As long as cables are installed properly and not exposed to environmental conditions beyond their design basis, they are generally durable and rel­atively long lived, typically lasting 40-50 years (Hashemian, 2010; IAEA, 2011). In fact, compared to other I&C components, cables have historically experienced few problems. A Japanese study, for example, found that most nuclear power plant I&C cables will maintain their electrical function capa­bilities over 60 years of operation (Hashemian, 2010).

The IAEA defines a ‘mild’ operating environment as one that ‘would at no time be significantly more severe than the environment that would occur during normal plant operation, including anticipated operational events.’ In contrast, a ‘ harsh ’ environment is one that results from a design basis accident (DBA) involving, for example, a loss-of-coolant accident (LOCA) or the failure of a high-energy line or main steam line (IAEA, 2011). Mild and harsh operating environments can be distinguished from unanticipated operating conditions such as those caused by poor installation, operation or movement of the cable. All of these can accelerate cable ageing and degra­dation (IAEA, 2011).

Cable ageing is a subset of cable degradation and primarily consists of cracking, embrittlement, or other changes to the cable jacket or insulation material. In most cases, these changes are produced by a combination of physical age and environmental stressors such as temperature or radiation exposure (AMS Corp., 2010). Cable circuits can be subjected to any or a combination of the following stressors: oxidation, water intrusion, contami­nation, vibration, thermal variations, electrical transient, voltage variations, temperature, installation damage, and handling and physical contact (AMS Corp., 2010). However, the three principle ageing factors for cables are (1) elevated ambient temperature or humidity; (2) cyclic mechanical stress; and (3) exposure to radiation (Hashemian, 2010). Cable degradation is mainly dependent on environmental factors such as temperature, radiation, humid­ity, or contaminants (IAEA, 2011).

K. L. MURTY, North Carolina State University, USA, S. GOLLAPUDI, Massachusetts Institute of Technology, USA, K. RAMASWAMY, Bhabha Atomic Research Center, India, M. D. MATHEW, Indira Gandhi Center for Atomic Research, India and I. CHARIT, University of Idaho, USA

DOI: 10.1533/9780857097453.1.81

Abstract: The time-dependent deformation of materials or creep governs the useful life of many engineering structures. It assumes even higher significance in the case of structures constituting a nuclear reactor, wherein materials bombarded with neutrons develop defects that assist faster diffusion leading to greater plastic deformation. As a result, an understanding of the creep deformation process and factors controlling it is necessary for gauging the usefulness of materials in a nuclear reactor as well as for predicting life-times of various structures. Thus in this work we discuss the various mechanisms of creep, the rate controlling factors, deformation mechanism maps and useful life prediction methodologies. We also identify a few cases where direct application of simple creep correlations might not be feasible. Finally, we discuss the various factors that control the creep behavior of materials in light water reactors.

Key words: creep, diffusion creep, dislocation creep, deformation mechanism maps, modeling, zirconium alloys, stainless steels, irradiation creep.

3.1 Introduction

Creep is time-dependent plastic strain under a constant load/stress at a given temperature and often becomes the life limiting criterion for many structures that experience loads and temperatures, and becomes signifi­cant for materials in light water reactors (LWRs) due to imposed radiation effects. A thorough understanding of the plastic deformation behavior of materials is essential for the sound design of engineering structures. Fail-safe designs are based on the ability to predict the response of a structure to applied loads and ensuing plastic deformation. While brittle materials such as ceramics fail after relatively low plastic strains, a significant number of engineering materials such as metals and alloys are characterized by large scale plastic deformation leading to failure. The extent of deformation is controlled by intrinsic factors such as bond strength, presence of secondary phases and defect concentration. At the same time extrinsic factors such as applied loads, temperature, deformation rates and geometry of the structure also determine the amount of plastic deformation. It has been well estab­lished that high applied loads and temperatures generally accelerate the rate of plastic deformation. This is because high temperatures and stresses provide the necessary activation energy required for defects to overcome barriers to plastic deformation. While plastic deformation at room tempera­ture or low homologous temperatures (T/Tm) occurs when the applied stress exceeds the yield stress ay, deformation at high temperatures can occur at stresses significantly smaller in comparison to the yield stress. The branch of metallurgy which attempts at understanding material deformability at high homologous temperatures and small applied stress has come to be known as creep. The kinetics of deformation processes become important with increasing temperatures and hence creep is defined as the time dependent plastic deformation of a material under constant load or stress.

