Observing fresh water freezing in highly detailed 3D printed fracture replicas

The study of groundwater freezing in fractures of crystalline rock is essential for understanding subsurface flow dynamics during ice age events. Freezing significantly alters the hydraulic permeability, which is particularly relevant for the safety assessment of a nuclear waste repository in crystalline rock. German law mandates safety evaluations for at least one million years, during which multiple ice ages are likely to occur and potentially causing freezing in fractures.

Due to the challenges of directly observing fresh water freezing in real rock, an alternative measurement approach using 3D scanning of fracture surfaces and subsequently 3D printing of fracture replicas was investigated. Surface data of natural fractures in granitic rock were captured with a high-resolution 3D scanner. After post-processing the datasets, a digital model of a test cell for flow tests in the lab including the fracture surfaces was created and printed. The cell was fabricated using a high-resolution Formlabs Form 3 printer with a clear resin. This material was chosen for its transparency and durability under low temperatures.

The printed test cell was placed in a climate chamber and equilibrated to a temperature of- 5°C. Freezing processes were monitored by an industrial camera using a 0.05% methylene blue solution, which changes color from dark blue to transparent during crystallization. Initial tests did not account for the expansion of the tracer solution during freezing which proved to occur in significant proportions separating the two fracture halves. In subsequent tests, this shift was restricted using a metal frame. A series of scenarios with varying temperatures and inflow rates were tested.

It turned out that both, the freezing pattern and the point where freezing originated, varied between individual experiments. Occasionally, freezing even began in the tubing, causing blockages and leading to premature termination of the experiments.

To analyze the images, threshold segmentation was applied to the digital photographs using Matlab. This resulted in a binary array that represents the state of the methylene blue solution for each pixel. Each entry in this array corresponds to an area of approximately 150 by 150 µm. Although the image processing technique is advanced, the sensitive test setup was still affected by external disturbances. These disturbances resulted in spikes and other artifacts in the images. While some of these issues were mitigated, it was not possible to eliminate all disturbances and achieve a completely undisturbed test.