The zirconium oxide thickness can be measured by the eddy current (EC) technique. However, before measuring the fuel rod oxide thickness it is important to remove as much crud from the rods as possible. This may be done by brushing the rod before the oxide is measured by the EC probe. If crud is not removed, the oxide thickness value normally obtained is too large. In many cases the crud deposited onto the fuel rods may be ferro­magnetic and therefore the EC technique may fail to give reliable results unless the EC equipment has been designed to compensate for ferromag­netic crud. Eddy Current technique can also be used to detect fuel claddings defects, for example non-penetrating cracks.

T. COUVANT, EDF R&D, France

DOI: 10.1533/9780857097453.1.70

Abstract: Corrosion is one of the major obstacles to extending the lifetime of nuclear power plants within agreed safety requirements. A large variety of the structural metals present in primary and secondary circuits of pressurized water reactors (PWRs) suffer corrosion. Uniform corrosion, flow-accelerated corrosion (FAC), pitting, stress corrosion cracking (SCC), environmentally assisted fatigue and hydrogen embrittlement can all affect the major components of PWRs, despite stringent selection of materials for component manufacture. Remedies can vary: adjusting water chemistry, reducing superficial strains and stresses, replacing materials or changing microstructures. Experience in the field has demonstrated that increasing chromium content is an efficient strategy: to date nickel alloys containing 30% chromium exhibit very good resistance to corrosion such as SCC. It can be shown that tendency to corrosion can largely depend on manufacturing conditions.

Key words: corrosion, austenitic alloys, pressurized water reactors, primary water, cracking.

2.1 Introduction

We begin the chapter with an outline of the history and fundamental prin­ciples of corrosion.

2.1.1 History

Corrosion and its effects have been observed since the first steps in metal­lurgy. Corrosion damage increased with the use of iron over the centuries. In 1830, de la Rive (1801-1873) showed that bimetallic junctions suffered fast corrosion due to impurities present in zinc. Later, Faraday (1791-1867) cor­related the current flow with the associated rate of corrosion. In the 1930s, Wagner (1901-1977) showed that the uniform dissolution of metals did not require separate anodic and cathodic sites but that metal dissolution and the accompanying cathodic reaction can occur randomly with respect to space and time over the surface. In the 1950s, Pourbaix (1904-1998) edited a series of major diagrams giving the domain of stability of many elements as a function of potential and pH.

From the 1950s the importance of corrosion to the economy became increasingly evident. Today, corrosion is one of the major degradations to overcome in order to extend the lifetime of nuclear power plants in agree­ment with safety requirements.

Kachanov represented continuum damage as an effective loss in mate­rial cross-section due to the formation and growth of internal voids. Consequently the internal stress corresponding to a nominal externally applied load increases with increasing damage. Kachanov assumed that damage could be represented by a quantity which he called the ‘continuity.’ The continuity is essentially the ratio of the remaining effective area A to the original area A0 . With accumulation of damage, the resulting internal stress (o) increases from initial value o0 to a value given by

°t = °o A [3.54]

The continuity term was later modified by Rabotnov and was called the damage parameter a, where

[3.55]

By assuming a power-law dependence of stress, the creep rate at constant temperature was described as

where m and p are material parameters. At time t = 0, a = 0 and the above equation assumes the power-law form. As a increases, the creep rate increases and when it achieves a critical value, the creep rate tends towards infinity and failure follows.

In order to describe the evolution of damage, Kachanov assumed that dam­age is a function of the initial stress a0. This was later generalized by Rabotnov who assumed that the damage is instead a function of the instantaneous stress and described the rate of change of damage through the following:

dm _ BO

~dt ~ (1 — m)r. [3.57]

Solving the above two equations gives the creep strain in the following form:

[3.58]

where ec is the instantaneous creep strain, eR is the rupture strain, t is the time and tR is the time to rupture. The shape of the creep curve described by Equation [3.58] is as shown in Fig. 3.23.

The damage tolerance parameter X is given by the following equation:

Я

1 + r — p

The material fails in the steady-state creep regime when X = 1. Ashby and Dyson97 have demonstrated that each damage micromechanism has a char­acteristic X and a characteristic shape of the creep curve. This implies that

the creep curve would assume different shapes for different values of X. Phaniraj et al.9S have established a correlation between the ratio of time to Monkman-Grant ductility (tMGD) and time to rupture (tR) and the damage tolerance parameter as given by

Figure 3.24 is based on this Equation [3.60] and shows that tMGD/tR is essen­tially constant for X > 4. The fMGD was suggested as time for onset of true tertiary creep damage and was considered to be an important parameter in identifying the useful creep life of a material. It also describes the time for which minimum creep ductility is ensured. Hence Phaniraj et al. contend that the stress to cause tMGD in 105 h can be used as a useful design criterion for creep of elevated temperature components.

