Shielded containments are commonly referred to as hot cells, the word ‘hot’ being used as a synonym for radioactive. Hot cells are used in both the nuclear power and the nuclear medicines industries. They are required to protect individuals from radioactive isotopes by providing a safe contain­ment box in which they can control and manipulate the equipment required.

Hot cells are used to inspect spent nuclear fuel rods and to work with other items, which are high-energy gamma ray emitters.

The methods utilized in a hot cell examination vary with the objec­tive of the program, but usually begin with a visual inspection. Visual inspections are used to characterize the external surfaces of a fuel rod, including the crud and oxide layer, and to identify features of interest for subsequent examinations. High resolution diametral profilometry and EC lift-off measurements are also used to characterize the thickness of the crud and oxide layers relative to axial and azimuthal position, partic­ularly if such measurements were not performed before transporting the rod to the hot cell.

EC measurement with encircling, pancake or pencil coils can detect the axial and tangential location of cracks and other damage that might not be readily discernible in visual inspections; for example, small or incipient cracks due to PCI.

Neutron radiography — If a test reactor is adjacent to a hot cell, neutron radiography can be performed on the fuel rod to non-destructively locate regions with different hydrogen levels in the cladding. Depending on the method used for capturing and displaying the neutrographic image, accu­mulations of hydrogen are indicated by light or dark areas. Neutron radi­ography can also reveal fuel washout in the case of degraded failed fuel. In addition, neutron radiography provides geometrical information of the fuel column for the selection of cutting positions of samples for ceramographic examinations.

Destructive examinations — The fuel and cladding can be examined by means of a number of destructive methods including:

• Fission gas collection and analysis.

• Cladding oxide thickness (ID and OD) measurement.

• Hydride concentration and morphology.

• Second phase particle size and distribution.

• Microstructure characterization.

• Ceramographic characterization of fuel pellets.

• Micro-gamma scanning across the diameter of fuel pellets.

• Electron microprobe analysis of the fuel pellets and the pellet-cladding interface.

The fission gas that was released from the fuel to the free volume within a rod is normally collected and analyzed as the first step in the destructive examinations of sound fuel rods. The collection process involves punctur­ing the cladding and collecting the free gas in one or more containers of known volume, pressure and temperature. This process not only captures the gas from the open volume within a fuel rod, but also provides data for the calculation of the net internal void volume and the total quantity of gas in the open volume. The gas sample is analyzed for composition by tech­niques such as gas chromatography to identify the concentration of Xe, Kr, CO-CO2, CH4 and other constituents. The results are adjusted for decay and combined with calculations or measurements of exposure to establish the fraction of the generated fission gas that was released to the free volume within the fuel rod.

Metallographic mounts of the cladding and fuel may be prepared, pol­ished and photographed. The mounts may be examined for clad oxide layer (external and internal), fuel/clad interactions, fuel restructuring, rim effect and agglomerate behaviour. The mounts may also be etched to reveal grain boundaries for an analysis of the grain size.

Hydrogen/hydride analysis — A hot vacuum extraction process may be used to determine clad hydrogen content such as that employed by the LECO method. (LECO is a corporation which provides instrumentation for elemental determination in organic and inorganic materials.) When such analyses are done, it is important to separate the total hydrogen evolved from the sample into concentrations from the surface layer(s) and from the metal itself (which is of interest). Specifically for samples with thick oxides, water molecules adsorbed at the cracked and porous zirconium oxide may contribute to the recorded hydrogen content and give a metal hydrogen content that is too large. It is therefore crucial that the adsorbed water mol­ecules are removed before the hydrogen content in the sample is measured. This can be done by grinding of the oxide. but in this case care is needed so as not to remove massive hydrides at the outer metal surface of a fuel rod. For components without a surface heat flux there is no concentration of hydrides at the outer metal surface and there is lower risk of removing hydrides from the sample by grinding.

In addition to gas-extraction methods, detailed destructive post-irradia­tion examination (PIE) can be used to characterize the local oxide thick­ness and hydrogen concentration. Backscattered electron imaging (BEI) of polished cladding in the SEM may be used together with image analysis to determine the local hydrogen concentration and radial hydrogen profile through the cladding wall. This is a relatively new application of BEI which allows the relative positions of hydrides and oxides to be located more pre­cisely than the gas-extraction methods.

Alternatively, the hydride distribution and orientation in the cladding material is determined by optical examination after etching the sample to reveal the hydride locations.

Crud analysis — The morphology of the crud may be examined by electron probe microanalysis (EPMA).

Mechanical tests — Mechanical tests of differently shaped samples from

the fuel cladding can be carried out in the hot cell. Some of the tests that can

be done with irradiated cladding are:

• Room and elevated temperature axial tensile test yield strength (YS), ultimate tensile strength (UTS), uniform elongation (UE) and total elongation (TE).

