Junhe Cui
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Junhe Cui has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Jacob Fish
Department of Civil Engineering and Engineering Mechanics Columbia University New York New York USA
7 shared publications - Jaan-Willem Simon
Computational Applied Mechanics University of Wuppertal Wuppertal Germany
3 shared publications Anna Kantert
2 shared publicationsJacob Fish
2 shared publicationsMichele Valsecchi
2 shared publicationsTiansheng Liu
2 shared publications
Latest Publications
- 19 October 2026
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 464
Eigenstate-based homogenization for inelastic heterogeneous media with imperfect interfaces
- 19 October 2026
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 464
Eigenstate-based homogenization for inelastic heterogeneous media with imperfect interfaces
- Open Access
- 16 July 2026
- Published by Elsevier in SSRN Electronic Journal
Eigenstate-based homogenization for inelastic heterogeneous media with imperfect interfaces
This work extends eigenstate-based homogenization (EBH), previously formulated for perfectly bonded interfaces, to general heterogeneous composites with imperfect, damageable interfaces. The resulting framework captures coupled bulk inelasticity and interfacial damage evolution within a unified EBH formulation. A key feature of the proposed framework is its minimal representation of the fine-scale RVE solution: both bulk and interface nonlinearities are described by scalar eigenstates associated with precomputed coarse eigenmodes. Built on transformation field analysis, the formulation accurately recovers both homogenized and fine-scale fields using far fewer unknowns than the full finite element discretization.The central computational obstacle in extending EBH to imperfect interfaces is the severe ill-conditioning induced by penalty-based cohesive zone models. This poor conditioning makes monolithic fixed-point iterations, which are highly efficient for bulk inelasticity alone, prohibitively slow once interfacial damage initiates. To overcome this challenge, we develop a block FP–DR scheme: the well-conditioned bulk inelasticity problem is solved by fixed-point (FP) iterations, whereas the ill-conditioned interface damage problem is solved by Newton iterations combined with LU decomposition for the dense linearized systems of equations, herein referred to as the DR solver.Convergence studies are conducted for both monolithic and block solution schemes. The results show that convergence of the monolithic FP solver depends strongly on the interface regularization and is no longer guaranteed once a critical damage level is reached. In contrast, although the convergence rate of the block EBH schemes remains affected by the maximum interface damage and the interface regularization, these quantities no longer constitute necessary conditions for convergence.Exploiting the rapid initial convergence observed in monolithic FP solvers for bulk inelasticity, we further introduce a loosely converged block EBH variant that achieves significant computational speedup without sacrificing solution accuracy.Numerical studies on fiber-composite and polycrystalline rock RVEs demonstrate that the proposed EBH framework, particularly the block schemes, delivers high-fidelity multiscale predictions for diverse interface-dominated problems at a small fraction of the computational cost of full finite element analysis.
- 16 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEigenstate-based homogenization for inelastic heterogeneous media with imperfect interfaces
This work extends eigenstate-based homogenization (EBH), previously formulated for perfectly bonded interfaces, to general heterogeneous composites with imperfect, damageable interfaces. The resulting framework captures coupled bulk inelasticity and interfacial damage evolution within a unified EBH formulation. A key feature of the proposed framework is its minimal representation of the fine-scale RVE solution: both bulk and interface nonlinearities are described by scalar eigenstates associated with precomputed coarse eigenmodes. Built on transformation field analysis, the formulation accurately recovers both homogenized and fine-scale fields using far fewer unknowns than the full finite element discretization.The central computational obstacle in extending EBH to imperfect interfaces is the severe ill-conditioning induced by penalty-based cohesive zone models. This poor conditioning makes monolithic fixed-point iterations, which are highly efficient for bulk inelasticity alone, prohibitively slow once interfacial damage initiates. To overcome this challenge, we develop a block FP–DR scheme: the well-conditioned bulk inelasticity problem is solved by fixed-point (FP) iterations, whereas the ill-conditioned interface damage problem is solved by Newton iterations combined with LU decomposition for the dense linearized systems of equations, herein referred to as the DR solver.Convergence studies are conducted for both monolithic and block solution schemes. The results show that convergence of the monolithic FP solver depends strongly on the interface regularization and is no longer guaranteed once a critical damage level is reached. In contrast, although the convergence rate of the block EBH schemes remains affected by the maximum interface damage and the interface regularization, these quantities no longer constitute necessary conditions for convergence.Exploiting the rapid initial convergence observed in monolithic FP solvers for bulk inelasticity, we further introduce a loosely converged block EBH variant that achieves significant computational speedup without sacrificing solution accuracy.Numerical studies on fiber-composite and polycrystalline rock RVEs demonstrate that the proposed EBH framework, particularly the block schemes, delivers high-fidelity multiscale predictions for diverse interface-dominated problems at a small fraction of the computational cost of full finite element analysis.
