Modelling the behaviour of host rock for nuclear waste

Nuclear energy is nowadays a crucial issue and the storage of radioactive waste raises many questions. Over the past few decades, many countries have subsequently built underground laboratories allowing them to test hypotheses on a large  scale. In Belgium, a decision was taken in 1974 to create the HADES laboratory in the layer of clay situated under the Nuclear Energy Study Centre in Mol, with the first excavation works taking place in the 1980s. The various studies carried out over the past three decades have led to the development of various concepts. This is especially the case regarding the behaviour of the rock that will host the waste. This is due in particular to the development of calculation tools which currently allow the user to model these behaviours fairly accurately, by taking into account the most complex situations, especially in the long term over several tens of thousands of years.

The Laboratory of Geomechanics and Engineering Geology (ArGEnCo department at the Faculty of Applied Sciences, University Liège) is one of the few teams in Europe working on this numerical area. Three doctoral theses have just been defended, which demonstrate the laboratory’s competence in this field. Thanks to their input, several major steps have been accomplished in the modelling of the fracture zones that appear during excavation, the behaviour of rock under the effect of ventilation, and the behaviour of concrete supports and the plugs designed to seal off the galleries and fill in the technological gaps.

The Laboratory of Geomechanics and Engineering Geology (ArGEnCo department at the Faculty of Applied Sciences, University Liège) studies the behaviour of soil and rock under various stresses (excavation of galleries or foundations, for instance). Its internationally renowned expertise has led to the development and use of a finite element method that enables the numerical modelling of complex phenomena. Among the areas studied over the past 30 years is the burying of long-lived, highly toxic nuclear waste. Several countries are testing storage solutions in galleries excavated in deep geological repositories. In Belgium, this storage is studied in the underground HADES laboratory, located 225 metres beneath the Centre d’Etude de l’Energie Nucléaire (CEN) in Mol. This research is continuing in close collaboration with the Belgian universities that have developed competences in this field. Including ULg, where three doctoral theses were defended at the end of 2015 and at the beginning of 2016. "We studied the behaviour of the host rock and the materials that might be used around it – support, plugs – for efficient storage", says Professor Robert Charlier, co-director of the Laboratory of Geomechanics and Engineering Geology and promoter for these theses alongside Frédéric Collin.


The type of host rock chosen by countries such as Belgium, France and Switzerland is argillaceous, while other countries, especially those in Scandinavia, prefer granite. The storage of long-lived, highly radioactive nuclear waste is governed by the law of “staying within one’s own borders”. Hence the increase in test sites, each with its own specificities that obviously have to be taken into account, although this doesn’t rule out international collaborations. The three theses from Liège fall within this scope, one of them referring to the Belgian site in Mol and the other two to the French site, where ANDRA (National Agency for Radioactive Waste Management) carries out experiments in Bure (Haute-Marne).

Rock fracturing

In the first recently-defended thesis(1), Benoît Pardoen focuses on the argillaceous rock that hosts the underground ANDRA laboratory at the site in Bure. He particularly studied the phenomena that occur when this clay is excavated, and during exploitation and maintenance phases. These galleries have to be ventilated, which dries out the rock and thus alters its state; as for the waste, it releases heat, another element to be taken into account during long-term simulation.
The first part of the thesis deals with phenomena that appear when the rock is excavated. "Clay, especially the one in Bure", Benoît Pardoen explains, "is an anisotropic material; the characteristics of the phenomena that appear will therefore be different depending on the direction of the excavation. This was one of the first steps in my research."

The experiments carried out at the ANDRA site, in particular, showed that during excavation, damaged zones and fracture zones develop around the galleries. These zones take the form of elliptical rings oriented either vertically or horizontally in relation to the direction of the excavation. Something that is difficult to explain. As part of his thesis, Benoît Pardoen was able to model the appearance of these phenomena, both the vertical (the easiest to model) and horizontal ellipses. "This won’t influence the excavation", the young researcher concludes, "but it helps to explain what is happening and, depending on how we excavate, to assess the risk of interactions between fracture zones and the increase of the permeability of the rock surrounding the gallery."

