engineering, Matoshri College of Engineering & Research Centre, Nashik, Maharashtra, India
  3. Assistant Professor, Department of Artificial Intelligence and Data Science Engineering, MET, BKC, Nashik, Maharashtra, India
  4. Associate Professor, Department of Mechanical Engineering, Parvatibai Genba Moze College of Engineering, Pune, Maharashtra, India
  5. Assistant Professor, Department of Information Technology Engineering, PVGCOE & SSDIOM, Nashik, Maharashtra, India
  6. Assistant Professor, Department of Mechanical Engineering, Amrutvahini College of Engineering, Sangamner, Maharashtra, India
  7. Associate Professor, Department of Mechanical Engineering, Rajarshi Shahu College of Engineering, Tathawade, Pune, Maharashtra, India
  8. Professor, Department of Mechanical Engineering, Indira College of Engineering & Management, Pune, Maharashtra, India

Abstract

Polymer matrix composites (PMCs) deployed in aerospace, marine, automotive, and renewable-energy structures are continuously subjected to coupled environmental stressors — ultraviolet (UV) radiation, moisture ingress, thermal cycling, and mechanical loading — that progressively degrade their mechanical performance. Conventional accelerated ageing tests and empirical lifetime models are time-consuming, destructive, and poorly suited to in-service, asset-specific degradation forecasting. This paper proposes a Digital Twin (DT) assisted intelligent prediction framework that fuses a physics-based degradation kernel (coupled Arrhenius reaction-rate and Fickian moisture-diffusion equations) with a physics-informed Long Short-Term Memory (PI-LSTM) network to continuously estimate and forecast the residual mechanical strength of polymer composites under multi-factor environmental exposure. The digital twin ingests streaming sensor data (temperature, relative humidity, UV dosage, strain) from the physical asset, updates its internal state through a Bayesian fusion layer, and issues short- and medium-horizon degradation forecasts with quantified uncertainty. The proposed architecture, mathematical formulation, and experimental protocol are described in detail. To demonstrate the workings of the framework prior to full-scale physical validation, illustrative synthetic exposure data were generated and used to benchmark the proposed PI-LSTM digital twin against four baseline predictors (linear regression, support vector regression, random forest, and a physics-agnostic LSTM). On this illustrative dataset the proposed model achieved the lowest root-mean-square error and mean absolute error among the compared methods, indicating the potential of embedding physical degradation kinetics inside a learning-based digital twin.  To make the physical basis of the framework explicit, the paper also details the degradation mechanisms — thermal ageing, moisture absorption, hydrolysis, oxidation, photo-oxidation, chain scission, secondary cross-linking, plasticisation, fibre–matrix interfacial degradation, microcracking, debonding and delamination — through which temperature, humidity, ultraviolet radiation, oxygen and chemical exposure alter tensile strength, tensile modulus, elongation at break and impact resistance, and specifies how these variables are encoded as inputs, and the corresponding property-retention ratios as outputs, of the neural-network predictor embedded in the digital twin.The paper concludes with a discussion of the framework’s practical implications, current limitations, and a roadmap for validation against real accelerated-ageing and field-exposure datasets.

Keywords: digital twin; polymer composite; degradation prediction; environmental exposure; physics-informed neural network; LSTM; structural health monitoring; accelerated ageing; hygrothermal ageing; predictive maintenance; degradation mechanisms; mechanical property retention; neural network prediction; IoT-based monitoring.

How to cite this article: Nilesh V. Ingale, Mahesh T. Dhande, Pankaj Deshmukh, Ritesh S. Fegade, Ashwini L. Patil, Bhausaheb Varpe, Vithoba Tale, Rupendra Nehete. Digital Twin Assisted Intelligent Prediction of Polymer Composite Degradation Under Environmental Exposure. Journal of Polymer & Composites. 2026; 14(04):-.
How to cite this URL: Nilesh V. Ingale, Mahesh T. Dhande, Pankaj Deshmukh, Ritesh S. Fegade, Ashwini L. Patil, Bhausaheb Varpe, Vithoba Tale, Rupendra Nehete. Digital Twin Assisted Intelligent Prediction of Polymer Composite Degradation Under Environmental Exposure. Journal of Polymer & Composites. 2026; 14(04):-. Available from: https://journals.stmjournals.com/jopc/article=2026/view=253369

