The IAEA safety fundamentals document2 states that:

The generation of radioactive waste must be kept to the minimum practicable level by means of appropriate design measures and procedures, such as the recycling and reuse of material.

‘Waste minimisation’ or (more appropriately, perhaps) ‘waste avoidance’ is derived from the principle of sustainability and there two fundamental ways of achieving it. The first is through the control of practices that lead to unnecessary radioactive waste generation. An example is the carrying of non-essential items such as packaging on replacement plant items (filters for instance) into NPP contamination-controlled areas. The second is through the decontamination of (mainly metallic) waste, which may be done chemically, mechanically and/or by re-melting. The aim is to reduce contamination to below clearance levels so that the metal may be recycled as scrap. By avoiding the need to dispose of material as radioactive waste, considerable cost savings can be made. Care should be taken to maintain a distinction between waste avoidance and volume reduction although, as we shall see, when it comes to re-melting one does rather merge into the other.

PCGE presents no data on the costs of carbon storage (‘sequestration’) probably because it is common practice for CO2 to be sold to the petroleum industry, which pumps it into wells to improve oil recovery. While it may seem strange that one is simultaneously acting to both reduce CO2 emissions and to increase them (by producing more oil) there is no doubt that there is a net benefit in terms of emissions. This practice means that, at the present time, carbon dioxide sequestration is essentially free. If carbon capture becomes widespread, however, and especially when it is deployed in regions where no use can be found for it, this may no longer be the case. This may represent a hidden bias in the calculation in favour of coal+CC.

First of a kind (FOAK) and Nth of a kind (NOAK) technology

The PCGE data do not distinguish between FOAK and NOAK technology. Where FOAK plants can be inferred (e. g. advanced reactor designs, supercritical coal combustion, etc.) there is often no obvious difference in overnight cost between these and more standard technology. Nevertheless, it is clear that considerable cost savings can be made from replication. A report for the UK government by Mott MacDonald indicates a reduction in costs of between 20 and 40% in moving from FOAK to NOAK. These savings result from removal of the FOAK premium and reductions in the supply chain cost, contingencies and unscheduled over-runs. This report provides a ‘headline’ figure (i. e. without contingency) for engineering procurement and construction for a FOAK plant of $4750 per kWh(e). After removing the 15% contingency that is included in the nuclear overnight cost shown in Table 5.1, this suggests that, when applied to Europe and North America, the value used here probably lies somewhere between NOAK and FOAK.

Electromagnetic isotope separation was used in combination with gaseous diffusion to build the first atomic weapons. Uranium in the form of solid uranium tetrachloride (UCl4) is put into an evacuated chamber and heated to high temperature to produce UCl4 vapour (boiling point of 792 °C at 101 kPa). The vapour is then bombarded with an electron beam to break down the molecules to generate U+ ions and other species. The charged particles are accelerated by an electrical potential to form an ion stream. The ion stream is passed through a strong magnetic field, which causes the ions to start to travel in a circular path by acting upon the particles according to the charge to mass ratio. The charge to mass ratio of the 235U+ ion is slightly higher than for the 238U+ ion so that the magnetic field will cause those ions to travel in a tighter radius, the result being that the ions of the two isotopes will separate into two beams. Careful positioning of collectors then allows the two ions to be separated. The principle of ion generation, magnetic separation and collection is essentially the same as found in mass spectrometers.

The technology provides effective separation but has a number of drawbacks, in particular:

• Not all of the UCl4 is converted to ions, so that a significant proportion of the starting material is not collected.

• Many of the ions ‘miss’ their designated collector, so that they are again not collected.

• The material is deposited within the equipment so that it quickly becomes contaminated and must be cleaned out both to allow continued operation and to recover feed material that would otherwise be wasted.

• It is a batch process with a significant time between runs.

• Many thousands of units would likely be required for a commercial scale enrichment facility.

For these reasons, the technology has never progressed to civil, commercial use.

The fuel cladding is a tube whose length, diameter and wall thickness depend on reactor type (Table 9.2) and assembly design. It has to withstand corrosive high temperature water on the outside, stresses and strains imposed by expanding pellets in contact with the tube on the inside, and corrosive fission products released from the pellets. (Modelling of these phenomena is discussed in Chapter 14.) The tube is manufactured in several steps, Fig. 9.9, starting from an ingot produced by 2-3 times vacuum melting of reactor grade zirconium, recycled scrap and the alloying elements for obtaining the desired product. Several drawing and heating steps are required before the hollow billets are reduced to the size of the cladding, guide tube or water rod. The beta-quenching to form second phase particles and the final heat treatment (stress relief annealing, partial or full recrystallisation) are important steps, which influence corrosion, creep and yield properties of the material.

