Liquid water uptake in unconfined Callovo Oxfordian clay-rock studied with neutron and X-ray imaging

Abstract

The Callovo Oxfordian clay-rock (COx) is studied in France for the disposal of radioactive waste, because of its extremely low permeability. This host rock is governed by a hydromechanical coupling of high complexity. This paper presents an experimental study into the mechanisms of water uptake in small, unconfined, prismatic specimens of COx, motivated by the comprehension of cracking observed during concrete/COx interface sample preparation. Water uptake is monitored using both X-ray tomography and neutron radiography, the combination of these imaging techniques allowing material deformation and water arrival to be quantified, respectively. Given the speed of water entry and crack propagation, relatively fast imaging is required: 5-min X-ray tomographies and 10-s neutron radiographs are used. In this study, pairs of similar COx samples from the same core are tested separately with each imaging technique. Two different orientations with respect to the core are also investigated. Analysis of the resulting images yields with micro- and macro-scale insights into hydromechanical mechanisms to be obtained. This allows the cracking to be interpreted as a rapid breakdown in capillary suction (supposed large both to drying and rebound from in situ stress state) due to water arrival, which in turn causes a loss of effective stress, allowing cracks to propagate and deliver water further into the material.

Appendix: In situ imaging techniques

Full-field imaging is a very precious tool in experimental mechanics especially for the observation and study of inhomogeneous phenomena. The use of penetrating radiation allows the inside of studied specimens to be revealed, which is a great improvement over photography for the representativeness of the measured field. The attenuation of penetrating radiation beam (of a given energy) as it travels through matter can be modelled using the Beer–Lambert law, expressed in its differential form in Eq. 2, the integration of which gives Eq. 3:

$$\begin{aligned} \frac{\mathrm{d}I}{\mathrm{d}x} & = -\,\mu \ I(x) \end{aligned}$$

(2)

$$\begin{aligned} I\, & = \,I_0\,e^{-{\mu _m}{\rho }x} \end{aligned}$$

(3)

where I is the resulting beam intensity after interaction with the material, $I_0$ is the reference beam intensity without interaction (no sample), $\mu _m$ is the mass attenuation coefficient ($\mu _m\,=\,\mu /\rho$, where $\mu$ is the linear attenuation coefficient which is a material property), $\rho$ is the density of the material and x is the linear distance that the beam travels inside the specimen.

Furthermore, the use of penetrating radiation enables the use of tomographic techniques, which give access to the attenuation field—a 3D “image” of $\mu$.

In this work we present full-field measurements of water imbibition into a clay-rock observed with two highly complementary techniques using different types of penetrating radiation—X-ray tomography and neutron radiography—both of which are briefly below.

1.1 X-ray tomography

X-rays are photons with high-energy that interact with the electron cloud surrounding each atom’s nucleus. Therefore the probability of an X-ray photon being attenuated is related to the atomic number Z, which itself is roughly proportional to the density of the material.

Tomography consists in sampling the unknown attenuation field within a specimen from a large number of different orientations, which allows an accurate reconstruction of the desired attenuation field using back-projection.

Tomography was developed with X-ray radiation, and X-ray tomography remains the most developed form of tomography with high-performance laboratory scanners and dedicated synchrotron installations.

Within mechanics in general, and geomechanics in particular, X-ray tomography has become a de-facto technique for time-resolved 3D analysis. For example, [13] presents the in-situ deformation mechanisms in Callovo Oxfordian clay-rock in triaxial compression using X-ray tomography with important results regarding strain localisation. The measurement of strain in this material is possible due to the natural inhomogeneities within the rock, that present a convenient pattern tracking with image correlation. The observed patterns reveal that the different components that make up the rock must have significantly different densities. X-ray tomography (using a lab-based micro-focus scanner) will be used in this work to acquire a number of 3D X-ray attenuation fields during the imbibition process, allowing the deformation of the rock to be measured. However, due to the relatively low density of water, this technique will have difficulty measuring the invasion of water into the sample.

