Several hundred different cable types and sizes are used throughout a typ­ical light water plant. Plant conditions determine which type of cable is used; for example, cables for control rod drive mechanisms must withstand higher temperatures and have additional shielding capacity (IAEA, 2011). I&C cables are by far the most common cable type (Hashemian, 2010). Instrumentation cable, which includes thermocouple (T/C) extension wires, is a low-voltage (< 1 kV), low-ampacity cable used to transmit digital or ana­log measurement signals from transducers such as resistance temperature detectors (RTDs) and pressure transmitters. Control cable, also low-voltage and low-ampacity, is used in the circuits of control (rather than monitoring)

Wire insulation

Braided shield

Cable jacket

Foil shield

Conductor

6.1 Cable components.

components such as control switches, valve operators, relays, and contactors (Hashemian, 2010).

A complete cabling system (see Fig. 6.1) may include any or all of the following components: conductor, insulation, shield, jacket, terminations, penetrations, splices, connectors, and/or end devices (sensor, transmitter, detector, motor, etc.) (AMS Corp., 2010). However, the main components of an I&C or low-voltage power cable are conductors, electrical insulation or dielectric, shielding, and the outer jacket.

Power cables and I&C cables both operate by providing a conductive route for an electric circuit by using metallic conductors — typically cop­per or aluminum that are insulated with a polymer and have different configurations such as coaxial, triaxial, twisted pair, or multi-conductor arrangements of single-strand or bundled wires (AMS Corp., 2011 ; U. S. NRC, 2001). The cable insulation and jacket are made of different poly­mers, including polyethylene (PE), cross-linked polyethylene (XLPE), polyvinyl chloride (PVC), ethylene propylene diene-monomer (EPDM) rubber, ethylene propylene rubber (EPR), Hyplon, Lipalon, and others (AMS Corp., 2010 ; 2011). More than three-quarters of cable insulation and jacket used in nuclear plants is constructed from such polymers (U. S. NRC, 2010a). Another type of cable, fiber-optic, is used to transmit signals based on optical fiber technology. Though its outer jacket is similar to cop­per and aluminum cable, fiber-optic cables have unique ageing, degrada­tion, and failure characteristics (U. S. NRC, 2010a). As such, they are not covered here.

Connectors are also part of the conductor in a cable circuit. A multitude of connectors, terminations, terminals, splices, etc., join the conductor to other cables or electronic equipment. The failure of these components
may appear as a problem with the conductor in test data and may be the result of corrosion, loose terminations, and other faults (AMS Corp., 2010).

Cable manufacturers qualify cables for a specific service life (e. g. 40 years for nuclear power plant cables) and specific voltage class at a given max­imum ambient temperature (U. S. NRC, 2010a). Service life is affected by everything from voltage and temperature rating of the cable and the mate­rial and thickness of its insulation and conductor jacket, to the conductor size and construction (e. g. solid or stranded), the type of metal and coat­ings used in the conductor, the cable configuration (e. g. single or multiple), and the presence of ground conductors, shields, braids, or binding and filler material (U. S. NRC, 2001).

Because I&C cables are used at low current, their typical operating tem­peratures are between about 40°C and 65°C (IAEA, 2011). In contrast, power cables can operate at 80-90°C because of continuous current flow, which generates ohmic self-heating, and the higher voltages and currents used to power medium — and high-voltage equipment such as pump motors (IAEA, 2011 ).

Because of their typical length, cables can experience multiple operating environments as they travel through different areas of the plant, including harsh temperature, radiation, humidity, and moisture conditions, which may include submersion in water (IAEA, 2011; U. S. NRC, 2001).

Since 1997, cracking incidences of pressurizer heaters were encountered in French PWRs.20,21 The heater sheath (outer diameter of 22 mm, 2 mm thick) is made of 316 L stainless steel. The lower part of the heater is attached to electrical connectors, out of the pressurizer, while the upper part, introduced into the pressurizer, consists of a coaxial heating element coiled round a copper mandrel. After the final assembly of the heater, the sheath is cold swaged to reduce the gaps between the heating element and the sheath and therefore, to improve the thermal exchanges. Under operat­ing conditions, the heaters are exposed to hydrogenated and non-polluted primary environment at 345°C. Nevertheless, the temperature at the outer surface of the sheath, in nominal condition, could reach 360°C. Nine cases of SCC were found after 12 destructive examinations out of the 1200 fail­ures observed on heaters. When SCC leads to primary leakage (boron traces on connectors), the heater has to be replaced no later than the next outage. Elemental analyses (energy-dispersive X-ray spectroscopy) did not reveal any trace of pollutant at the surface of the retired heaters. Therefore, it was concluded that SCC occurred in the nominal hydrogenated primary environment. No chromium depletion was present at the grain boundaries of the stainless steel. Such depletion can result in the precipitation of chro­mium carbides at grain boundaries and promotes IGSCC in an oxidizing environment.