In the past, CANDU reactors and BWRs particularly suffered from so-called pellet-clad interaction (PCI) failures (Cox, 1990). The mechanism is often a stress

corrosion cracking (SCC) process induced by localised mechanical stresses and the fission product iodine. The problem was solved by introducing graphite — coated cladding in CANDU reactors and cladding with an inner liner in BWRs. The latter consists of a layer of material (e. g. zirconium with 400 ppm Fe) with better SCC resistance and higher ductility than Zry-2. This SCC barrier may be protected by an additional, thinner layer against rapid corrosion and deterioration from the inside should the cladding be breached and water enter the fuel rod. The liner is co-extruded with the substrate and metallurgically bonded to it.

The same technique is used to obtain an outer, corrosion resistant layer on PWR Zry-4 cladding. The Zry-4 tube is metallurgically bonded with an extra-low tin (Sn) outer layer about 100 pm thick. In this corrosion-resistant layer, the Sn level is below and the Fe and Cr levels are above the range specified by ASTM for Zry — 4. Such DUPLEX cladding combines high corrosion resistance with the good mechanical properties of Zircaloy-4 (Garzarolli et al., 2000).

P. G. BOCZAR, IndependentConsultant(formerlyDirector, Reactor Core Technology Division, Chalk River Laboratories, AECL), Canada

Abstract: This chapter begins with a description of the CANDU reactor, describing the main plant features, its fuel and fuel cycle and the control and safety aspects of the reactor. The chapter then summarizes evolutionary trends in the reactor design, followed by a discussion of the advanced fuel cycles that could be employed in existing or future CANDU reactors, including LEU, recovered uranium, DUPIC, MOX, actinides and thorium.

Key words: CANDU, EC6, ACR-1000, advanced CANDU reactor, SCWR, supercritical water reactor, advanced fuel cycles, recovered uranium, MOX, thorium, actinide burning.

11.1 Introduction

A distinguishing feature of the CANDU®[19] power reactor is its use of natural uranium fuel. This fuel has many advantages, including simplicity, low fuel cycle cost, high uranium utilization and ease of fuel manufacturing localization. In the back-end of the fuel cycle, the higher volume of used nuclear fuel is offset by lower decay heat and radiotoxicity compared to used LWR fuel (Boczar et al, 2010). The Advanced CANDU Reactor™ (ACR-1000™[20]) and the heavy-water moderated, pressure-tube supercritical-water-cooled reactor (PT-SCWR) represent evolutions of the traditional CANDU design. The features of the CANDU reactor that enable the use of natural uranium fuel also facilitate the use of a wide variety of fuels and fuel cycles, including recovered uranium (RU) from reprocessed light water reactor (LWR) fuel, low enriched uranium (LEU), MOX, minor-actinide bearing fuels and thorium. Fuel cycle flexibility is hence another hallmark of the CANDU design. While there are several variants of the CANDU reactor design, this chapter will use the 700 MWe-class CANDU 6 reactor for illustration (AECL, 2005) with the 37-element fuel bundle. (In terminology, a CANDU fuel element is equivalent to an LWR fuel rod; a CANDU fuel bundle is equivalent to an LWR fuel assembly, although much smaller.)

This section only covers the key aspects of partitioning and transmutation technology relevant to this chapter (see Chapter 17 for further information). The type of fuel form is dependent not only on the kind of reactor (thermal or fast), but also on the type of cycle (heterogeneous or homogeneous) and the nature of the envisaged reprocessing. Different options have been considered for transmutation of advanced fuels (Ogawa et al., 2005; Warin and Boullis, 2008). The main issue is the very different chemical behaviour of the various actinides. The ‘classic’ fuel forms like oxides, successfully used up to now for U — and (U, Pu)-based fuels, are not directly applicable to MA-bearing fuels. In order to obtain a high level of transmutation, the fuel should be irradiated up to very high burn-ups, and it should be particularly resistant to radiation damage. Key points for the choice of a transmutation fuel are also thermal conductivity and density. Carbides and nitrides meet these requirements. However, carbides and nitrides of Am are volatile and extremely difficult to treat, while there is still limited knowledge about the behaviour of carbides and nitrides of Cm (Bomboni, 2009). Another important aspect to consider is the production of He in MA-based fuels due to the transmutation of 241Am, which could cause an excessive swelling rate. These problems have led to research on U, Pu or MA-bearing oxides, nitrides (Arai et at., 2008), carbides (NEA, 2005) and metals (Pasamehmetoglu, 2008) for thermal or fast reactors.