The earliest studies on time dependence of plastic strain were carried out by Andrade.1 The time dependence of elongation under tensile loads was investigated at constant temperature. Andrade observed that the total deformation could be divided into three periods: (a) immediate extension upon loading (mainly elastic with relatively small instantaneous plastic), (b) an initial flow which gradually disappears and (c) a constant flow which takes place throughout the elongation. Subsequent studies by Hanson and Wheeler2 showed the presence of a period where the extension increases continuously until fracture. This period was found to occur following the period of constant flow and was understood to be due to decreased cross-sectional area accompanying the elongation. At constant loads, the cross-sectional area decrease leads to the increase in effective stress and a corresponding increase in strain rate.

Although addition of alloying elements is never based solely on the diffusiv — ity criteria, the resulting creep rate of the alloy is the outcome of the diffusiv — ity of the elements added and understanding the diffusion phenomenon in these alloys will help in fine-tuning the concentration of the solute added. For instance, though the strengthening effect of Nb in Zr increases with Nb con­centration, the optimum level to obtain a low steady-state creep rate (<10-8 /s) as measured from the stress was found to be around 2.5 wt.%.1 24 Nb with low diffusion rate and high solubility limit can effectively enhance the creep strength of Zr-2.5Nb alloy.125 The creep strengthening effect by molybdenum in zirconium is reported to be superior to that of niobium (for comparable alloy fraction) in the temperature range 350-600°C.126 Further, the modifica­tion in the creep mechanism by addition of Nb is clearly seen (Fig. 3.16) to transition from climb-controlled creep as in class-M to viscous glide creep as in alloy class by correlating experimental results on Nb-added Zr-alloy sheet127 and Zircaloy-4 cladding.128 Addition of 1 wt.%Nb to Zircaloy-4 reduces the creep rate by about 100 times and a region with n = 3 is introduced, absent in Zircaloy-4 and which behaves like pure metal (class M).

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

DOI: 10.1533/9780857097453.2.246

Abstract: This chapter highlights integral performance of zirconium alloy fuel bundle components used in nuclear power light water reactors (LWRs). In particular we focus on those behaviours which result in performance issues, and in experimental techniques which are used to quantify the performance. Details in this chapter complement those in the previous chapter on the properties of in-reactor zirconium alloy materials.

Key words: zirconium alloys, nuclear reactors, accidents, dimensional stability, irradiation, mechanical properties, corrosion, inspection, high burnup.

5.1 Introduction

The previous chapter described material properties of zirconium alloys in light water reactors (LWRs). The performance of fuel bundle components is often driven by a combination of singular material properties; for example, mechanical strength and irradiation creep. This chapter extends material behaviour to include integral performance of fuel bundle components such as fuel rods, channels and guide tube assemblies under both normal opera­tional and accident conditions. In all cases, component inspection is needed to verify expected or explore abnormal performance. A thorough under­standing of both material and component behaviour is needed to assure safe and efficient reactor operation.

5.2 Materials performance during normal operational conditions

We proceed with sections covering ways in which materials perform during normal operating conditions in the main reactor types.