• Cladding room and elevated temperature ring tensile test (YS, UTS, UE and TE).

• Burst testing.

• Cladding thermal creep rate test.

• Hardness testing.

Uniform corrosion proceeds over the entire surface area of the material exposed to the environment leading to a general slow thinning accom­panied by a release of corrosion products. Uniform corrosion is usually relatively easy to measure and predict. In the primary circuit of PWRs, released ions are transported and may be activated when they reach the reactor pressure vessel (RPV), which is a major problem to be overcome.

Therefore, uniform corrosion is minimized combining appropriate materi­als (nickel alloys and stainless steels), surface finish, passivating treatments and alkaline pH.

Flow-accelerated corrosion or flow-assisted corrosion (FAC) is a mecha­nism in which the passive layer dissolves in fast flowing water, without any mechanical erosion. As a consequence, the underlying metal continuously corrodes to recreate the protective oxide. FAC rate decreases when the flow velocity decreases and when the pH increases. FAC stops as soon as oxygen is dissolved in water. FAC affects carbon steel piping of the secondary cir­cuit where water or wet steam circulates.

Pitting is a localized corrosion forming holes at the surface of the metal, induced by the local depassivation of an area, which becomes anodic while a large area becomes cathodic. The acidity inside the pit is sustained by the spatial separation of the cathodic and anodic half-reactions, which creates a potential gradient and the transport of anions into the pit. The presence of surface defects, such as scratches and local changes in chemical compo­sition promote pitting which causes little loss of material but it may lead to deep corrosion in a component. Pitting mainly affects materials such as austenitic stainless steels exhibiting a good resistance to uniform corro­sion thanks to their good passivation. However, the presence of chlorides and oxygen at relatively low temperature (typically 80°C) may weaken the passive layers and enhance pitting via an autocatalytic process: Cl — ions start to concentrate in the pits for charge neutrality and promote the reaction of positive metal ions with water to form a hydroxide corrosion product and H+ ions. The increasing acidity within the pits accelerates the process.

Stress corrosion cracking (SCC) is a progressive failure affecting metals subjected to a tensile stress (residual or applied) while they are exposed to a corrosive environment. SCC occurs in specific and limited conditions in terms of water chemistry, material and loading. SCC usually involves a long incubation period prior to initiation, followed by a slow crack exten­sion stage and transition in a fast crack propagation stage leading to failure. Stress concentrations, cold work and irradiation promote SCC. The mate­rial most susceptible to SCC is Alloy 600 (in both primary and secondary waters). However, stainless steel becomes susceptible to SCC in primary water under specific conditions: polluted environments (oxygen plus chlo­rides), high level of cold work and irradiation. The susceptibility of nickel alloys to SCC strongly decreases when the chromium content of the mate­rial increases, especially above 20%. Basically, SCC results in the oxide ingress, usually at grain boundaries, which locally weaken the material. The oxide penetration is enhanced by the presence of strain and is affected by precipitation. If local stresses are sufficient to fail weakened grain boundar­ies, a crack extension occurs.

Environmentally assisted fatigue occurs under the combined actions of low frequency cyclic loading and oxidation. Therefore, ingredients are very similar to those involved in SCC mechanisms in the sense that a synergy operates locally between oxidation and mechanics. The major difference with SCC is the nature of the loading: cyclic loading strongly promotes strain localization in shear bands. The movement of dislocations (defects allowing the non-reversible deformation of the metal) enhances oxide ingress and failure, especially when shear bands emerge at the surface, breaking the pas­sive layer. Therefore, strain rate is one of the key controlling parameters of the mechanism.

Last, it should be noted that one of the corrosion products is able to embrittle metals. Indeed, as reported before, the cathodic reaction (reduc­tion of water) produces hydrogen which partly enters into the metal and interacts with the microstructure. The entry of hydrogen is limited by the growth of the passive layer; also its transport and interactions with the metal strongly depend on temperature. Hydrogen embrittlement is the process by which a localized accumulation of a sufficient level of hydrogen can eventually lead the metal to fracture under residual or applied stress. Therefore, situations limiting the transport of hydrogen (low temperature, presence of traps) promote such embrittlement. Other mechanisms of introducing hydrogen into metals exist, such as manufacturing (welding) and irradiation.