Before concluding we present a few examples where the concepts dis­cussed in the previous sections may not be directly applied. Rather subtle modifications to the models are necessary in order to simulate the actual behavior of the material.

3.4 Case studies illustrating the role of other factors

In the following section, the effects of impurities, second phases and multi­axial loadings on creep of materials are discussed with examples taken from various classes of materials including ionic solids.

Figure 4.61 gives schematic growth curves for Zircaloy illustrating several points. Note that L-textured (longitudinal, or in the original rolling direction) material grows, while T-textured (transverse to the rolling direction) mate­rial shrinks; when taking into account shrinkage of a component in the third direction (N, normal to the rolling direction), this behaviour results in approx­imately constant volume. The long direction (L) of a component is the most important: for instance the length of a fuel rod, channel box or GT. Note that cold worked (CW or SRA) material grows at a high and almost linear rate, while recrystallized (RXA) material grows in a 3-stage process, with the final high rate being called ‘breakaway’ growth. The various stages can be directly correlated to the irradiation-produced microstructure described earlier. For RXA Zircaloy, at low fluences where only <a> component loops exist, growth is small (~0.1%) and saturates. When <c> component loops begin to appear the growth rate increases and becomes nearly linear with fluence in the range 6-10 x 1025 n/m2, E >1 MeV. For L-texture material growth can reach 1% at 20 x 1025 n/m2. In initially cold worked (CW) or stress relieved material (SRA), <c> component dislocations occur as part of the deformation-induced struc­ture and more are formed during irradiation (Holt et al, 1996). The growth

rate is nearly linear with fluence and the magnitude is almost linear with the amount of initial CW. In heavily-worked material (typically 70-80% in a fuel rod) a growth of 2% can be reached by 20 x 1025 n/m2 (corresponding to a burnup of about 100 MWd/kgU). Figure 4.62 gives some values of irradia­tion growth for Zircaloy materials of different heat treatments, reflecting the amount of residual CW and dislocation density. An overview of factors affect­ing growth is given by Fidleris et al. (1987).

Texture

It can be argued (Hesketh et al, 1969; Alexander et al, 1977) that the magni­tude of growth strain in any given direction of a polycrystalline material can be related to the crystallographic texture and is proportional to a growth anisotropy factor Gd, given by

G=l-3fc, [4.1]

where fd is the resolved fraction of basal poles, fc, in the d-direction. The anisotropy factor depends on the assumptions that each grain behaves as an independent single crystal and that the volume change due to irradiation growth is zero.

At high burnup and high temperature (greater than about 360°C, 633K) and perhaps also in a heavily cold worked material, the familiar (1-3f) and

constant volume assumptions may not be valid. At low fluence the two assumptions are reasonable, but at high fluence the transverse strain is not zero (as would be predicted by the fx value) and the sum of the strains is strongly positive. It is also noted that cold worked material and recrystal­lized material have similar growth behaviour at high temperatures. It is fur­ther noted that the temperature of this irradiation is at the upper range (>378°C) expected for even a hot PWR.

For the high temperature data presented in Fig. 4.64, STEM studies revealed grain boundary cavities and occasional IG voids (Tucker et al. ,1984) which may explain the observed change of volume. Other studies have not reported cavities or voids at very high fluence at 290°C (363K) (Mahmood et al, 2000) or high fluence at 350°C (623K). Holt & Causey (2004) reported that for Zr-2.5Nb there is a small volume increase (0.05-0.1%) at low flu — ence, but at high fluence the volume change was close to zero.

Visual inspection involves examining the cable throughout its length during a formal plant walkdown, a useful practice when, as is often the case, degra­dation is visible to the naked, well-trained eye (IAEA, 2011). Visual inspec­tion can identify changes in physical/visual appearance, surface texture, and damage as a result of manufacturing or operation (U. S. NRC, 2010b). More sophisticated techniques can then be used to determine the degree of age­ing more accurately.

The advantages of visual inspection are that it is low cost and easy to perform, requires no specialized equipment, does not require that samples be removed from the cable, and can be performed on operating equipment in-situ. Its disadvantages include the requirement that the cable be accessi­ble and visible, inspectors must be trained to evaluate what they are looking at (subjectivity), it generally only provides information on the cable jacket, and it does not provide quantifiable results (no trending possible) (AMS Corp., 2010; IAEA, 2011).

