TL;DR
Researchers have developed a fully coupled hygro-thermo-mechanical (HTM) model to explain "contour scaling"—the mysterious way historic sandstone monuments shed their outer layers. The study reveals that repeated wetting and drying cycles, combined with the weight of the structure, are enough to trigger subsurface fractures even without the help of freezing temperatures or salt crystallization.
What Problem Does This Paper Address?
Historic monuments, such as the Cathedral of Lausanne, are often built from clay-bearing sandstones like the Swiss "grey molasse." These stones suffer from a devastating form of decay called contour scaling, where large flakes of stone fracture parallel to the facade and eventually detach.
For decades, scientists attributed this primarily to ice lensing or salt crystallization. However, scaling often occurs in climates where these factors aren't dominant. The challenge has been understanding how the microscopic swelling of clay minerals (smectites) within the stone's "cement" translates into macroscopic structural failure. This paper seeks to model the "initiation" phase of this damage—how micro-cracks coalesce into a failure zone deep within the stone.
Methods
The authors developed a sophisticated numerical framework that couples two distinct physical domains:
- Hygro-Thermal (HT) Modeling: Using the software Delphin, they simulated how moisture and heat move through the porous stone. This accounts for liquid water transport, vapor diffusion, and the "moisture storage function" of the stone.
- Mechanical Modeling: Using the FE solver Abaqus, they calculated stress, strain, and damage. They employed Continuum Damage Mechanics (CDM), where a scalar variable ($D$) represents the degradation of the material's stiffness.
- The Coupling: This is the paper's core innovation. Moisture causes the stone to swell and soften (HT $\rightarrow$ M), while the resulting damage alters the stone's porosity and pore-size distribution, which in turn changes how it absorbs and holds water (M $\rightarrow$ HT).
To find the most damaging weather patterns, they used Bayesian optimization with a surrogate model (Gaussian Process Regressor) to identify the "worst-case" climatic conditions for accelerated testing.
Datasets
The study didn't use a massive external dataset but rather relied on high-fidelity experimental data for material calibration:
- Mercury Intrusion Porosimetry (MIP): To determine pore-size distribution.
- Sorption Isotherm Tests: To map how much moisture the stone holds at different humidity levels.
- Lab Tests: Water uptake, drying, and longitudinal swelling tests were performed on Villarlod molasse samples to calibrate transport and expansion parameters.
- Literature Data: Mechanical properties (Young’s modulus and tensile strength vs. moisture) were sourced from previous petrophysical studies of Swiss sandstones.
Key Insights and Ideas
- The "Subsurface" Paradox: Damage doesn't start at the exposed surface. Because the surface is free to expand, it doesn't build up enough stress to crack. Instead, the damage localizes 3–5 cm deep, where the material is "constrained" by the rest of the block.
- Moisture Trapping: As the stone degrades, its ability to transport and store water changes. The model showed that moisture becomes "trapped" in the damaged zones (6–10 cm deep), creating a persistent damp layer that further weakens the stone.
- The Necessity of Load: One of the most striking findings was that moisture cycles alone aren't enough to cause scaling. Damage only occurred when the external compressive load (the weight of the wall above) was applied simultaneously with the climatic cycles.
- Shear Bands: The model predicted that damage eventually organizes into 45° shear bands, providing a mechanical explanation for how micro-cracks eventually form the continuous fractures seen in the field.
Key Results
- Validation: The predicted damage depth (3–5 cm) and moisture accumulation depth (6–10 cm) align remarkably well with on-site impedance measurements and observations at the Cathedral of Lausanne.
- Worst-Case Climate: The optimization identified specific parameters (e.g., a rain flux of 0.0018 $dm^3/m^2s$ and a drying surface temperature of 28.8°C) that maximize the risk of damage propagation.
- Stiffness Decay: The model successfully captured the "softening" of the stone, showing that as damage accumulates, the strain energy increases, leading to a feedback loop of further degradation.
Limitations and Future Work
The authors acknowledge several areas for expansion:
- 2D vs. 3D: The current model uses a 2D plane-strain assumption. Real-world blocks have 3D boundary conditions that might influence the edges of the scaling zones.
- Fracture vs. Damage: The model tracks "dispersed damage" (weakening) rather than the growth of a discrete, sharp crack. Future work could integrate the Discrete Element Method (DEM) to simulate actual fragment detachment.
- Chemical Effects: The model focuses on physics and mechanics; it does not yet account for chemical weathering or the long-term effects of biological growth (like lichens) on moisture transport.
- Ice and Salt: While the paper proves scaling can happen without them, the authors note that freezing and salt are still important accelerators that should be integrated into the HTM framework in future iterations.
Summary updated