The constitutive relationships identified in the previous sections are appli­cable for a wide variety of metals and alloys. However the strain rates of deformation might assume values different from model predictions even while the parametric dependencies remain the same. This was discussed in case study (3.7.2) and was attributed to the presence of impurities. On the other hand, there are cases where the strain rates of deformation as well as parametric dependencies can turn out to be different. Such instances are encountered while dealing with two — or multi-phase alloys that exhibit pre­cipitation or dispersion strengthening. Strain compatibility issues as well as differences in deformation rates of individual phases contribute to discrep­ancies in experimental observations and traditional creep model predic­tions. To this end, analytical models have been proposed to understand the creep behavior of multi-phase alloys.103-105 Here we present a case where the second phase is rigid and is added to enhance the overall strength of the alloy. For particle strengthened alloys, stress exponents higher than those predicted by established creep models106 and/or anomalous varia­tion of stress exponent with stress are observed.107 This is rationalized by the introduction of a friction or resisting stress also known as back stress

as demonstrated by Li et al. im As noted in Fig. 3.27a, the stress exponent decreases with increasing stress.

By introducing a friction stress (t0) the creep behavior of this alloy could be described by the following equation:

where у is the strain rate of deformation, т is the applied stress, t0 is the threshold stress and the rest of the terms are as described previously. Following the introduction of the threshold stress, the creep strain rates, Y1/n, are plotted on a linear scale against the applied stress. Here n is chosen as that value where a linear correlation between jlln and т is obtained. Extrapolation of the straight line to the x-axis provides a value for the threshold stress (Fig. 3.27b). The threshold stress is attributed to the pres­ence of the second phase which acts as an obstacle to dislocation motion. Such threshold stress based models have been considered in 60 s in analyzing the creep behavior of precipitation and dispersion strengthened alloys.108

The most simple view of the growth mechanism is that it is due to interstitial <a> loops (with displacement vectors or Burgers vectors pointing in the a-direction) lying on prism planes, and vacancy <c> loops (with displace­ment vectors or Burgers vectors pointing in the c-direction) lying on basal planes (Buckley 1961). Figure 4.67 gives a schematic illustration.

The simple vision is that planar arrays of vacancies cause shrinkage in the direction normal to the plane, and planar arrays of interstitials cause expan­sion. However, all vacancies and interstitials do not end up as loops. As dis­cussed in earlier sections on irradiation creep, they can also be deposited at grain boundaries, solute atoms and network dislocations (that is, dislocations introduced by deformation rather than irradiation) having <a> or <c+a> Burgers vectors (i. e., line dislocations having a net strain in the <a> or <c+a> lattice directions). In the final analysis, irradiation growth is due to aniso­tropic deposition of vacancies and interstitials at these sinks and is strongly influenced by the anisotropic diffusion of self interstitial atoms (SIAs) in the basal plane, or more exactly, in the directions normal to the c-axis.

The mechanisms for irradiation growth parallel those for irradiation creep, with the notable exception of any applied stress.

In general, growth is now believed to occur when there is an anisotropic distri­bution of sinks receiving a net flux of vacancies and this anisotropy is different from that of the distribution of sinks receiving a net flux of the self intersti­tial atoms (SIAs). The dilations associated with the addition of lattice planes accompanying the precipitation of SIAs then do not cancel the contractions associated with the removal of lattice planes associated with the condensation of vacancies. For a complete understanding of irradiation growth one must identify the possible sinks and explain their evolution, identify the source

of the anisotropy and explain the partitioning of the PDs amongst the sinks. (Holt, 1988)

The key unique mechanistic feature of irradiation growth is the initiation and growth of <c> component dislocations (either as irradiation-produced loops or deformation-induced networks). Without them breakaway growth does not occur. It also appears to be essential that diffusion of SIAs be anisotropic and favour the directions in the basal plane. More details are given in Holt et al. (1996), Woo (1988), Holt etal. (2000), Holt (2008) and Christien and Barbu (2009).

Irradiation growth will occur to some extent in all zirconium alloys. To minimize the effects of growth several actions are important:

• minimize residual stresses,

• plan for the effects of cold work,

• minimize hydrogen absorption,

• plan for the effects of temperature,

• understand and control alloy chemistry,

• control texture.

LCR, TDR, and FDR are cable testing techniques that measure imped­ance in cables in order to detect anomalies. There are two basic types of impedance tests — lumped data and distributed measurement — based on the ability of the test to localize its measurement. Lumped data tests like LCR typically identify anomalies in cables with greater accuracy than dis­tributed measurement methods. But once the fault is detected, the distrib­uted measurement methods (TDR and FDR) can determine the distance to the fault.

The LCR test uses an LCR instrument or meter at specific frequencies to make impedance measurements along the cable at specific frequencies to verify the characteristics of the cable conductor, insulating material, and the end device. The results are evaluated to determine if they are as expected for the type of circuit being tested. Imbalances, mismatches, or unexpectedly high or low impedances between the cable leads indicate problems caused by cable degradation and ageing, faulty connections and splices, or physi­cal damage. For example, abnormal capacitance measurements indicate a change in cable dielectric or insulation. In addition to providing information about cables, connectors, and end devices, LCR measurements can identify circuit problems such as moisture or loose connections (Hashemian, 2010; IAEA, 2011).