Another option is that of fuel dispersed in a matrix. Inert matrix fuel (IMF) consists of a neutron-transparent matrix (generally with good thermal conductivity) and a fissile phase that is either dissolved in the matrix or incorporated as macroscopic inclusions. The matrix plays a crucial role in diluting the fissile phase to the volumetric concentrations required for reactor control. The same role is played by 238U in conventional low enriched uranium (LEU) or MOX fuel. The key difference is that replacing fertile 238U with a neutron-transparent matrix eliminates plutonium breeding as a result of neutron capture. Oxides, metals, carbides and nitrides have been found to be suitable matrix materials. As an example, silicon carbide (SiC) and yttrium-stabilized zirconia (YSZ) are two typical IM materials for He-cooled high-temperature reactor technology (IAEA, 2006).

The concept of a dedicated, moderated assembly for MA burning in FRs is also relevant to transmutation. In order to better exploit the high flux of FRs, introducing a MA-bearing fuel mixed with moderating material in certain zones of the core can maximize the transmutation rate of these nuclides. Many different materials have been considered as moderators for these dedicated assemblies: hydrides are a particularly promising option, since they are very good moderators because of their H content. However, one disadvantage is their relatively ‘low’ (at least as far as high-temperature reactors are concerned) operating temperatures (<800 °C).

As far as the reprocessing technology of advanced fuels is concerned, a few processes, developed at the laboratory scale, have already reached the target efficiency (see Table 13.6 for some examples of processes developed in Europe). Pyrochemical methods, for example, are a very attractive option for high burn-up fuels, since they are based on fuel dissolution in molten salts from which individual actinides are selectively precipitated by electrorefining. A few aqueous processes also show promise, but these are based on organic molecules that are vulnerable to radiolysis, while molten salts are highly stable in all possible conditions.

Whereas aqueous processes require mainly oxide fuels, pyrochemical processes allow the treatment of any fuel form and the recovery of virtually every kind of element. Nevertheless, they are currently characterized by low recovery efficiencies, very corrosive reagents and high secondary losses (NEA, 2006a).

AIROX-like processes, in which UO2 fuel is powdered by oxidation to U3O8 and re-enriched without separation of either solid fission products or HMs, could offer a potential alternative (see Fig. 13.3). Using these processes, it should be possible to decouple deployment of FRs from development of expensive commercial reprocessing of LWR SNF (Greenspan, 2007; Feinroth et al., 1993). The carbon dioxide oxidation (CARDIO) process is another innovative reprocessing method (Greenspan, 2007). It is a dry process for UC spent fuel. The process can be subdivided into three stages:

1 UC spent fuel can be converted into UO2 via UC + 3CO2 ^ UO2 + 4CO at T>670°C

2 Applying the AIROX process it is then possible to remove volatile fission products

3 Applying the carbothermic reduction of oxide fuel in a high-purity inert atmosphere (UO2 + 3C ^ UC + 2CO) it is possible to produce UC fuel again.

Given the coolant pressure and the irradiation history, i. e. the evolution of the axial distributions of pin power, bulk coolant temperature and fast neutron flux with time, a fuel performance code calculates the evolution of the thermo­mechanical state of a fuel pin. This involves modelling a large number of phenomena. As described in Section 14.2, these include: (a) standard phenomena associated with thermo-mechanical behaviour of the fuel and cladding materials, i. e. heat transfer by conduction, convection and radiation, thermal expansion, creep, elasticity, plasticity, fatigue, phase changes and melting; (b) phenomena related to the presence of a neutron flux, i. e. cladding hardening, embrittlement, axial growth and void swelling; (c) phenomena related to fissioning, neutron capture and the generation of fission products, i. e. (non-uniform) heat generation, the generation and release of fission gas (Xe and Kr) and helium, and fuel densification and swelling; (d) phenomena related to microstructural changes in the fuel, i. e. formation of high burnup structure, grain growth and restructuring; (e) phenomena related to radial temperature gradients in the fuel pellets, i. e. pellet cracking and fuel fragment relocation, pellet wheatsheafing, axial extrusion, dish filling, oxygen migration and plutonium redistribution; (f) chemical phenomena, i. e. fuel-clad bonding, stress-corrosion cracking and cladding oxidation, erosion and dissolution.