References

  1. Yang Y, et al. Machine learning-driven digital twin for strength prediction of dissimilar adhesive joints under environmental aging. J Compos Mater. 2025. doi:10.1177/00219983251342147.
  2. Karimi M. Adaptive digital twin framework for monitoring and predicting the performance of composite adhesive joints. J Adhesion. 2025. doi:10.1080/00218464.2025.2535687.
  3. Aysa, et al. Integrating machine learning and digital twin for strength prediction of CFRP/aluminum adhesive joints under hygrothermal conditions. Polym Compos. 2025. doi:10.1002/pc.29928.
  4. Integrating machine learning and digital twin in additive manufacturing of polymeric-based materials and products. Prog Addit Manuf. 2025. doi:10.1007/s40964-025-01257-4.
  5. Towards the development of multiscale digital twins for fiber-reinforced composite materials using machine learning [preprint]. ResearchGate; 2026.
  6. Integrating machine learning and simulation for composite damage detection within a digital twin framework. In: AIAA SciTech Forum; 2025. doi:10.2514/6.2025-1803.
  7. Digital-twin-enhanced quality prediction for composite materials. ScienceDirect. 2023.
  8. Machine learning-enabled multiscale modeling platform for damage sensing digital twin in piezoelectric composite structures. Sci Rep. 2025. doi:10.1038/s41598-025-91196-5.
  9. Kulkarni MV, et al. Data-driven digital twin model for real-time strength estimation in polymeric materials. J Polymer Compos. 2026;14(3).
  10. Pires R, et al. Data generation and deep neural network predictions for aged mechanical properties. Polym Eng Sci. 2025. doi:10.1002/pen.27196.
  11. Artificial neural network approach for assessing mechanical properties and impact performance of natural-fiber composites exposed to UV radiation. Polymers (Basel). 2024. PMCID:PMC10892044.
  12. Physics-informed neural network-based prediction of multi-factor coupled thermal-oxidative aging behavior in polyamide66-glass fiber composites. Chin J Polym Sci. 2026. doi:10.1007/s10118-025-3509-1.
  13. Nezafatkhah S, Margoto OH, Sassani F, Milani AS. Understanding natural and accelerated weathering degradation mechanisms of glass and natural fiber composites: a review. J Compos Mater. 2026. doi:10.1177/07316844251337240.
  14. Machine learning-driven paradigm for polymer aging lifetime prediction: integrating multi-mechanism coupling and cross-scale modeling. Polymers (Basel). 2025;17(22):2991.
  15. Das PP. A multimodal prediction framework for moisture aging assessment in polymer matrix composites [dissertation]. Arlington (TX): University of Texas at Arlington; 2025.
  16. Effects of accelerated weathering on degradation behavior of basalt fiber reinforced polymer nanocomposites. 2020. PMCID:PMC7694794.
  17. Integrating wide and deep neural networks with squeeze-and-excitation blocks for multi-target property prediction in additively manufactured fiber reinforced composites [preprint]. arXiv. 2025. arXiv:2512.22397.
  18. Grieves M. Digital twin: manufacturing excellence through virtual factory replication [white paper]. 2014.
  19. Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput. 1997;9(8):1735-1780.
  20. Palanisamy S, Murugesan TM, Palaniappan M, Santulli C, Ayrilmis N. Use of hemp waste for the development of mycelium-grown matrix biocomposites: a concise bibliographic review. BioResources. 2023;18(4):8771-8780. doi:10.15376/biores.18.4.Palanisamy.
  21. Ravichandran G, Ramasamy K, Manickaraj K, Kalidas S, Jayamani M, Mausam K, et al. Effect of Sal wood and Babool sawdust fillers on the mechanical properties of snake grass fiber-reinforced polyester composites. BioResources. 2025;20(4):8674-8694. doi:10.15376/biores.20.4.8674-8694.
  22. Kar A, Saikia D, Palanisamy S, Pandiarajan N. Effect of fiber loading on the mechanical, morphological, and dynamic mechanical characteristics of Calamus tenuis fiber reinforced epoxy composites. J Vinyl Addit Technol. 2025;31(1):224-240. doi:10.1002/vnl.22167.
  23. Pandiarajan P, Baskaran PG, Palanisamy S, Karuppusamy M, Marimuthu K, Rajan A, et al. Enhancing polyester composites with nano Aristida hystrix fibers: mechanical and microstructural insights. BioResources. 2025;20(4):9257-9281. doi:10.15376/biores.20.4.9257-9281.
  24. Karuppusamy M, Kalidas S, Palanisamy S, Nataraj K, Nandagopal RK, Natarajan R, et al. Real-time monitoring in polymer composites: Internet of things integration for enhanced performance and sustainability—a review. BioResources. 2025;20(3):8093-8118. doi:10.15376/biores.20.3.Karuppusamy.
  25. Govindarajan PR, Shanmugavel R, Subramanian K, Palanisamy S, Santulli C, Fragassa C. Effect of stacking sequence on mechanical and water absorption characteristics of jute/banana/basalt fabric aluminium fibre laminates with diamond microexpanded mesh. Int J Polym Sci. 2024;2024:3835788. doi:10.1155/2024/3835788.
Ahead of Print Subscription Original Research
Volume 14
04
Received 23/07/2026
Accepted 20/08/2026
Published 26/08/2026
Publication Time 34 Days

Login

My IP

PlumX Metrics