The most popular and effective cable testing technique today, TDR, is used to locate problems along a cable, in a connector, or at passive devices at the cable end by sending a test signal through the conductors in the cable and measuring its reflection. It works on the same principle as radar. A pulsed or swept DC signal is sent through the cable, and its reflection is measured to identify the location of any impedance discontinuity or change in the cable and the end device (load). It measures the time taken for the signal to travel down the cable to where the impedance change is located, and return. This propagation time for a known distance is then converted, and depending on the type of display used, the information can be presented as a waveform and/or a distance reading (IAEA, 2011).

Any significant change in impedance along the cable will cause a reflec­tion that will appear on the TDR signature as a peak or valley whose ampli­tude depends on the characteristics of the cable impedance. Depending on the impedance of the load, the TDR trace representing the end of the cable may step up or step down. That is, reflected voltage waves occur when the transmitted signal encounters an impedance mismatch or discontinuity (fault) in the cable, connector, or end device. Any such change in impedance along the cable due to a short, open, shunt, or other electrical effect can thus be identified and located using the TDR test. A rise in the reflected wave is indicative of an increase in impedance, and a decrease in the reflected wave is indicative of a decrease in impedance. Thus, the peaks and dips in a TDR plot are used to identify the location of normal and abnormal electri­cal effects throughout the cable (Hashemian, 2010).

The TDR test is typically performed using a pulse generator, which produces a step pulse, and a recorder, oscilloscope, or automated computer-controlled data acquisition system, which captures the reflected wave. The test signal is applied between pairs of lead wires, a cable shield, and a ground plane, and the results are displayed as a plot of the reflected wave versus time or distance (Hashemian, 2010).

Yielding diagnostic information about the cable conductor and any con­nector or connection, the TDR method relies on comparisons with a base­line TDR. Its success therefore typically shows significant improvement if there is a baseline TDR for comparison (IAEA, 2011). In light water reactors, the TDR method is useful for testing instrumentation circuits, motor and transformer windings, pressurizer heater coils, thermocouples, RTDs, motor-operated valve cables, neutron detector cables, and other components that are normally inaccessible, such as in high temperature and high radiation zones. The simplest and perhaps most important appli­cation of TDR is to locate an open or short lead, moisture, or problems such as erratic behavior along a cable or in an end device (e. g. resistance temperature detector).

The TDR test non-destructively identifies and locates cable defects and dis­continuities on an installed cable in-situ, providing trendable measurements. However, end terminations of the cable must be disconnected in order to per­form the test (U. S. NRC, 2001; 2010b). In addition, the size of the wave that the TDR method sends down a tested cable is limited by the bandwidth of the pulse and sampling circuitry. Because it sends only very broad DC pulses, the TDR method can locate only DC open — or short-circuit conditions.

Like TDR, the frequency domain reflectometry (FDR) technique can measure the distance to and severity of a fault in a cable conductor, connec­tors, and end device. However, because the FDR technique uses a selected set of much smaller or narrower bandwidth frequencies, it is also able to locate RF faults in cables, unlike the TDR test. The FDR technique can also help identify degradation in cable insulation material. There are three types of FDR which calculate distance based on the sine wave property they mea­sure — namely, frequency, magnitude, and phase.

In FDR, a stepped (or variable) frequency sine wave generator sends stepped-frequency sine waves down the cable. These waves are reflected back from the cable end as well as from any faults encountered along the cable, and are sensed by either a frequency counter, received signal strength indicator, or another technology for measuring high or intermediate fre­quency voltage magnitudes. Using pulses of discrete frequencies can iden­tify and locate small faults in connectors or cables, making possible a more realistic picture of cable condition than TDR provides.

FDR measures reflection responses in the frequency domain and then converts the data into the time domain using an inverse Fourier transform. Similarly, FDR data can be acquired by using a TDR to measure the reflected wave over the large bandwidth and then using Fourier transform to convert from time to frequency domains. Decreasing the time required for a signal to change from a specified low value to a specified high value (rise time) of a TDR test will increase its accuracy, as will increasing the bandwidth of an FDR test. Similarly, increasing the number of frequency samples in an FDR test increases its maximum range, as does increasing the period between the rise and fall of the pulse in a TDR pulse (Hashemian, 2010).

Like TDR, FDR is a non-destructive technique that can send a swept sig­nal through miles of cable without attenuation as long as the cable under test is shorter than the signal wavelength (IAEA, 2011).