The active length of the fuel pin, i. e. the part containing the fuel pellets or bar, is usually represented by a series of axial zones (or segments). In each axial zone the fuel is divided into radial annuli (or rings), usually of equal volume, but sometimes of equal thickness. The cladding may also be divided into two or more annuli, especially if liner or duplex cladding is being simulated. The free volumes associated with the fuel-clad gap, pellet dishes and chamfers, pellet cracks, the pellet/bar bore (if any) and any upper and lower plena are also generally modelled.

Further details of the thermo-mechanical modelling depend upon the sophistication of the simulation, which is typically summarised as 1/-D, 2-D or 3-D, and whether the code is steady state (where stored heat is ignored) or transient (where stored heat is taken into account).

In the 1/-D representation, only radial, i. e. no axial or circumferential, heat flow is assumed and the fuel annuli are all considered to be subject to the same axial strain (the so-called plane strain assumption). The latter, in conjunction with an assumption of axi-symmetry, allows shear stresses to be ignored, such that only the principal stresses along the radial, circumferential and axial directions are non-zero. The thermal (energy conservation) and mechanical (force balance, stress-strain relationship and strain-displacement relationship) equations are typically solved by a finite difference scheme. Coupling between the axial zones (which explains the ‘half dimension’ in the 1/-D) is restricted to the coolant enthalpy, pin internal pressure and gas transport. A 1/-D code cannot simulate phenomena caused by shear stresses, such as pellet wheatsheafing, clad ridging, axial extrusion and pellet cracking (although the fact that the pellets are cracked is taken into account, as described below). An example of a 1/-D code is TRANSURANUS (Lassmann, 1992).

In the 2-D representation, which is effectively only applicable to pelleted fuel, there is radial and axial modelling of a fuel pellet in each axial zone (axi-symmetry, but not plane axial strain, is still assumed). The thermal and mechanical equations are typically solved by a finite element technique. The advantage is that 2-D phenomena such as pellet wheatsheafing, clad ridging and axial extrusion can be modelled explicitly. The disadvantage is the increased complexity and therefore also the slower running time. An example of a 2-D code is FEMAXI (Suzuki and Uetsuka, 2002).

In the 3-D representation, there is full three-dimensional modelling of the fuel pellets/bar and cladding. As in the 2-D representation, the thermal and mechanical equations are typically solved by a finite element technique. The advantage over 2-D codes is that phenomena such as the azimuthal cladding stress concentration over radial fuel cracks, or pellet-cladding eccentricity, which cannot be modelled when axi-symmetry is assumed, can be simulated. The disadvantage is the increased complexity and therefore also the slower running time. Due to the intricacies of the 3-D representation, advanced numerical techniques are generally required in the solution scheme. An example of a 3-D code is BISON (Hansen et al., 2009).

In reality, codes are often some hybrid of the 1/-D, 2-D or 3-D representations. For example, in the ENIGMA code for modelling LWR and AGR oxide fuel, which nominally has a 1/-D representation: (a) the effects of shear stresses are approximated using models for axial extrusion and for pellet wheatsheafing, which feed back calculated strain increments into the main solution scheme (Gates et al., 1998); (b) the azimuthal cladding stress concentration over radial fuel cracks is calculated using a parasitic model (Jackson et al., 1990). Thus, the key phenomena, which cannot be modelled with the 1-D plane axial strain assumption employed in the code’s main solution scheme, are instead modelled by other means.

The cracked nature of fuel pellets complicates the mechanical analysis of ceramic fuel. There are two main approaches to modelling the effects of pellet cracks on the stresses and strains (Bailly et al., 1999), both of which are only approximate. The first method models directionally dependent (anisotropic) fuel elastic constants (Young’s modulus and Poisson’s ratio). The second method models ‘crack strains’ (the dimensions of the cracks as fractions of the corresponding pellet dimensions) in the stress-strain relations, which relieve the stresses when the rupture stress is exceeded. With 2-D and 3-D codes there is a, potentially more accurate, third approach, which is to model the cracking itself, together with the resultant effects on stresses and strains — this has been demonstrated (albeit with a commercial finite element software package) by Williamson and Knoll (2009).

The thermal and mechanical equations and their solution for a typical F/2-D fuel performance code, together with modelling the effects of pellet cracking, are described in detail by Olander (1976).