Benoît Pardoen also focused on hydromechanical coupling, i.e. the influence of fracturing on permeability. If the rock cracks, there is an increase in permeability and therefore the creation of preferential pathways for radionuclide migration through the rock. Here too, the aim is to reproduce large-scale measurements. With modelling, it is possible to deal with time scales, which can’t be done in a laboratory: at test sites, such as those in Bure or Mol, sensors record a series of parameters but this has only been happening over the past few decades. And in geological time, that is very short. "We’re sometimes doing calculations over 100,000 years”, Benoît Pardoen points out.
Thousands of years during which a whole range of phenomena will occur. "Radioactive decay produces heat which will dissipate over thousands of years”, Robert Charlier explains. "If we want to know how this heat will be dissipated and affect the rock mass, we have to build models over periods of several thousands of years. The waste loses its toxicity after several tens of thousands of years, up to 100,000 years. Another issue: the waste is sealed in stainless steel canisters… which nevertheless oxidise in the end! In a thousand years, the steel will have corroded, which requires oxygen! They are stored in an airtight environment which will be saturated in water in a thousand years time. Therefore, the steel will draw oxygen from the water to oxidise and will release hydrogen; gases will be produced which can increase the pressure. So we have to know how these gases are going to be able to migrate…” Which Benoît’s thesis helps to explain.

"Which", as Professor Charlier points out, "leads us to another problem: the plugs that are supposed to seal the galleries and fill the cracks. If we use a plug that is impermeable to gas, the gallery will become like a champagne bottle and may cause problems. So, should the plug prevent the gas from escaping or not? This question certainly goes against the preconception that has been the norm for a long time: nothing should ever escape from these galleries! But over time, this way of thinking has evolved and the idea of recoverability currently prevails. Maybe in 50 years time, when knowledge will have evolved, we will be able to make the waste inert and no longer toxic. In that case, we have to be able to recuperate it! Therefore, we have to reseal the galleries in the most impermeable way possible but, at the same time, we want to be able to reopen them without them collapsing. It is a major stake and our modelling work helps to take the necessary decisions."

The thesis defended by Benoît Pardoen also deals with another important problem, that of the ventilation required during excavation works and also while the waste is being stored, before the galleries are closed. This ventilation dries out the rock but once the galleries have been resealed, the rock will be saturated in water again.  How does the clay behave with regard to this desaturation and resaturation?  To understand it, it was necessary to model the hydraulic behaviour of the rock.

The concrete blocks in Mol

Fatemeh Salehnia’s thesis(2) also deals with the modelling of the rock behaviour although there are two major differences compared with the previous one. First of all, this thesis uses data from the Belgian site in Mol. Also excavated in clay… but one type of clay isn’t the same as another: the one in Mol is more “superficial” than the one in Bure (200 m depth compared with 500 m), the age isn’t the same, nor the porosity (7-8 % in Bure, almost 50% in Mol), and the permeability is low in Mol but it is 10 to 100 times lower in Bure. All differences that require special studies. The second major difference: in Mol, there are supports, which is not the case in Bure. "My work", the young researcher explains, "consisted of studying not just the behaviour of the rock but also that of the supports and the interaction between the two types of material". The rock in Mol is far more deformable that the one in Bure. It is therefore essential that these galleries are provided with supports, whose design has evolved over the years and according to various studies. The solution that was finally adopted was a support in the form of concrete blocks.

Modelling the behaviour of concrete when interacting with rock was one of the key points of Fatemeh Salehnia’s thesis. We know that the resistance of a vault especially depends on the pressure exerted on it and engineers take this resistance into account to meet the expectations of durability: the galleries must remain intact as long as possible to give future generations the opportunity to recuperate the wastes. It is therefore essential to model the behaviour of the concrete blocks in connection with the pressure exerted by the surrounding rock in the long term (several dozen or hundreds of years). As in the case of Bure, damaged zones and fractures appeared in the rock during excavation. Their structure was also studied. The result, as demonstrated by Fatemeh Salehnia, engenders a far more heterogeneous pressure on the support than it is usually the case. As the support is composed of concrete blocks – the voussoirs – its behaviour is particularly complex.  