Reverse time domain reflectometry (RTDR) is a technique developed by CHAR Services Inc., a division of Analysis Measurement Services Corporation (AMS). It tests the quality of the shielding around the conduc­tor of a coaxial or triaxial electrical cable. The RTDR method estimates the distance to a fault in a cable by coupling a repetitive DC pulse to the shield and allowing the pulse to travel the length of it. The time delay between the DC pulse and when the signal is received can be measured by simul­taneously monitoring the cable signal path (Hashemian, 2010). These time delays make it possible to identify the point at which the electromagnetic interference (EMI) couples into the cable system. This reveals the location of degraded connectors or cable shields because such interference usually couples at cable connections or terminations that tend to degrade through ageing or damage (IAEA, 2011).

A stress exponent value of 1 suggests the mechanism of creep to be Coble, N-H or H-D controlled. Recent studies have shown that the Spingarn-Nix
(S-N) mechanism with a stress exponent value of 1 could also be a viable creep mechanism.17 This mechanism could be identifiable through micro­structural evaluation following creep deformation revealing the limitation of only considering the creep parameters. In this section, the mechanism of Coble, N-H, H-D and S-N creep will be discussed.

E110 alloy (Zr-1%Nb) has been the universal fuel cladding in the Voda Voda Energo Reactor (VVER) design, which is a pressurized water reactor (Cox et al. , 2004). In the early VVER-440 (MWe) reactors the fuel assem­blies were hexagonal arrays of 126 E110 clad fuel rods, with 10 or 11 stainless steel grid spacers, enclosed in an E125 hexagonal wrapper (which performs the same function as a BWR channel) (Smirnov et al, 1994) see Fig. 4.6a. In the VVER-1000 type reactors each assembly contains 312 fuel rods, a central E125 alloy support tube and 18 stainless steel guide tubes for control and shut-down rods. Except for the first VVER-1000 (Novo-Voronezh-5) such VVER-1000 fuel assemblies do not have ‘wrappers’.

The increasing temperatures and presence of steam will cause the intact cladding to oxidize on the outer diameter (OD) and the burst cladding to oxidize on both the OD and ID (two-sided oxidation) (Strasser et al., 2010b). The oxidation process at the high LOCA temperatures will increase the oxygen and hydrogen content in the cladding, reducing its ductility and resistance to rupture. Two sided oxidation can have significant effects on the post-quench ductility (PQD) of the cladding as a result of high but localized hydrogen pickup in addition to the oxidation (Strasser et al., 2010b). The cladding continues to oxidize until the ECCS becomes effec­tive and a peak-cladding temperature (PCT) is reached. The maximum PCT is regulated to be a maximum of 2200°F by the USNRC and 1200°C internationally. The length of time the system may remain at the PCT is determined by the reactor system and regulated by the equivalent clad­ding reacted (ECR) limit, defined as the total thickness of cladding that would be converted to stoichiometric ZrO2 from all of the oxygen con­tained in the fuel cladding as ZrO2 and oxygen in solid solution in the remaining metal phase.

There are several models that have been proposed to describe the rate con­trolling mechanism in the five power-law creep regime. The general con­sensus is that the five power-law creep regime is diffusion-controlled. This is evident from the equivalence between the activation energy for creep and that for self diffusion. In addition, factors affecting self-diffusion such as phase transformation, superimposed hydrostatic pressure, etc., simi­larly influence creep-rates thereby rendering support to the fact that the creep-rate is proportional to self-diffusivity (DL). Thus, all models that have been proposed to explain the five power-law creep regime are built around the concept of a dislocation climb-controlled creep mechanism.

The earliest model to describe creep by dislocation climb was proposed by Weertman5663 who considered the creep processes to be a result of the glide and climb of dislocations, with climb being the rate controlling process. The glide motion of dislocations is impeded by long range stresses due to dislocation interactions and the stresses are relieved by dislocation climb and subsequent annihilation. The rate of dislocation climb is determined by the concentration gradient existing between the equilibrium vacancy concentration and the concentration in the region surrounding the climb­ing dislocation. Creep strain, however, arises mainly through the glide of dislocations. In the glide-climb model (Fig. 3.12) a dislocation produced by the Frank-Read source glides a distance L’ until a barrier of height ‘h’ is

encountered which it has to climb so that another dislocation can be gener­ated by the source.

A rather simple way of deriving the relation between the strain rate and the applied stress and temperature is given by referring to Fig. 3.12: thus the total strain is given by,

Ay = strain during glide-climb event = Ayg + Ay. « Ayg = p b L

h

t = time of glide-climb event = tg + tc ~tc = ,vc = climb velocity

vc

■ = Ay=pbbL

t h / vc

where vc <x ACv e-Em/kT, Em = activation energy for vacancy migration Y = 9bvc, where ACV = C+ — C — = C0 e lkT = C0 2sinh ^ ^ j

sothate = apbLv = apbLC0e~EmlkT 2sinh| |

H h c H h v { kT)

At low stresses, sinh (oV/kT) ~ oV/kT

e = A1pb(L/h)C0e~EmlkT (V/kT)

є = AipbLDL ~ A2pcrLDl. [3.31]

h kT h

Assuming that the dislocation density (p) varies as stress is raised to the power 2 (a2), we find that

£ = Aa’DL. [3.32]

This is known as ‘natural creep law’ and Weertman showed that L/h in Equation [3.31] varied as o15 so that

e = ADLo45.