246

Cable testing and diagnostics is not a new field, and many standards have been developed that provide guidance on different methodologies for assessing their performance (AMS Corp., 2010). Over the years, the nuclear industry has suffered from a variety of plant issues resulting from reactive cable age­ing management practices and failed cables, including plant trips, damage to plant equipment, radiation exposure to maintenance personnel, increased outage activity, and more (AMS Corp., 2010). Because cables have been the source of serious accidents, national and international organizations such as the U. S. Department of Energy (DOE), U. S. Nuclear Regulatory Commission, International Atomic Energy Agency (IAEA), International Electrotechnical Commission (IEC), American Society for Testing and Material (ASTM), Institute of Electrical and Electronics Engineers (IEEE), and Electric Power Research Institute (EPRI) have sponsored research and development projects, developed cable testing techniques, and written stan­dards and guidelines to preserve the integrity, health, and reliability of the cables used in nuclear and other applications (Hashemian, 2010).

As a result, today cable testing is often recommended in nuclear indus­try standards and guidelines as a method for performing predictive mainte­nance and managing the ageing of I&C equipment. For example, the IAEA stipulates that ‘aged I&C cables are expected to fully function to carry the I&C signals to a control room for normal operation, Design Basis Event (DBE) management and recovery.’ Similarly, IEC standard 62465 (‘Aging of Electrical Cabling Systems’) outlines requirements for in-situ testing tech­niques to detect problems in cable conductors and cable insulation material. IEEE standards for testing fire travel and cables under fire conditions are similar to several of the IEC standards. There are also ANSI and ASTM standards covering general cable testing as well as specific cable tests such as partial discharge testing (AMS Corp., 2010).

Both the regulatory and industry pressure to manage cable ageing in light water reactors has only intensified as plants have been granted license renewal to operate cables for an extended qualified life as part of their efforts to extend the initial design life of a nuclear power plant from 30-40 years to 60 years (Hashemian, 2010). In some countries, plants have been able to replace some of their critical cables as an ageing management strategy. For example, the Beznau nuclear power plant in Switzerland has implemented a comprehensive cable maintenance program and has thereby emerged as a leader in cable ageing management in the worldwide nuclear power industry (AMS Corp., 2011).

Today, the nuclear power industry can obtain guidance for managing and testing plant cables and wiring from an extensive collection of various cable specifications and cable testing standards. This collection presents plant staff with a variety of recommendations for testing, monitoring, and managing the maintenance of plant cables (AMS Corp., 2010). However, because replac­ing cables is expensive, radiation intensive, and typically impractical, utili­ties operating nuclear power plants are not adopting wholesale replacement of cables as a strategy. Rather, they are searching for ageing management techniques that can identify cable problems and areas where maintenance or replacement is needed (AMS Corp., 2011).

On a regulatory level, cable ageing has not been an afterthought, and the U. S. NRC, DOE and others have sponsored ongoing research to improve currently available techniques so as to enhance preventative mainte­nance and proactive management of cable ageing. Others, such as the U. S. Department of Defense (DOD), NASA, the National Institute of Standards and Technology (NIST) and numerous international organizations have also sponsored and performed research and development (R&D) on cable condition monitoring and residual life estimation (AMS Corp., 2010).

As an example of increased regulatory concern over cable ageing, in February 2007 the NRC issued NRC Generic Letter 2007-01; ‘Inaccessible or Underground Power Cable Failures that Disable Accident Mitigation Systems or Cause Plant Transients. ’ This letter required responses from license hold­ers to the issues of undetected ageing problems associated with underground power cables that had resulted in plant shutdowns and unusual transients. After evaluating licensee responses, the NRC summarized the current plant circum­stances with respect to cable condition monitoring in its recommendations: ‘Plants undergoing license renewal have agreed to a cable testing program for the extended period of plant operation for a limited number of cables that are within the scope of license renewal, but only a few have established a cable test­ing program for the current operating period. The data… show an increasing trend of cable failures. These cables are failing within the plants’ 40-year licens­ing periods. Licensees have identified failed cables and declining insulation resistance properties through current testing practices; however, licensees have also reported that some failures may have occurred before the failed condition was discovered… The 10 CFR Part 50 regulations require licensees to assess the condition of their components, to monitor the performance or condition. in a manner sufficient to provide reasonable assurance that they are capable of fulfilling their intended functions, and to establish a test program to ensure that all testing required to demonstrate that components will perform satisfactorily in service is identified and performed’ (AMS Corp., 2010).