- Open Access
- 24 June 2026
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 461
Thermomechanically coupled visco–hyperelastic–plastic model with self-heating for thermoplastics
- 24 June 2026
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 461
Open AccessThermomechanically coupled visco–hyperelastic–plastic model with self-heating for thermoplastics
- Open Access
- 20 April 2026
- Published by Elsevier in SSRN Electronic Journal
Thermomechanically Coupled Visco-Hyperelastic-Plastic Model with Self-Heating for Thermoplastics
Thermoplastic polymers play a pivotal role in modern engineering applications owing to their meltprocessability and tunable thermomechanical properties. However, conventional constitutive models often struggle to predict the material response under realistic manufacturing conditions, as they frequently neglect the strong coupling between finite viscoelasticity, crystallinity-dependent hardening, and mechanically induced self-heating. To address this, this work establishes a thermodynamically consistent visco-hyperelastic-plastic framework specifically designed for semi-crystalline polymers. Formulated within a corotational intermediate configuration, the model employs a modular twopotential structure that ensures thermodynamic consistency by design. A key novelty lies in the explicit embedding of microstructural information, where material parameters are derived as continuous functions of temperature and crystallinity. This formulation captures how processing-induced crystallinity modulates the material's resistance against thermal softening. The constitutive update is implemented via automatic differentiation and vectorized on GPUs, ensuring algorithmic robustness and computational efficiency. Validation against PA6 dogbone specimens demonstrates that the framework quantitatively reproduces the characteristic temperature evolution-from initial thermoelastic cooling to dissipative heating and accurately captures the thermomechanical coupling mechanisms that trigger strain localization. Crucially, this transition emerges spontaneously from the entropic constitutive structure without empirical cooling terms, validating the thermodynamic consistency. This spontaneous emergence of the thermoelastic cooling effect serves as a validation of the rigorous thermodynamic consistency embedded within the proposed framework. Consequently, the proposed formulation provides a predictive and extensible computational tool for the coupled analysis of thermoplastic components.
- 20 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessThermomechanically Coupled Visco-Hyperelastic-Plastic Model with Self-Heating for Thermoplastics
Thermoplastic polymers play a pivotal role in modern engineering applications owing to their meltprocessability and tunable thermomechanical properties. However, conventional constitutive models often struggle to predict the material response under realistic manufacturing conditions, as they frequently neglect the strong coupling between finite viscoelasticity, crystallinity-dependent hardening, and mechanically induced self-heating. To address this, this work establishes a thermodynamically consistent visco-hyperelastic-plastic framework specifically designed for semi-crystalline polymers. Formulated within a corotational intermediate configuration, the model employs a modular twopotential structure that ensures thermodynamic consistency by design. A key novelty lies in the explicit embedding of microstructural information, where material parameters are derived as continuous functions of temperature and crystallinity. This formulation captures how processing-induced crystallinity modulates the material's resistance against thermal softening. The constitutive update is implemented via automatic differentiation and vectorized on GPUs, ensuring algorithmic robustness and computational efficiency. Validation against PA6 dogbone specimens demonstrates that the framework quantitatively reproduces the characteristic temperature evolution-from initial thermoelastic cooling to dissipative heating and accurately captures the thermomechanical coupling mechanisms that trigger strain localization. Crucially, this transition emerges spontaneously from the entropic constitutive structure without empirical cooling terms, validating the thermodynamic consistency. This spontaneous emergence of the thermoelastic cooling effect serves as a validation of the rigorous thermodynamic consistency embedded within the proposed framework. Consequently, the proposed formulation provides a predictive and extensible computational tool for the coupled analysis of thermoplastic components.
- 24 January 2026
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 452
Eigenstate based homogenization
- 24 January 2026
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 452
Eigenstate based homogenization
- 4 April 2024
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 425
Data-Physics Driven Three-Scale Approach for Ultra-Fast Resin Transfer Molding (UF-RTM)
- 4 April 2024
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 425
Data-Physics Driven Three-Scale Approach for Ultra-Fast Resin Transfer Molding (UF-RTM)
- 19 September 2023
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 417
Data-physics driven multiscale approach for high-pressure resin transfer molding (HP-RTM)
- 19 September 2023
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 417
Data-physics driven multiscale approach for high-pressure resin transfer molding (HP-RTM)