A surface annealing heat treatment was developed (induction heating) to counteract the initiation of SCC on the original cold-worked outer layer exposed to the primary water. The goal of the heat treatment is to anneal the surface of the material, decreasing strain-hardening and residual stresses without any damage of the electrical properties of the heater element. As a result, the Vickers micro hardness decreased from 320 HV1 (higher than the threshold necessary to initiate SCC)22 down to 200 HV1 (below the threshold necessary to initiate SCC) at the surface of heat treated heaters, and resid­ual stresses were removed as shown by corrosion tests in MgCl2 medium.

1.3 Conclusion

Despite the original stringent selection of the materials used to manufac­ture the components, uniform and localized corrosion occurs in PWR envi­ronments. Remedies can be of different types: adjusting the water chemistry, reducing superficial strains and stresses, replacing materials or changing microstructures. In particular, experience in the field demonstrated that an increase in chromium content is an efficient strategy: to date nickel alloys containing 30% chromium, used to replace 16% chromium nickel alloys, have exhibited very good resistance to localized corrosion, such as SCC. The history of the degradations shows that for a given type of material, the ten­dency to corrode can largely depend on the manufacturing conditions.

The creep behavior of unirradiated material is taken as a benchmark to pos­tulate its performance in the reactor. Though these out-of-pile tests may not be representative of their in-reactor behavior, they have been successfully used to grade various materials during alloy development programs and to

gain basic understanding about the material behavior. It has been recog­nized that hoop strain in a clad tube is a vital parameter in the breach of fuel clad, and evaluation of their creep and burst behaviors is very important to assess the integrity of the tube.117 Steady-state creep-rates at relatively high (>5 x 10-4E) followed the same behavior as described earlier exhibiting power-law creep behavior with exponential dependence at higher stresses and were identified as due to dislocation glide-climb creep mechanisms. At low stresses, viscous creep with the characteristic n = 1 was indeed reported as expected. Figure 3.29 summarizes the various sets of results in terms of Dorn parameters for Zr-alloys.118 Bernstein119 observed that both Zircaloy-2 and pure Zr exhibit a stress exponent value of unity at low stresses which increases to 4.6 and 6, respectively, at higher stresses. The data produced by MacEwen et at.120 also showed that the n increases with stress (for compa­rable dE). On the contrary, data from Ardell and Sherby121 for a-Zr with comparable purity and in the comparable low stress range, but at slightly higher temperature, showed a stress exponent value of 7.5 and the n value reduced at higher stresses indicating operation of series mechanisms (see

0-6 10-5 10-4 10-3

ст /G

3.29 Steady-state creep of a-Zirconium.123

Section 3.4). The reason for these differences is not clear. While the results of Prasad et al.122 indicated a stress exponent value close to 1 for pure zirco­nium at low stress levels (1-3 MPa) revealing the operation of Coble creep, the mechanism of creep at low stresses (0.2-14 MPa) at intermediate tem­peratures is ascribed by Ruano et al., to grain boundary sliding than to dif­fusion mechanism.123

Major sources are many, including:

• ZIRAT Annual Reports and ZIRAT Special Topic Reports, A. N.T. International, Molnlycke, Sweden (www. antinternational. com).

* PR, partially recrystallized; RX, recrystallized; SR, stress relieved. ** Fitting factor (calculated by Garzarolli et al., 2011a).

*** Hydrogen pickup fraction (calculated by Garzarolli et al., 2011a). Source: Garzarolli et al. (2011a).

Notes: Proposed advantage relative to current Zry-2; C — corrosion; SC — shadow corrosion; HPUF — hydrogen pickup fraction; HPUF HB — at high burnup; G — irradiation growth; PQ — beta quenched.