Regulators are increasingly urging that cable ageing be taken into account to ensure that plants continue to operate safely throughout the remainder of their original licenses and during any extended operation (AMS Corp., 2011). For example, the U. S. NRC published the Regulatory Guide 1.218 in April 2012 to describe the technique that the NRC staff considers accept­able for monitoring the performance of electrical cables that are important to safety (U. S. NRC, 2012). The title of this regulatory guide is ‘Condition Monitoring Techniques for Electric Cables Used in Nuclear Power Plants’.

The first microstructural evidence for diffusional creep was provided by Squires et a/.34 who carried out creep studies on Mg-0.5Zr at 123 and 113K. The initial microstructure had a uniform distribution of inert ZrH. particles and investigation of the microstructure of the crept specimen, as shown in Fig. 3.6a, depicts the presence of regions denuded of the inert par­ticles. These denuded zones mostly formed near transverse grain boundar­ies. Squires et a/.34 attributed the formation of denuded zones to diffusional creep of the Mg alloy. Under the application of a stress, Mg atoms diffuse from parallel grain boundaries to the transverse grain boundaries causing a slight elongation of the grains. The inert particles do not travel along with the Mg particles and their absence adjacent to the transverse grain bound­aries causes the formation of denuded zones. In subsequent years, denuded zones have been observed by other groups in Mg-Zr35 and Mg-Mn36 alloys. Even though the formation of denuded zones as a consequence of diffu­sional creep appears reasonable, it has been a matter of regular debate.31-40 Jaeger and Gleiter41 carried out experiments on a bamboo structured cop­per coated with Al2O3 film. Diffusional creep experiments were carried out on copper at a temperature of around 1348 K. At the conclusion of the creep experiment, it was observed that the alumina film fractured in a few places. The fracturing of the alumina film was ascribed to the deforma­tion incompatibility of the alumina film and the copper beneath. The copper grains elongated under the application of the stress whereas the alumina film did not deform to the same extent causing fracturing of the film. The

elongation of the copper grains was ascribed to diffusional creep deforma­tion. Recently McNee et al.42 carried out detailed experiments on OFHC grade copper in the diffusional creep regime. A surface scratch technique was employed to establish the operation of a diffusional creep mechanism. In addition to measurements of the surface scratch displacements, grain boundary grooves were identified and subsequently quantified through an atomic force microscopy (AFM). Grooves, as shown in Fig. 3.6b, were formed predominantly on boundaries transverse to the applied stress. Grain boundary grooves were thus suggested as a microstructural feature char­acteristic of diffusional creep. While denuded zones, elongated grains and grain boundary grooves are essentially features developed due to N-H or Coble creep, the features associated with H-D and S-N creep are different. Dislocations cross slipping43 as shown in Fig. 3.6c and slip-bands sheared by grain boundaries,17 Fig. 3.6d, are suggested to be evidence of H-D and S-N creep, respectively.

Irradiation temperature does have an effect on microstructure — for instance higher irradiation temperature results in larger <a> loops, <c> loops do not

form at 77°C (350K), and Zr2(Fe, Ni) SPPs do not become amorphous above an irradiation temperature of about 100°C (453K) (see e. g. Griffiths et al, 1996). In addition, post-irradiation temperatures cause effects that give insight to the microstructure stability.

Damage in the form of <a> loops appears to be stable in post-irradia­tion annealing conditions to about 400°C (673K). Figure 4.18 (Adamson & Bell, 1986) shows that 1 hour at 400°C is a threshold condition for damage in size and density of <a> loops. Above that temperature, or quite likely longer times at that temperature, results in a marked increase in loop size and decrease in loop density. A temperature of 550°C (823K) for 1 h is suf­ficient to reduce the loop density to zero. This is accompanied by a dramatic decrease in hardness, as discussed below. Complementary data (Cheng et al, 1994) indicate no changes in <a> loops after 200 days at 316°C (588K).