Another major advance in the thesis was the fact that the concrete’s viscosity was taken into account. "This wasn’t at all planned when I began my research", Fatemeh Salehnia remembers. "But we had to consider it when we wanted to begin to do long-term simulations because we couldn’t explain the obtained answers". So what is the viscosity of concrete? To understand this notion, think of a shelf stacked with books. After a certain amount of time, even if you don’t add any extra weight, the shelf will acquire a permanent curve. And this deformation will be irreversible. The same is also true, to varying degrees, for all materials including concrete: it is just a question of time. Part of the simulations the researcher did, meant she had to take into account the phenomenon of concrete’s viscosity otherwise some of the answers in the long term would be incomprehensible. It is a phenomenon that must be taken into account if you want to make calculations over a long period of time.

Bentonite plugs

Back to France and the Bure site for the third thesis(3), defended by Anne-Catherine Dieudonné. This time, the focus is no longer on the host rock but on the plugs that could be used to seal the fractured galleries studied in the previous theses. These plugs are composed completely or partly of a clay known as bentonite, which swells when it absorbs water. All clays have this property but to highly varying degrees. Kaolinite swells very little or not at all; while smectite, the main component of bentonites, is a clay that reacts very strongly with water, hence the use of bentonite in applications such as the stabilisation of excavations or drilling for oil. "In the case in hand", Anne-Catherine Dieudonné explains, "we are using dry bentonite, which is extremely compacted and contains very little water. It will swell, through natural or artificial hydration, come into contact with the rock – there isn’t a concrete support structure in the French experiment –, exert pressure on this gallery and thus seal the fractures in the damaged zones and form a seal. This allows the galleries to be hydraulically sealed in just a few years." The technique has been known for a long time but it fells out of favour for a while. However, it recently became the subject of renewed interest requiring new studies, hence the thesis presented in Liège. The stake is important because when it comes to placing a plug such as this, it is necessary to know for how long it will be effective. Experiments showed that little was known of rehydration kinetics. The work of Anne-Catherine Dieudonné therefore consisted of trying to better understand the kinetics and how the material’s permeability evolves. A tricky job because bentonite has a bimodal distribution of porosity, with two classes of porosity. One on a very small scale: that of argillaceous particles, in layers of nanometric thickness; and the other on a larger scale: aggregates. It is the latter that mainly contributes to the material’s permeability. The thesis defended by Anne-Catherine Dieudonné shows that if water is used to make bentonite swell in a confined space, i.e. at a constant volume, the water obstructs the macroporosity because the bentonite can’t increase in volume. Thepermeability is therefore reduced and hydration occurs more and more slowly, as well as the sealing process. On the other hand, if hydration takes place at free volume, the bentonite can swell and permeability remains intact. The host rock is therefore sealed more quickly. The modelling of this phenomenon is very complex because, in reality, the two types of porosity come into play at the same time. There are indeed places where bentonite can swell freely and others where it can’t.

A second stake, also dealt with in the thesis, is to determine what pressure will ultimately be exerted by the plugs on the walls of the host rock: will it be homogenous or not? Enough to seal the cracks or not? Will this create new cracks if the pressure is too great? The answers to these questions will obviously determine the implemented solutions. Complex and refined, they will undoubtedly do a near perfect job of covering up the cracks but their cost and their complexity probably mean that they will never be deployed. On the other hand, if they are too simple and incomplete, they will be cheap and easier to apply but undoubtedly won’t provide a suf

(1) Hydro-mechanical analysis of the fracturing induced by the excavation of nuclear waste repository galleries using shear banding, Pardoen Benoît, University of Liège, 2015, doctoral thesis.

(2) From some obscurity to clarity in Boom clay behavior: Analysis of its coupled hydro-mechanical response in the presence of strain localization, Salehnia Fatemeh, University of Liège, 2015, doctoral thesis.

(3) Hydromechanical behaviour of compacted bentonite: from micro-scale behaviour to macro-scale modelling , Dieudonné Anne-Catherine, University of Liège, 2016, doctoral thesis.