Mechanical testing is a subset of life-testing techniques that involves inspecting cables for cracks or changes in color, texture or hardness, mass loss, visco-elasticity properties, or size (swelling, shrinkage, deformation). Among the most conventional and popular means of mechanical cable testing are mea­suring the elongation-at-break of the cable and its tensile strength when pulled apart. The elongation-at-break test measures the strain on the cable when it breaks and is a recognized standard for assessing the health, integrity, and func­tionality of a cable insulation material (IAEA, 2011). This test is performed by stretching a ‘dog bone’-shaped cable sample until it breaks. The elongation-at — break test yields information on the tensile strength and modulus of elasticity of the cable, but the percentage of elongation is the most important criterion in evaluating cable health. When the percentage elongation-at-break is less than 50%, the cable is considered to be unhealthy — potentially unable to survive DBA conditions (AMS Corp., 2010; IAEA, 2011).

The tensile test measures the stress needed to break the cable. For poly­meric materials like thermoplastics, tensile strength only begins to fall after substantial ageing has already occurred. Both the elongation-at-break and tensile strength tests can be performed using a tensile testing machine.

A third mechanical test, measuring compressive modulus, involves check­ing the ductility of the cable insulation or jacket material to determine if the cable has become dry, brittle, or prone to crack. Developed in the mid-1980s by the Electric Power Research Institute (EPRI), this test is performed with a device known as a cable indenter, which uses a small probe to press against the cable jacket or insulation. A PC-based system analyzes cable hardness by measuring the probe force and polymer deformation, thus pro­viding diagnostic insights (Hashemian, 2010).

The difficulty with these classic life-testing techniques is that they can check for problems only at the locations on the cable where the cable is tested. Such passive maintenance methods can thus fail to detect problems or hot spots in other areas. Similarly, the elongation-at-break and tensile strength test are also destructive to the tested material and require that the cable be removed from operation for testing (IAEA, 2011). For these reasons, mechanical life-testing techniques should be combined with other measurements, such as electrical or chemical functionality.

The steady-state strain rate of creep deformation, at a given temperature, has been found to be directly dependent on the applied stress. The functional dependence of strain rate on stress can be expressed by Norton’s law13

[3.13 ]

where K isa constant and n is the stress exponent. Similarly, for a constant applied stress, the rate of creep deformation increases with increasing tem­perature. The effect of temperature can be understood by including an extra term indicated in the following equation

[3.14]

where Kj is another constant and Qc is the activation energy of creep defor­mation. The activation energy term is included due to the fact that the creep deformation is considered to be a first order reaction rate process. The mag­nitude of the activation energy is dependent upon the physical mechanism governing the deformation process.

The effect of stress and temperature is clearly illustrated by Fig. 3.3. With increasing stress and temperature, the instantaneous strain at the time of stress application increases, the steady-state creep rate is increased and the rupture lifetime is diminished.

R. B. ADAMSON, Zircology Plus, USA and P. RUDLING, ANT International, Sweden

DOI: 10.1533/9780857097453.2.151

Abstract: This chapter highlights the various uses and properties of zirconium alloy cladding and structural components used in nuclear power light water reactors. Specific attributes including dimensional stability, corrosion resistance, irradiation effects and mechanical properties are discussed in detail.

Key words: zirconium alloys, nuclear reactors, dimensional stability, radiation effects, mechanical properties, corrosion.

4.1 Introduction

Zirconium alloys are used as the prime structural material in light water reactors (LWRs). As such, they have to meet several requirements: low neutron absorption cross section; corrosion resistance in 280-350°C water; resistance to radiation in both mechanical behaviours and dimensional sta­bility; reasonable strength, ductility and fabricability; affordable cost; and availability in large quantities.

Unalloyed zirconium was used as the structural material in the prototype core for nuclear submarines in 1953 (Rickover, 1975 in Adamson, 2010). However, variability in corrosion resistance, strength and cost issues prompted development of a stronger, more corrosion-resistant alloy named Zircaloy-2. This alloy was used in the first nuclear powered submarine, Nautilus 1954, and in the first commercial electricity-generating reactor, Shippingport 1957. Today, a variety of zirconium alloys (see below for details) are used in all LWRs throughout the world.

This chapter covers the following topics relevant to the uniqueness of zirconium alloys: Section 4.2 on fuel assembly design; Sections 4.3-4.6 on material and performance issues; Section 4.7 covers future trends in materi­als; and Section 4.8 provides sources of further information.