On the other hand, <c> component dislocations are quite resistant to change over the whole temperature range where <a> loops disappear. Yang (1989) and Kruger (1990) have shown that 1 h at 560°C (833K) or 575°C (848K) causes little or no change in <c> loop density or size. One hour at 675°C results in a 50% reduction in <c> loop density, while 1 h at 750°C (1023K) results in removal of all loops.

Figure 4.18 indicates hardness decreases in concert with changes in the <a> loop size and density. This is an indication that <c> loops do not have influence on the hardness. A summary is given by Adamson (2006). An addi­tional study (Ribis et al, 2007), confirms the results of Adamson and Bell

4.18 Post-irradiation microstructure (<a> loop density and size) and hardness of Zircaloy-2 irradiated to a fluence of 6.5 x 1024 n/m2 (E > 1 MeV). Upper: TEM after annealing at indicated temperatures. Lower: density, size and hardness as functions of annealing temperature (Adamson & Bell, 1986).

(1986), and add modelling equations for the recovery process. Bourdiliau et al. (2010) go a step further and show that there is a direct relation between recovery of hardness and recovery of ultimate tensile stress (UTS) for both SRA Zircaloy-4 and Zr1Nb. However the recovery for Zr1Nb is more slug­gish than for Zircaloy-4, as shown in Fig. 4.19. Zr1Nb does not fully recover the irradiation-induced hardening, primarily due the effects of the thermally stable, irradiation-induced phase which forms in that alloy.

Post-irradiation annealing also has effects on irradiation-affected SPPs. The observed phenomena give important insights into, for instance,

corrosion mechanisms. For Zircaloy Yang (1989), Kruger (1990) and Cheng et al. (1994) report that post-irradiation annealing causes SPPs to recrystal­lize, to regain Fe and Ni, and to form under specific conditions of time and temperature. Minimal effects are observed for 316°C (589K) for 30 days, but for 200 days significant amounts of Fe diffuse back to the precipitates. At 400°C (673K) Fe diffuses back to precipitates in less than 10 days, and Fe-rich precipitates form at grain boundaries. At higher temperatures >560°C (833K) amorphized SPPs recrystallize, Fe and Cr diffuse back to SPPs, and re-precipitation occurs in the matrix and grain boundaries. Recent studies by Vizcaino et al. (2010) tend to confirm the earlier results.

There are several reasons why poolside and/or hot cell examinations are undertaken on fuel assembly (FA) components (Rudling & Patterson, 2009). These are:

1 Root cause investigations of failed FA components

(a) A ‘failed’ FA component has a wider meaning in this respect. It not only means that the component has physically failed but it could also mean that the component does not behave satisfactorily, for example FA bowing that is so large that control rods cannot be inserted.

2 Maintaining good fuel reliability by:

(a) Providing baseline data before a change in operational environ­ment of the fuel.

(b) Getting early warnings of potential issues.

3 Fuel vendor design and licensing data such as:

(a) Providing data to material models and fuel performance codes

(b) Verification of the good performance of a new fuel design

(c) Assessment of the effects of changes in the operating environment; for example, water chemistry improvements or higher exposures

The in-pool examinations are usually non-destructive, but can also involve destructive operations such as breaking a flow tab off a spacer or cutting coupons from a channel for measurement of hydrogen concentration. Hot cell examinations normally start with non-destructive followed by destruc­tive examinations. The costs for hot cell examinations are much more expen­sive than those carried out in pool. However, certain material characteristics can only be assessed in a hot cell. In the following subsections examples of different examination techniques and results obtained are discussed. The interested reader is referred to Rudling & Patterson (2009).