Source: A. N.T International (2011) and Garzarolli et al. (2011a).

• The series of Zirconium in the Nuclear Industry, International Symposiums, ASTM International, West Conshohocken, PA, USA, held every 2-3 years.

• Zirconium Production and Technology: The Kroll Medal Papers 1985­2010, editor, R. B. Adamson, ASTM International RPS2, ASTM I, West Conshohocken, PA, USA, 2010.

• Proceedings of the LWR Fuel Performance Meeting/Top Fuel/WRFPM, held annually in the United States, Europe or Asia.

• References given in Section 4.10

• The next chapter of this book — ‘Performance and Inspection of Zirconium Alloy Components in Nuclear Power Light Water Reactors’ P. Rudling, ANT International, Molnlycke, Sweden and R. B. Adamson, Zircology Plus, Fremont, CA, USA.

4.7 Acknowledgements

The authors sincerely thank our colleagues in the expert network staff of

ANT International: Brian Cox, Friedrich Garzarolli, Charles Patterson and

Alfred Strasser. Their discussions, their expertise, their comments and their

contributions to the ZIRAT programme reports have greatly contributed

to this chapter.

The servicing and maintenance of the miles of I&C, low — and medium-voltage cables in each light water plant has historically been reactive in nature. Such reactive efforts have successfully resolved connector problems, corrected signal-to-noise ratios, and improved grounding and shielding. However, they have done little to identify the condition, age, or remaining useful life of cables, especially the insulation material (AMS Corp., 2010). Not enough research has been completed to identify a useful, practical method, procedure or technique for accurately evaluating the ageing condition of plant wiring or correlating the condition of cables to measurable electrical, mechanical, or chemical properties (AMS Corp., 2010). In 2010 , the U. S. NRC (NRC DG1240) stated that ‘research and experience have shown that no single, nonintrusive, currently available condition monitoring method can be used alone to predict the survivability of electric cables under acci­dent conditions’ (U. S. NRC, 2010b).

Because of the safety-related importance of I&C cables functioning effec­tively on an ongoing basis, efforts to use prognostic techniques to predict residual life in cables continue.

Such techniques attempt to establish relationships between condition indicators and ageing stressors (IAEA, 2011). To predict future perfor­mance, a trendable indicator and a well-defined end point are essential. From them, a trend curve can be used to estimate the time remaining before the end point is reached (U. S. NRC, 2001). Used with appropriate mate­rial ageing models and knowledge of environmental conditions, such trend data can be used to estimate residual cable lifetimes, but only when suffi­cient data has been generated to validate predictive ageing models (IAEA, 2011). Currently, both the NRC and DOE are sponsoring research at AMS, national laboratories such as Sandia National Laboratories (SNL), Oak Ridge National Laboratory (ORNL), Idaho National Laboratory (INL), and elsewhere to address cable aging and cable qualification issues.

In recent years researchers have developed analytical ageing models based on experimental data from cable samples that have been subjected to accelerated ageing. For example, the power law extrapolation model extrapolates radiation ageing data obtained under isothermal conditions at several dose rates. Similarly, the superposition of time-dependent data model combines data from both thermal and radiation ageing to account for both dose rate effects and the synergistic relationship between radiation and thermal ageing. The superposition of end-point dose data model also uses a superposition approach to radiation and thermal ageing data, but can be used in materials where a single dominant degradation mechanism is lacking (AMS Corp., 2011).

There have been recent efforts toward integrated cable residual life analysis systems that combine existing methods to provide cable testing, ageing assessment and cable management as part of a plant-wide cable age­ing assessment program (AMS Corp., 2010) (see Table 6.2). For example,

Analysis and Measurement Services (AMS) Corp. of the United States is developing methods to ‘calibrate’ results from classical testing methods so they can be categorized, evaluated consistently, and if necessary improved. Correlations between measurable parameters and the health and condition of the cable using classical ageing tests such as the elongation at break (EAB) test would be identified. The classical tests would then be integrated with promising new cable testing technologies, such as the wireless AgeAlert™ micro-sensors (AMS Corp., 2010) (see Table 6.3). Testing methods are then categorized according to their capability to show a particular fault, faults or developing cable conditions that indicate degraded performance (see Fig. 6.2). These tests are performed using laboratory and plant-aged cables of the types found in nuclear power plants.