Zirconium and hafnium (used as a neutron absorber) are unique among materials used in LWRs in that they have the hexagonal close packed (HCP)

151

Degradation of failed fuel rod is a situation where the leakage path(s) through the damaged cladding increases to the point where the fuel itself is dispersed into the primary system (Strasser et al., 2008). This may occur if the rods degrade to such a point that the water contacts the fuel pellet, par­ticularly if the contact also involves active flow of the water over exposed fuel pellets, one example being a large axial cladding crack. Steam will not be able to cause fuel washout while water can by oxidising the fuel grain boundaries thereby causing disintegration f the fuel grains. Normally, util­ities are much more concerned about fuel washout than high iodine and noble gas release. This is because it may take up to ten years to clean the core from the tramp uranium resulting from the fuel dissolution, while the high iodine and noble gas activities released from the failed rod will be elimi­nated when the failed rod is extracted from the core.

Degradation has historically been more of an issue in BWRs than in PWRs (Strasser et al. , 2008). Failed rods in PWRs may degrade, but the amount of dispersed fuel is lower than in a BWR. The rationale may be that the coolant chemistry in a PWR is more reducing than in BWRs. During the period 1992-93, six plants in the United States and Europe were forced into unscheduled outages because of concerns about failed Zr-sponge liner fuel (IAEA, no. 388, 1998). This is a liner produced from Zr sponge material to which no alloying elements have been added; its major impurities are oxygen (about 600-900 wt. ppm) and iron (about

150-500 wt. ppm). In all these cases, the very high off-gas activities and significant loss of fuel pellet material resulted from only one or two failed rods. Other plants in the United States and Europe also elected to shut down during and slightly after this interval to remove failed fuel assem­blies and avoid the risk of large residual contamination from tramp ura­nium. More recently, the risk of degradation and residual contamination has been reduced by the use of corrosion-resistant liners in BWR fuel to the extent that forced and voluntary outages are less common.

Two different types of degradation scenarios have been identified, namely the development of two different types of cracks (Strasser et al, 2008):

1. Transversal breaks (also called guillotine cuts or circumferential break) occurring in BWRs, PWRs and VVERs.

2. Long axial cracks (axial splits), which can occur in BWRs due to the movement of control blades but may also occur in PWRs that are sub­jected to significant control rod movements during operation. Axial split is a term introduced by GE and represents a failed rod that either has an off-gas level larger than 5000 jiCi/s (185 MBq/s) or a total crack length that is larger than 152 mm (6 inches).

Transversal breaks in BWRs — normally occur in low to intermediate bur — nup rods in the bottom part of the rod with a primary failure in the upper part of the rod (see Fig. 5.6) (Strasser et al. , 2008). The primary defect will allow water/steam to gain access to the rod interior (1 in Fig. 5.6) where the steam will oxidize the fuel clad inner surface forming a zirconium oxide the thickness of which will decrease with distance from the primary defect (2 in Fig. 5.6). At the same time a hydrogen partial pressure is being built up in the pellet-cladding gap. At a critical distance from the primary defect, the steam partial pressure will be insufficient to protect the clad inner surface

JzrH^©

from hydrogen ingress thus causing secondary hydriding (3 in Fig. 5.6) (e. g. Olander et al., 1997). If the hydride precipitates along the whole fuel clad circumference, the fuel rod may fracture transversally due to the hydride embrittlement effect (4 in Fig. 5.6).

Transversal break in PWRs/VVERs — are caused by a mechanistic develop­ment similar to that of BWRs (Strasser et al, 2008). However, the second­ary hydride defects tend to form in the upper part of a PWR/VVER rod. The processes involved in developing a transversal break in a PWR rod are shown in Fig. 5.7 .

1. Axial cracks in BWRs — Formation of long axial cracks has three prereq­uisites, (Strasser et al, 2008):

la. A sharp primary defect such as a PCI crack or cracks in hydride blisters formed due to a primary defect. However, in this case the hydride blister is very local and does not exist along the whole fuel clad circumference, as seen in formation of transversal breaks.

lb. A fuel cladding hydrogen content larger than the hydrogen solid solubility.

lc. A stress intensity (K,) at the crack tip above the critical value for crack extension. K, will increase with clad tensile stress level which in turn depends on:

pH2/pH2O >(pH2/pH2o)critica^ ZrH166

‘2

5.7 Schematic description of the events resulting in transversal break formation (Strasser et al., 2008).

1c1. The initial pellet-cladding gap prior to the power ramp, which depends on:

lcla. Burnup since the gap is decreasing with increased bur — nup due to fuel swelling and fuel clad creep-down. This is the reason that axial cracks do not form in low burnup fuel since the fuel pellet-clad gap is so large.