This equation with n = 4.5 and D = DL agreed closely with the experimental results on pure aluminum. Subsequently, this has been generalized with n close to 5 and is referred to as the five power-law creep.

At high stresses, Equation [3.30] predicts an exponential stress dependence:

sinh | 0V1« exp I 0V j so that є = A1pbLDLeoV/kT УkT) УkT) h

and є = A о2 DLe°VlkT ~ A’DLe°V/kT,

as is commonly noted in the PLB regime. Both the power-law and exponen­tial stress regimes can be combined into a single equation as proposed by Garofalo6

є = ADl (sinhBo)n, [3.35]

which describes both the power-law creep regime at low stresses and expo­nential stress dependence at high stresses.

Another model that considered the non-conservative motion of disloca­tions was proposed by Barrett and Nix.64 This model came to be known as the ‘jogged screw dislocation’ model. The rate controlling mechanism is the motion of screw dislocations containing edge jogs. The edge jogs impede the motion of the screw dislocations and the non-conservative motion of the edge jogs becomes the rate controlling mechanism. This model is sim­ilar to the Weertman model in the sense that the rate of climb of the edge jogs is dependent on the concentration gradient established by the climb­ing jogs. In the original model Barrett and Nix assumed the jogs to be of atomic height, but recently Viswanathan et al.65 have shown that these jogs could be several times larger than atomic dimensions. The modified jogged screw model proposed by Viswanathan et al. has been used to satisfactorily explain the creep behavior of titanium aluminides65 and some titanium and zirconium alloys.66,67

Ivanov and Yanushkevich68 were the first to identify subgrain boundaries as important rate controlling features. The subgrain walls were suggested as obstacles to the motion of dislocations emitted within the subgrain. Subsequent plastic deformation could occur only when the dislocations
were annihilated at the subgrain boundaries. This annihilation process is climb controlled.

There are a few other network and recovery based models that appear attractive. These models consider the dislocation networks (Frank net­works) present inside the subgrains to explain the hardening and recovery during creep.

We proceed here with descriptions of corrosion phenomena, which often limit the lifetime of core components.

4.5.1 Types of corrosion and comparison between PWRs and BWRs

Corrosion of zirconium alloys used in the core of nuclear power plants (and the accompanying absorption of hydrogen in the zirconium metal matrix)

is of prime interest when considering performance of the core components and therefore the performance of the entire reactor. For instance, for PWRs a practical corrosion limit exists (about 100 pm oxide thickness which is associated with a critical amount of hydrogen absorption) that has driven a material change away from Zircaloy-4.

The technical literature and many conferences are full of papers deal­ing with corrosion issues. Of particular interest is the series ‘Zirconium in the Nuclear Industry: International Symposium’, ASTM STP, American Society of Testing and Materials, which provides most relevant details. Also reviews of the topic are given in ZIRAT reports: Adamson et al. (2002), Cox et al. (2004), Adamson etal. (2007); and ‘Waterside corrosion of zirco­nium alloys in nuclear power plants’, International Atomic Energy Agency (IAEA)-TECDOC-996, January 1998. The most recent open literature review of mechanisms is by Cox (2005).

Corrosion of zirconium alloys is an electrochemically-driven process affected by the microstructure and microchemistry of the alloy surface, the nature of the oxide layer that forms, the temperature at the metal/oxide interface, the chemistry and thermohydraulics of the corrodent water, the effects of irradiation and the effects of time. Table 4.7 gives information on the various types of commercial power reactor systems currently being used throughout the world. In comparing BWRs with PWRs, with corrosion mechanisms in mind, the main features are:

• BWR coolant boils; PWR coolant does not. This has an important effect at the oxide/water interface.

• PWR coolant contains a high concentration of hydrogen; BWR coolant does not. Complementarily, BWR coolant contains a high concentration of oxygen, PWR coolant does not. This has an important effect on cor­rosion processes.

• PWR components generally operate at higher temperatures than BWR components. Corrosion processes are temperature dependent.

• Both reactor types employ chemical additions to the coolant which may affect corrosion and buildup of deposits on fuel rods.

It also should be noted that BWR zirconium alloys continue to be primarily Zircaloy-2 or slight variants of Zircaloy-2. PWR zirconium alloys no longer tend to be Zircaloy-4, for reasons of insufficient corrosion resistance (and hydriding resistance) at high burnup, but have moved toward zirconium alloys with Nb additions.