The correlations between the changes measured by the various methods and condition or age of the cable will form the foundation of a database that will be the core of the integrated cable testing and analysis system (AMS Corp., 2010). This database would contain the information to provide default configuration settings for the various devices that could be tested, optimized data acquisition parameters for the equipment under test, control of data acquisition hardware, and the ability to analyze and store the results of the testing. The program for the AMS integrated cable testing practice would incorporate eleven different modules (see Fig. 6.3): user interface; test lead compensation; test data acquisition; data storage; data qualifica­tion; data review; statistical analysis; historical data trending; similar equip­ment data comparison; report generation; default equipment setting (AMS Corp., 2010 ).

The result will be a user-friendly and technically feasible solution for examining low — and medium-voltage plant cables and wiring to determine their ageing condition and residual life (AMS Corp., 2010).

6.2 Testing and analysis techniques in an integrated cable condition monitoring system.

The S-N model32 is based on the fact that dislocation climb, when assisted by grain boundaries, can occur at activation energies smaller than those of lattice diffusion. Since grain boundary diffusion is much faster than lattice diffusion, climb rates are increased in the proximity of a grain boundary. The S-N model thus relates to the ideas of diffusional creep and dislocation climb at grain boundaries. This model, also known as the slip-band model, provides a physical mechanism to explain the observation of activation energies equal to the grain boundary self diffusion and a stress exponent equal to 1. According to this model, creep occurs by shearing along slip — bands blocked by grain boundaries. The creep strain at the boundary is in turn accommodated by diffusional flow. A schematic of the slip-band model is shown in Fig. 3.5. Under the application of a shear stress, the slip-band/

grain interfaces slide, generating compressive (C) and tensile (T) tractions at the grain boundary. In order to relieve these tractions, atoms flow from regions under compression to regions experiencing tensile stresses. This atomic flow is accompanied by grain boundary sliding that causes the shear of a continuous slip-band. A mathematical analysis using Fick’s law for dif­fusion of vacancies yields the following expression for the strain rate due to the absorption of dislocations into the boundary:33

where X is the slip-band width, 2 is the slip-band length, П is the atomic volume, 8 is the grain boundary thickness and the rest of the terms are as defined before. The slip-band length, /, can be considered equal to the mean linear intercept grain size, d. Recently Gollapudi et a/.11 have studied the feasibility of the slip-band model as a viable creep mechanism in a titanium based alloy.

Corrosion resistance in zirconium alloys is intimately related to the presence of second phase particles (SPPs) formed in the zirconium matrix by delib­erate additions of alloying elements. The precipitates are usually incoherent crystalline intermetallic compounds, meaning that their physical structure is unrelated to the Zr matrix in which they are imbedded. In as-fabricated Zircaloy-4 the most common SPP is Zr(Fe, Cr)2, while in Zircaloy-2 they are Zr(Fe, Cr)2 and Zr2(Fe, Ni). For the ZrNb type alloys the most common is pNb (which is not an intermetallic) and for the ZrSnNbFe alloy types are Zr(Nb, Fe)2 and PNb. Table 4.6 gives a more complete description, also indi­cating some neutron irradiation effects.

At normal LWR temperatures (270-370°C, 543-643K) the SPPs change under irradiation in a combination of two ways — amorphization and dissolution.

Amorphization means that the original SPP crystalline structure is con­verted to an amorphous structure. Amorphization is a complex process, described in some detail by Griffiths et al. (1987); Yang (1989);Motta (1997); Bajaj et al. (2002); and Taylor et al. (1999). It occurs when an intermetallic compound accumulates enough irradiation-induced defects to cause it to thermodynamically favour an amorphous rather than a crystalline structure. The rate of amorphization depends on the relative rates of damage creation and damage annealing in the SPP; therefore important parameters are neu­tron flux, irradiation temperature and SPP chemistry. A critical tempera­ture exists above which the annealing processes are fast enough to prevent the damage accumulation of defects needed for transformation. For typical reactor irradiations amorphization of both Zr(FeCr)2 and Zr2(Fe, Ni) occurs readily at temperatures near 100°C (373K) (although Fe is not related from the SPPs into the Zr matrix, as discussed later). At typical (LWR) tem­peratures (300°C, 573K) and neutron flux, Zr(Fe, Cr)2 becomes amorphous but Zr2(Fe, Ni) does not. Above about 330°C (603k) neither SPP becomes amorphous.