lclb. The corrosion properties of the cladding inner surface (Edsinger, 2000). The pellet-cladding gap decreases if the corrosion properties of the cladding inner surface are poor, resulting in formation of a thick porous oxide layer in the failed rod. The decrease in gap is related to the zir­conium oxide having a larger specific volume than that of the zirconium metal. It also turns out that, if the corro­sion resistance of the cladding inner surface is poor, then formed oxide is less dense due to the many cracks and pores which will decrease the pellet-cladding gap further. The first type of Zr-liner materials used in the nuclear industry were non-alloyed with very poor corrosion prop­erties. Once it was realized that the corrosion properties of the Zr liner have a large impact on the tendency to form axial cracks in failed fuel, all fuel vendors did alloy their liners to improve the corrosion resistance. However, it is important to ensure that the alloying additions will not degrade the PCI performance of the fuel cladding.

1c2. The magnitude of the rod power increase.

The axial split formation is schematically shown in Fig. 5.8 (Strasser et al, 2008). Initially, the control rod is inserted during the time when the primary defect occurs (1 in Fig. 5.8). The same scenario as for transversal breaks in BWRs occurs, but the secondary hydrides are distributed to several fuel clad locations which means that each hydride becomes too small to encom­pass the whole fuel clad circumference (2 in Fig. 5.8). The tensile stresses in the cladding which are necessary for crack propagation result from a power increase in the failed rod, for example, when a control rod adjacent to the failed rod is pulled out of the core. This will increase the temperature in the fuel stack resulting in a thermal increase of the pellet diameter. If these stresses become large enough the sharp defect may propagate if the result­ing K: exceeds the critical value for crack propagation (3 in Fig. 5.8). It is proposed that the mechanism for crack propagation forming an axial split is a delayed hydrogen cracking (DHC) type failure process (see e. g. Efsing & Pettersson, 1998; Edsinger, 2000; Lysell et al, 2000 for more details). The lower bounds of the crack velocities are in the range 4 x 10-8-5 x 10-7 ms-1 based on assumed constant growth rates in the time between first detection of the defect and removal of the fuel (Strasser et al, 2008).

5.8 Schematic showing the events resulting in axial split formation. The numbers in the figure relate to the sequence of the different events that may lead to an axial crack as described in the text (Strasser et al., 2008).

Axial cracks in PWRs/VVERs — Long axial cracks do not form in PWRs as readily as in BWRs (Strasser et al, 2008). The reason for the difference is that in PWRs, the power regulation is done slowly and without pronounced increases in local power by decreasing the boron coolant concentration, while power regulation in BWRs is done by a combination of control rod movements and variations in coolant flow, with the control blade move­ments leading to rapid increases in local power. However, axial cracks may form in PWRs/VVERs by essentially the same mechanism as formation of long axial cracks in BWRs due to (Strasser et al., 2008 ):

• A class II transient and/or

• Due to control rod movements in load-following plants.

5.3 Materials performance during accidents

Having considered normal operating conditions, we now move on to cover accident scenarios.

Fiducial markers are generally employed to study the contribution of grain boundary sliding to the total creep strain.5 3 GBS leads to shearing of the fiducial markers and the shear offset provides a measure of the strain contri­bution. Since the stress concentrations developed during sliding are relieved by dislocation emission, dislocation activity can be expected in the vicinity of the grain boundary. Recent work by Gollapudi et al.51 shows increased dis­location activity close to the grain boundary during deformation controlled by GBS. At the same time, dislocations emitted from a grain boundary are expected to travel across the grain until they encounter a grain boundary. These dislocations subsequently pile up which is relieved by dislocation climb. Dislocation pile-up close to the grain boundary was also observed by Gollapudi et al.51 Figure 3.9a and 3.9b provide microstructural features associated with creep in the GBS regime.

Knoop microhardness in a hot cell was used to determine hardness of zir­conium and Zircaloy as a function of fluence, purity and irradiation temper­ature (Tucker & Adamson, 1984). The ranking of hardness was the same in both unirradiated and irradiated materials. Hardness saturated with fluence, as it also does for tensile properties. Figure 4.26 gives some of the data. The general trends are the same for all zirconium alloys.

The effects of post-irradiation annealing on mechanical properties and radiation damage are given in an earlier section. In general both

4.25 Zircaloy-2 ductility as a function of irradiation and pre-irradiation annealing temperature for the simple tension plane stress and notched plane strain specimens shown. (Source: Reprinted, with permission, from Tomalin (1977), copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA 19428.)

characteristics anneal out at temperatures above about 400°C, although the rate of annealing is more sluggish for Nb-containing alloys.