1.2 Neutron radiography

Neutrons are neutral massive particles that interact directly with the nucleus of atoms. Neutrons generally have a much higher probability of interaction with light atoms like hydrogen, even though attenuation can be radically different between isotopes of the same atom (for example deuterium has a much lower attenuation than hydrogen).

Neutrons principally interact with nuclei by being scattered or absorbed; when a neutron is scattered its speed and direction change, whereas when it is absorbed a wide range of radiations can be emitted or fission can be induced in some specific elements. The nucleus may rearrange its internal structure and release one or more gamma rays, charged particles may also be emitted. The sensitivity of neutrons to hydrogen make it an ideal complement to X-rays, significantly facilitating the detection of water throughout the imbibition process. In the interaction of neutrons with water, the principal mechanism is elastic scattering.

To obtain low-noise radiographs at the pertinent resolution for this phenomenon, neutron imaging is slow (even compared to a lab-based X-ray scanner), and so it has been elected to perform neutron radiography for this study, since the time required for neutron tomography was estimated to be too long.

Fig. 13

Calibration of the beam absorption in $\hbox {H}_2\hbox {O}$ with neutron path length, on the left, experimental schematic, in the middle experimental data, top-right grey value in some zones of interest, bottom-right the fit of the Beer–Lambert law to the acquired data

Since a 3D field of neutron attenuation will not be reconstructed, great care has been taken to fully characterise the interaction of neutrons with water in the experimental conditions used for this test. Figure 13 shows a key step in this procedure where a hollow aluminium wedge has been scanned with neutrons before and after filling with $\hbox {H}_2\hbox {O}$ , in order to fit the attenuation coefficient $\mu _m$ of neutrons to water, net of the whole imaging setup. The left of the figure presents a schematic showing the setup. It is important to note that in order to maximise spatial resolution, in neutron imaging the sample is kept as close to the detector as possible ($\approx$ 10 mm from the closest edge of the sample). The middle of the figure shows the recorded radiographs of $I/I_0$ before and after filling the wedge with water. The change in attenuation of the beam between these two images is then extracted along the profile shown in violet in Fig. 13 middle-bottom. Together with the knowledge of the geometry of the wedge, this allows the relationship between the thickness of water in the direction of the beam (x in Beer–Lambert) and its attenuation ($I/I_0$), as shown in violet in Fig. 13 right-bottom. This experimental data is fitted with the Beer–Lambert law (Eq. 3) with $\mu$ being the only free parameter. The fit is made in the range [0, 2.3] mm, that is to say until half of the “mean-free-path length” ($\lambda$) of neutrons with wavelength of 1.8 Å in $\hbox {H}_2\hbox {O}$ (close to the peak of 3 Å in the beam used), as given in [19]; the fit gives $\mu = {0.310}\, {\hbox {mm}^{-1}}$.

It is immediately apparent that for more than 3 mm of $\hbox {H}_2\hbox {O}$ there is a significant deviation between the fitted Beer–Lambert absorption and the measured attenuation data, this is likely due to scattering of the neutron beam. For short path lengths in the water (well below the mean-path-length of 4.6 mm), the attenuation of the beam is low, therefore there is a significant amount of transmission ($I/I_0$). The non-transmitted beam is scattered and detected elsewhere, but since this is a small amount, the relationship between distance in water and attenuation is modelled well by Beer–Lambert. For longer path lengths, for example 10 mm, a 5% of the beam is expected to be transmitted. This implies that almost 95% of the beam will be scattered, and the scattered beam itself will be scattered again in a phenomenon known as multiple scattering. Some of this diffused beam will be detected at the position of the transmitted beam, thus increasing the measured flux to 19%. The effect of scattering is clearly visible in Fig. 13 right-top, where the measured intensity in red green and blue zones increases when water is added.

We therefore have a good degree of confidence in the Beer–Lambert law for up to 3 mm of water in the direction of the beam.