It is instructive to examine the transitions in creep mechanisms in solid solutions of class-A type such as the results depicted in Fig. 3.10 where we

(a) 10-2

10-3 10-4

^kT 10-5 DEb

note that the stress exponent in the intermediate stress region is 3.5 corre­sponding to viscous glide of dislocations or Weertman microcreep mecha­nism. Since glide and climb occur in sequence, when lower temperatures are approached the climb-controlled creep becomes dominant with five power law thus depicting the fact that the slower climb process controls creep. In

fact this low stress regime is associated with similar characteristics as the climb-controlled creep with distinct subgrain formation and relatively large primary creep region. On the other hand, dislocations may break away from the solute atmospheres at high stresses, thus entering a climb-controlled regime again as noted at higher stresses with higher n value. Following Murty’s work, this breakaway stress can be calculated from the equation74

wm c0

2ekTb3,

where Wm is the binding energy between solute atom and the dislocation, c0 is the solute concentration, and в typically ranges between 2 and 4 depend­ing on the shape of the solute atmosphere. Later, Langdon and co-workers75 showed that this relation is valid for a number of solid solution alloys. Assuming 0.23 eV as a reasonable value for Wm, the critical stress for break­away is estimated to be ~7.5 x 10-4E, which is in agreement with the experi­mental results obtained from various class-A alloys.65 At even higher stresses, another regime may appear involving low temperature climb-controlled creep with a stress exponent value of n + 2 (i. e. 7). This mechanism is asso­ciated with the climb processes involving dominance of dislocation core dif­fusion (Fig. 3.16b). However, this is often masked because the PLB regime starts in the near vicinity.

As lower stresses are approached, one expects to note viscous creep with n = 1 (Fig. 3.16b) either due to N-H or Coble creep mechanisms. Depending on the test temperature, one of the regions such as with n = 3 for viscous glide may completely disappear as noted in Fig. 3.16b. This could get further complicated if an intervening GBS regime with n = 2 appears between vis­cous creep and dislocation creep regimes.

The major sources of crud formation on fuel rod surfaces in BWRs (as in PWRs) are the metallic impurities in the coolant that result from the cor­rosion of reactor coolant system materials. Other sources are leakage of solids, liquids and gases into the system and impurities from the surface of components placed in the core (fuel assemblies, etc.).

Crud types identified on BWR fuel surfaces have generally been:

• tightly-adherent, dense Fe2O3 — generally a thin layer

• loosely-adherent Fe2O3 — various levels of ‘looseness’

• a combination of the two types — tight and loose layers

• tightly-adherent, dense Fe2O3 + Cu or CuO

• tightly-adherent, dense Fe2O3 + ZnO (ZnFe2O4).

The effect of the thin, tightly adherent Fe2 O3 layers on clad surface tem­peratures is minimal. The loosely adherent layers are generally very porous, filled with water and, as a result, have high thermal conductivity and only a small effect on clad surface temperatures.

Zn, injected to reduce activity transport, tends to form the spinel with Fe, as listed above, but is not known to have been the cause of fuel failures. A particularly detrimental crud type is Fe2O3 infiltrated with Cu. This has been associated with CILC cladding failures. The topic has been thoroughly reviewed in a ZIRAT report (Wikmark & Cox, 2001) and by Marlowe et al. (1985). An overview of CILC mechanisms is presented here.

In 1979 and the early 1980s, localized fuel cladding corrosion failures occurred in some plants that had copper alloy (Admiralty brass) condenser tubes and filter-demineralizer condensate clean-up systems. Such plants have higher levels of soluble copper in the water than those with stainless steel or titanium condenser tubes or with deep-bed resin clean-up systems. A few rods per bundle failed in susceptible bundles at burnups >15 GWd/MT. Over

90% of the failed rods contained (U, Gd)O2 fuel (i. e. they were ‘gadolinia rods’). However, most fuel reloads and fuel bundles were not affected, even in susceptible plants. Poolside and hot cell examinations revealed unusual crud scale deposits, with high copper concentrations, rather than the typical fluffy, Fe2O3 crud. Table 4.10 gives a breakdown of crud elemental composi­tion in a CILC-susceptible plant (Baily et al, 1985).

