Creep of Bulk C-S-H: Insights from Molecular Dynamics Simulations

Mathieu Bauchy, Enrico Masoero, Franz-Joseph Ulm, Roland J.‐m. PellenqarXiv 2015 · DOI · arXiv

Introduction

Concrete, the most widely used man-made material, exhibits a phenomenon known as "creep"—a slow, continuous deformation under constant load that can last for decades. This behavior, primarily originating in the calcium-silicate-hydrate (C-S-H) phase, poses significant challenges for the long-term structural integrity of infrastructure like bridges and dams. While understanding creep at the atomic scale is crucial for developing more durable and sustainable cements, traditional Molecular Dynamics (MD) simulations are limited to nanosecond timescales, creating a massive "time-scale gap" between simulation and reality.

Methodology: The Cyclic Perturbation Technique

To overcome this limitation, researchers developed a novel "time-acceleration" method. Using a 501-atom model of C-S-H and the reactive REAXFF potential, they applied a baseline shear stress ($\tau_0$) supplemented by small, cyclic stress pulses ($\pm\tau_p$). Following each pulse, the system's energy was minimized.

This "shaking" process allows the system to navigate its complex potential energy landscape, jumping over energy barriers that would otherwise require years of real-world time to cross. Effectively, each simulation cycle acts as a proxy for a specific duration of real time, enabling the observation of long-term viscoelastic behavior in just a few computational hours.

Evolution of shear strain (black) and potential energy (red) during the cyclic loading process. The inset illustrates the square-wave stress protocol applied to the system.

Results and Validation

The simulations revealed that the shear strain ($\gamma$) increases logarithmically with the number of cycles ($N$), mirroring the empirical logarithmic law observed in concrete structures over decades. Furthermore, the total shear strain was found to scale linearly with the applied baseline stress ($\tau_0$), indicating that the material behaves linearly under sub-yield conditions.

A critical finding was the determination of the "creep modulus" ($C$), a measure of resistance to creep. The simulation results for bulk C-S-H (zero porosity) were compared against experimental nanoindentation data from cement pastes of varying porosities. When the experimental data was extrapolated to zero porosity, the agreement with the MD simulation was excellent, validating the cyclic perturbation method as a predictive tool.

Shear strain vs. number of cycles on a logarithmic scale for different stress levels, showing the linear scaling of strain with stress. The inset compares the simulated creep modulus (red square) with experimental nanoindentation data (open squares).

Discussion and Conclusion

The study demonstrates that creep in C-S-H is not merely atoms sliding past one another, but rather the system's progression into deeper, more stable energy states—a process akin to artificial aging. The independence of the creep modulus from the applied stress suggests that creep is an intrinsic property of the C-S-H atomic arrangement.

While the current model is small (501 atoms), this methodology provides a powerful framework for bridging the gap between atomic simulations and engineering timescales. It offers a new pathway for the "rational design" of low-carbon, high-performance cementitious materials by allowing researchers to predict long-term durability before physical testing.

Reference: Bauchy, M., Masoero, E., Ulm, F.-J., & Pellenq, R. (2015). Creep of Bulk C–S–H: Insights from Molecular Dynamics Simulations. arXiv:1506.06435v1.

Summary updated

Abstract

Understanding the physical origin of creep in calcium–silicate–hydrate (C–S–H) is of primary importance, both for fundamental and practical interest. Here, we present a new method, based on molecular dynamics simulation, allowing us to simulate the long-term visco-elastic deformations of C–S–H. Under a given shear stress, C–S–H features a gradually increasing shear strain, which follows a logarithmic law. The computed creep modulus is found to be independent of the shear stress applied and is in excellent agreement with nanoindentation measurements, as extrapolated to zero porosity.
BibTeX
@misc{bauchy2015creep,
  author = {Mathieu Bauchy and Enrico Masoero and Franz-Joseph Ulm and Roland J.‐M. Pellenq},
  title = {{Creep of Bulk C-S-H: Insights from Molecular Dynamics Simulations}},
  year = {2015},
  eprint = {1506.06435},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1506.06435},
}
Figure