Spent fuel, after decay storage, is transported to the reprocessing plant using dedicated casks (Fig. 16.3(a)) and transferred to storage pools (Fig. 16.3(b)) where the fuel may be allowed to cool for a further 3 to 5 years. The consequent decrease in activity facilitates subsequent processing. The spent fuel is covered by about 4 m of water, which provides protection to staff. The demineralized water in the pool is constantly cooled and filtered.

When a fuel element has cooled sufficiently, it is removed from the pool. The first step is to break the fuel cladding in order to expose the fuel to a nitric acid solution. This is usually achieved by mechanical cutting (or shearing) of the fuel element. The result is a mixture of fuel rod sections of a few centimetres length and of debris largely consisting of broken fuel pellets. A trough enables the fragments to fall into a dissolver containing hot concentrated nitric acid.

The choice of nitric acid is underpinned by following criteria:

• very low corrosion of stainless steel, from which much of the equipment is made

• volatility of the acid that allows it to be removed by evaporation

• the high solubility of most nitrate salts

As concentrated nitric acid is an oxidant, most of the elements when dissolved are

at their maximum degree of oxidation (uranium is at level of valence 6). Plutonium,

however, is at level 4.

Most of the elements within the fuel are dissolved except:

• Gaseous fission products (krypton, xenon, iodine, some C-14), which emanate as off-gas from the dissolver along with steam and nitrogen oxides (NO and NO2), which that are formed by the reaction of nitric acid with the fuel. These gases are routed to the ‘head gases’ treatment unit where nitrogen oxides are recombined into nitric acid and returned to the process.

• Metallic fuel element components — hulls, end-pieces, grid debris, etc. — that remain undissolved owing to their composition (stainless steel alloys, zircaloy) are separated from the solution and are rinsed. They can be then compacted and conditioned as wastes.

• Insoluble fission products or fines remaining from dissolution. These mainly consist of ruthenium, molybdenum, rhodium, palladium and technetium and cladding particles produced by the shearing (shearing fines). These are usually separated by centrifugation, a process known as clarification. This separation prevents the fines from affecting the extraction operations.

In summary, from the point of view of the physics, most options seem to be viable, but the consequences on the fuel cycle (cost, feasibility, doses to workers, etc.) seem to be more crucial. For example, apart from an increase of costs due to the need to over-enrich the core fuel of a LWR and some limitation in the amount that may be loaded to avoid a deterioration of the reactivity coefficients, the recycling of grouped TRUs in this type of reactor is practically excluded due to the huge increase of neutron doses at fuel fabrication.

Similarly, in the case of FRs, the constraint on grouped TRU recycling due to neutronics and core safety is in principle manageable. However, even if much less dramatic, the consequences for the fuel cycle (e. g. at fuel fabrication) of the TRU loading in the fuel can be significant with dedicated TRU-burning FRs (critical or sub-critical, heterogeneous or homogeneous) being the most affected.

The development of appropriate fuels, loaded with more or less TRUs, has proven to be a very severe task; this is especially the case for viable U-free fuels (also known as fertile-free fuels or inert matrix fuels). In any event and whatever the strategy, MA-loaded fuel development is a very significant challenge that needs R&D and the availability of appropriate irradiation facilities.

Finally, the transmutation of LLFPs does not seem to be realistic in view of the limited transmutation rates and the burden, in terms of reactivity loss during the cycle, on core performance.

Importance of hydrogeology

Because water is the main vector for return of radionuclides to the surface, hydrogeology is the prime consideration in the determination of a suitable site for

deep disposal. Geology is also important, of course, but mostly because of the need to avoid sites that may be unstable due to volcanic and seismic effects and for its potential to influence hydrogeology. The driver for groundwater movement is gravity so that infiltration at a topographic high will cause sub-surface water to flow towards a drainage point such as a spring line, a river or the sea (Fig. 18.4). Even when basement rock extends all the way to the surface, flow is likely to be concentrated in the upper levels for a number of reasons: the near-surface rock may be weathered and therefore more permeable; the deeper rocks are compressed by the overburden leading to lower permeability; and the flow path through the deep rock is longer so that it has higher flow resistance. These circumstances may lead to a situation in which deep groundwater moves very slowly so that salts, leached out of the rock, form a concentrated solution. This increases the density of the water and produces negative buoyancy that may be a further hindrance to flow at depth. In the absence of salt-laden water at depth, a borehole drilled directly from the surface into the repository may exhibit artesian flow.