The type of oxides which form during corrosion in reactor water can be classified into several categories. The two most basic are uniform and nodu­lar corrosion. The ‘uniform’ category has an extension — ‘patch’ or acceler­ated uniform. The fourth category is ‘shadow corrosion’, which can look like thick uniform corrosion but has some characteristics of nodular corrosion. The fifth category is crud-related corrosion, which is a temperature driven process induced by poor heat transfer in crud-impregnated corrosion layers. These categories will be discussed later, but are introduced here. Table 4.8 (Garzarolli in Adamson et al, 2002) gives a useful summary of characteris­tics of various corrosion types.

Uniform corrosion occurs in both PWRs and BWRs. The oxide itself is uniform in thickness and consists of several different layers. For either in or out of reactor, the initial shape of the corrosion-versus-time curve is as shown in Fig. 4.41 in the pre-transition region. The first transition point occurs at around 2 pm oxide thickness in PWRs. The shape of the post-tran­sition curve in PWRs depends on several variables: initial SPP size, irradia­tion, amount of cold work, specific alloy, water chemistry, temperature, local thermohydraulics and hydride concentration. For Zircaloy-4 the corrosion

Parameter Western type PWR VVER (440/1000) MW CANDUb BWR RBMK0 1. Coolant Pressurized H20 Pressurized H20 Pressurized D20 Boiling H20 Boiling H20 2. Fuel materials (Pressure Zry-4, ZIRLOe, DUPLEX, Zr-alloy E110 Zry-4 (Zr-2.5Nb) Zr-2, Zry-4, Zr-alloy E110, tube materials) M5, Inconel, SSe Inconel, SS (Zr-2.5Nb) 3. Average power rating, kW/l 80-125 83-108 9-19 40-57 5 4. Fast neutron flux, Average, n/cm2.s 6-9 x1013 5-7 x1013 1.5-2 x1012 4-7 x1013 1-2×1013 5. Temperatures, °С Average coolant inlet 279-294 267-290 249-257 272-278 270 Average coolant outlet 313-329 298-320 293-305 280-300 284 Max cladding OD 320-350 335-352 310 285-305 290 Steam mass content, % 7-14 14 6. System pressure, bar 155-158 125-165 96 70 67 7. Coolant flow, m/s 3-6f 3.5-6 3-5 2-5f 3.7 8. Coolant chemistry9 Oxygen, ppb <0.05 <0.1 200-400 <20 Hydrogen (D2), 2-4 (3-10) 0.05-0.30 — PPm 25-50 30-60 cc/kg Boron (as boric acid), ppm 0-2200 0-1400 Li (as LiOH), ppm 0.5-3.5 0.05-0.6 1 — — К (as KOH), ppm — 5-20 — — NH3, ppm 6-30 NaOH, ppm 0.03-0.35

aVodaVoda Energo Reactor (VVER). bCanadian Deuterium Uranium (CANDU).

°Reaktor Bolshoi Mozhnosti Kanalov (RBMK). dZirconium Low Oxidation (ZIRLO). eStainless Steel.

‘Variation from lower to upper part of the core and from plant to plant.

9Zn in ppb quantities may be added for BWRs and PWRs; Pt and Rh in ppb quantities may be added for BWRs. Source: A. N.T. International (2011).

thickness at high burnup may reach or exceed 100 pm, while some of the newer alloys have less than half of that. As noted in Table 4.8, only uniform corrosion occurs in PWRs under non-boiling, hydrogenated conditions.

The study of uniform corrosion in the laboratory at temperatures rele­vant to reactor operation suffers from three problems: (1) irradiation effects are absent, (2) corrosion rates are very low at 280-360°C (553-633K) and (3) very long times (hundreds of days) are required to differentiate between material variables. The most widely used test appears to be 360°C (633K) pressurized water for very long time periods. An example is given in Fig. 4.42 where a series of Zircaloy-type alloys with different Sn contents are compared. Garde et al. (1994), show that in-reactor results after high burnup give the same ranking as the autoclave laboratory tests. Also, adding Li to the water is thought to improve comparisons for PWRs (Sabol et al., 1994 ).

Both Zircaloy-2 and -4 form a protective uniform oxide in typical BWRs and PWRs. In BWRs the various types of oxidation kinetic are shown in Fig. 4.43.