It was reported (Marlowe et al. , 1985) that three factors interacted to cause CILC fuel failures: reactor water chemistry, fuel duty and Zircaloy resistance to nodular corrosion. Marlowe et al. (1985) and Wikmark and Cox (2001) provide details and analysis. An interpretive summary is given here as an illustration of a crud-induced failure process. Failure in the gado — linia rods proceeds by the following steps:

1 Incubation phase (low to moderate power)

(a) extensive nodular corrosion occurs early in first cycle

(b) oxide nodules grow on some fuel rods to produce 90-100% coverage

(c) copper, in reactor water, deposits between oxide nodules

(d) copper deposition continues, crud grows within oxide nodules to form a thick sandwich structure (ZrO2/crad/ZrO2)

2 Failure phase (moderate to high power)

(a) cracks form within sandwich structure producing local, steam-insu­lated regions

(b) insulating effect accelerates cladding corrosion and hydriding

(c) cladding penetrations occur locally by formation of auto-catalytic corrosion pits or by cracking of hydrided Zircaloy in spalled regions.

Gadolinia rods are at low power during the first cycle and never become the highest power rod in a bundle (see Fig. 4.57). For reasons that are still not fully

understood, but perhaps related to the specifics of the boiling phenomena in a BWR core, low heat flux rods, or regions of rods, have been shown to be par­ticularly susceptible to nodular corrosion (Marlowe et al, 1985). Therefore, in gadolinia rods, Step 1b (in the incubation phase) is reached quickly. Figure 4.58 shows that when the nodular coverage reaches about 90%, copper depos­its copiously on the crudded rod surface. The most likely deposition location is between nodules, as it has been shown that microscopic heat flux increases between nodules and, therefore, good conditions for wick boiling are estab­lished (Wikmark & Cox, 2001). At this point, copper also begins to deposit in lateral cracks in the thick nodules. Figure 4.59 is an elemental X-ray map of a typical nodule, clearly showing copper (and zinc) in cracks in the nodule. At this point the ZrO2/CuO layer can be more than 100 pm thick and heat trans­fer through the nodule is inhibited. However, the conductivity of ZrO2/CuO is still quite high (Wikmark & Cox, 2001, Table 12) and gross overheating should not occur. The failure phase accelerates dramatically at step 2a, when new cracks in the oxide ‘sandwich’ form and become steam-filled. This can be facilitated by the deposited copper blocking normal ingress of coolant and egress of steam from the cracks, and by expansion of the CuO2 as the temper­ature increases. Once steam-blanketed regions form, clad temperatures can become very high, as steam has a very low conductivity compared to the crud or oxides present. Step 2c follows, as the steam blanketing and the resulting high temperatures dooms the cladding.

Two-step steam testing (Cheng et al,, 1987) showed that only a small percentage of material in failed fuel bundles is susceptible to nodular corro­sion, thus explaining why most rods do not fail.

Experience of CILC failures drove the BWR industry to develop nodule-resistant Zircaloy microstructures, to refine the 500°C-type steam test as a tool to predict in-reactor corrosion resistance and to tighten-up water chemistry specifications, particularly for copper.

It should be noted that there is no evidence that the presence of copper in the water enhances nodule nucleation. In fact at least one careful study indi­cates that copper either has no effect or improves nodular corrosion resistance (Ito et al, 1994; Shimada et al, 1997). Since nodular corrosion by itself has never been shown to affect fuel performance (assuming oxide spallation does not occur), elimination of either copper in the water or of nodular corrosion is claimed, with considerable justification, to eliminate CILC-type failures. In fact, the industry has been free of classical CILC failures for the past 15 years. Recent fuel failures in the United States may be caused by a crud-induced process, but the characteristics are different from CILC. Both, however, induce fuel rod failures by temperature-driven corrosion processes.