The amorphization process begins at the outside surface of the SPP and works its way inward with increasing fluence. This is illustrated in Fig. 4.11 (Etoh & Shimada, 1993) where the SPP on the left has an amorphous rim (dark area) and the one on the right, at higher fluence, is fully amorphous. There appears to be an incubation period prior to amorphization initiation, with the incubation fluence decreasing with temperature in the range 270­330 ° C (543-603K).

Reference Material As- fabricated Moderate burnup <330°C >330°C High burnup <330°C >330°C Adamson, 2000; Zircaloy-2 or Zircaloy-4 Zr( Fe, Cr)2 PA X A X Griffiths et al., 1996 PD PD D D Garzarolli, eta!., 1996(a); Zr2(Fe, Si) X X X X Adamson & Rudling, 2002 PD PD D D Garzarolli etal, 2002(a) Zr3Fe X X X X s PD Griffiths et ai, 1996; Zircaloy-2 Zr2(Fe, Ni) X X X X Etoh &Shimada, 1993 PD PD D D Zr-1 Nb Shishov et a!., 1996 E110 (3N b X Bossis et al., 2002 M5 (3N b X Gilbon et al., 2000 IE PN b Doriot etal., 2004/2005 M5 (RX) (3Nb Zr(Fe, Nb)2 X, PD, IE X, PD Shishov et al., 2007 E110(RX)(0.1 Fe) |3Nb Zr(Fe, Nb)2 X, PD, IE X, D E110(RX) (0.01 Fe) PNb X, PD ,IE Zr-2.5Nb Urbanic & Griffiths, 2000 (27% CW) o(in p phase) X, PD IE PNb (in a phase) Averin et al., 2000; E125 (RX, PRX) PNb X, PD IE PNb X, PD IE PNb Shishov et al., 1996 ZrSnNbFe Nystrand & Bergquist, 1997; ZIRconium Low Oxidation (ZIRLO) ZrFeNb PNb X, PD, PNb X Comstock et al., 1996 (Stress Relieved (SR)) Averin et al., 2000; E635 (RX, PRX) Zr(Nb, Fe)2 — PA, PD X, PD IE PNb A, D X, D IE PNb Shishov et al., 1996 Nikulina et al., 1996 (Zr, Nb)2Fe X, PD, IEZr3-4Fe X X X, IEZr3-4Fe Shishov et al., 2002 E635,RX(0.15Fe) Zr(Nb, Fe)2 PNb X, PD, PNb D, lEpNb Shishov et al., 2007 E635(RX) Zr( Fe, Nb)2 D, IE

Notes: All are crystalline (X) as-fabricated.

A — amorphous; D — dissolved; P — partially…; X — crystalline; S — stable; IE — irradiation-enhanced precipitation Source: A. N.T. International (2011).

On/m2	7x1025n/m2	1.2x1026n/m2
Amorphous state ‘%
Crystalline state
Amorphous state — шгтк. ‘M
Crystalline state
200 nm

4.10 The fluence dependence of the amorphous transformation of Zr(Fe, Cr)2 precipitate in recrystallized annealed (RXA) Zircaloy-2, neutron irradiated at 288°C (561K). Diffraction patterns indicate stages of the transformation (Etoh & Shimada, 1993).

Amorphization rate increases as temperature decreases, as neutron flux increases and as SPP size decreases. Literature evaluation therefore needs to consider reactor and material conditions of specific interest.

The fluence required to produce complete amorphization depends on neutron flux, temperature and SPP size, but for typical Zr(Fe, Cr)2 SPPs of initial size near 0.1 pm and the entire SPP is amorphous by the end of bundle life burnups <50 MWd/KgU (1 x 1022 n/cm2, E > 1 MeV). Interestingly, under well controlled conditions of flux and temperature, the amorphization rate of Zr(Fe, Cr)2 in Zircaloys can be used to estimate the neutron fluence (Motta & Lemaignan, 1992; Taylor et al, 1999; Bajaj et al, 2002).