Uniform corrosion continues at a very low rate out to high burnups. During reactor exposure the microstructure of the Zircaloys used in BWRs is continually evolving due to irradiation damage and SPP dissolution. In the range 30-50 MWd/kgU (6-10 x 1021 n/cm2, E > 1 MeV), changes in the microstructure induce an acceleration of uniform corrosion, as described earlier. First, patches of white oxide appear in the otherwise black or grey uniform background, as illustrated in Fig. 4.44. These patches remain very

Type of corrosion Comment Observed Irradiation effect in oxygenated coolant (BWR) Irradiation effect in hydrogenated coolant PWR Uniform Normal mode Out-of-pile, in-BWR and in-PWR 5-10x increased from beginning, low flux dependency 2-4x increased after 1st corrosion rate transition, low flux dependency Nodular Local breakdown of oxide protectiveness In-BWR and out-of-pile >500°C, Zry with large SPP Increases almost linearly with fast flux, little temp, dependency Not observed Shadow corrosion Probably driven by potential differences Only under irradiation and oxidative coolant conditions (BWR) Increases probably almost linearly with fast flux Not observed Crevice corrosion Change of environment in small gaps Out-of-pile, in-BWR and in-PWR Observed Observed Increased corrosion with fine SPP Reduction of protectivity Out-of-pile, in-BWR and in-PWR Increases almost linear with fast flux Increases almost linearly with fast flux, little temp, dependency Increased corrosion at high fluences Reduction of protectivity In-PWR ? Increases with increasing fluence above a critical threshold Increased corrosion at high hydride concentrations Lower corrosion resistance of hydride Out-of-pile, in-BWR and in-PWR Observed Flux dependent but less temperature dependent Source: A. N.T. International (2011) and Garzarolli in ZIRAT7 STR ‘Corrosion in Zirconium Alloys’, Adamson eta!. (2002).

4.42
Corrosion weight gain as a function of laboratory autoclave test exposure time for several Zircaloy-4 tube variants. (Source: Reprinted, with permission, from Garde et al. (1994), copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA 19428.)

4.43
Characteristics of different types of corrosion observed in BWRs. Oxide layer thickness in arbitrary units versus fuel assembly burnup.

4.44 Patch oxide formation on Zircaloy-2 after an exposure of 8.5 x 1021 n/cm2, E > 1 MeV. The oxide is thinnest at uniform black oxide (5) and patch oxide (4) and is thickest at coalesced patches (3). (Source: Reprinted, with permission, from Huang et al. (1996), copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA 19428.)

thin, about the same as the black uniform corrosion film. At some point the patches cover 100% of the surface and oxide thickening occurs at an accel­erated rate, labelled ‘late increased corrosion’ in Fig. 4.43.

Importantly, at the same or somewhat higher fluences, a marked increase in hydrogen pickup fraction occurs. This hydriding is potentially a more seri­ous issue than the corrosion increase and is discussed later.

Figure 4.45 gives optical micrographs and schematics of various types of corrosion. Schematics A and C show normal and increased uniform oxide, and the others illustrate nodular corrosion. Zr-Nb alloys and small-SPP Zircaloys (average SPP size less than about 0.1 pm) in general do not form nodules. In large-SPP Zircaloys, nodules initiate early in life and grow at a decreasing rate with fluence. In some cases, for aggressive water chemistries and susceptible material, nodules can coalesce to cover the entire surface. Nodular oxide thickness does not generally cause performance problems; however in severe cases spalled oxide can be a source of ‘grit’ in control drive mechanisms. Serious fuel failure problems can be induced by a combi­nation of heavy nodular corrosion and copper-zinc-laden crud, resulting in the crud-induced localized corrosion (CILC) phenomenon discussed later.

Susceptibility of Zircaloys to in-reactor nodular corrosion can be identi­fied by laboratory high temperature steam tests. The most effective testing procedures are variations of the ‘two-step test’ described by Cheng et al. (1987) where Zircaloys are exposed to steam at 410°C/1500 psi (683K/102 bars) for 8 h followed by 51071500 psi (783K/102 bars) for 16 h. In such tests, Zr-Nb alloys do not exhibit nodular corrosion.

The fifth type of corrosion, indicated in Fig. 4.43, is so-called shadow cor­rosion, described in more detail later. Shadow corrosion is induced on all zirconium alloys when they are in close proximity to many non-zirconium

Zircaloy

ZrO.

Normal

uniform

Nodular

Increased

uniform

4.45 Corrosion morphology for Zircaloy in BWRs (Adamson et al., 2007).

alloys such as stainless steel or Inconel. The oxide thickness is unusually large and often appears to be particularly dense and uncracked. For exam­ple shadow corrosion oxide induced by a stainless steel control blade bun­dle is shown in Fig. 4.46 . Shadow corrosion has ‘always’ been present in BWRs, but not in PWRs primarily related to the high PWR hydrogen con­centration which reduces or eliminates galvanic potentials between dissim­ilar alloy components. In BWRs shadow corrosion caused no performance issues until recently when at one reactor fuel failures were induced by unusually severe ‘enhanced spacer shadow corrosion’ (Zwicky et al., 2000). More recently, shadow corrosion has been alleged to be involved in BWR channel bow problems (Mahmood et al. , 2010). Both issues are addressed later in this chapter.