For the Zr-Nb type alloys neither the pNb nor Zr(Nb, Fe)2 SPPs become amorphous for irradiation temperature >330°C (603K). However, at 60°C (333K) Zr(Nb, Fe)2 does become amorphous at high fluences.

SPP amorphization in itself does not appear to affect material behaviour; however, dissolution of both amorphous and crystalline SPPs does influence corrosion, growth and mechanical properties, to be discussed later. At typi­cal LWR operating temperatures, SPP dissolution occurs relentlessly until the SPP essentially disappears.

As SPPs dissolve, the zirconium matrix becomes enriched (well beyond the normal solubility limit) in the dissolving element. For instance in Zircaloy-2, Fe leaves both Zr(Fe, Cr)2 and Zr2(Fe, Ni) SPPs as schemat­ically illustrated in Fig. 4.12 (Mahmood et al., 2000). This process is given in more detail by Takagawa et al. (2004) and in Fig. 4.13 . Here it is seen

4.11 Evolution of a Zr-Fe-Cr particle under BWR irradiation. Upper figures: TEM micrographs; middle diagrams: schematic illustration of amorphization; lower figures: schematic illustrations of the chemical compositions. (Source: Reprinted, with permission, from Takagawa et al. (2004), copyright AsTm International, 100 Barr Harbor Drive, West Conshohocken, PA 19428.)

that Fe rapidly diffuses from the amorphous rim into the matrix, while Cr diffusion is sluggish. At high fluence (~1 x 1022 n/m2, E > 1 MeV) complete amorphization and Fe-depletion has occurred, while the Cr level is still high. Only at very high fluence (~1.5 x 1022 n/m2, E > 1 MeV) is the Cr dispersed into the matrix, and the SPP essentially disappears.

The rate of dissolution depends on the SPP size (higher rate for smaller sizes), and the extent of dissolution depends on size and fluence. It has been demonstrated in a BWR that small (<.04 pm) SPPs can completely dissolve at low to moderate burnups, (Huang et al, 1996). Also in a PWR, but at temper­ature near 290°C, SPPs with an average size of 0.2 pm were >80% dissolved at moderate burnup (1 x 1026 n/m2, E > 1 MeV) (Garzarolli et al., 2002).

Modelling of the dissolution process gives insight into the alloying con­centration of the matrix (Mahmood et al, 1997). Figure 4.14 illustrates the model for release of solute into the matrix for various size SPPs. For the small SPPs (1R, 2R, 3R) all the Fe is released by moderate burnup.

For the channel material with very large (0.6 pm) SPPs only a small amount of Fe would be released even at high burnups. However, modern materials have SPPs with an average size < 0.3 pm.

In another study, experimental measurement of Fe released from Zr (Fe, Nb) 2 SPP in an E635 alloy containing 0.35% Fe during irradiation at 330-350°C is shown in Fig. 4.15 (Shishov et al, 2002). (In Fig. 4.15, fluence has been converted from E > 0.1 MeV to E > 1.0 MeV by dividing by 4.) Here it is seen that the Fe has diffused from the SPP to the alpha Zr matrix such that all of the Fe is in the matrix by moderate burnup. Extending to high burnup (2 x 1026 n/m2) in this case may only increase the probability of re-precipitation of Fe in the matrix. It should be noted that the ‘normal’ solubility of Fe in unirradiated Zr is <0.02 wt%.

Table 4.6 outlines changes in SPPs to be expected at moderate (50 MWd/ kgU) to high (100 MWd/kgU) burnup for various alloys now in use. To illus­trate interpretation of the table, consider the as-fabricated crystalline (X), Zr(Fe, Cr) 2 SPP for Zircaloy-2 or -4.

For moderate burnup at <330°C, the SPP would become partially amor­phous (PA) and partially dissolved (PD), depending on its initial size. At >330°C it would remain crystalline (X) and become PD, the extent of which would depend on initial size. At high burnup for <330°C it would very likely become totally amorphous (A) and could completely dissolve (D), depend­ing on its initial size. At >330°C it would remain crystalline (X), although it would become strongly fragmented as it eventually totally dissolved (D).

For the ZrNb and ZrSnNb alloys the most common SPPs are pNb and the (Laves phase) (L) Zr(Nb, Fe)2. Also observed is the T-phase (Zr, Nb)2Fe. Details of the SPPs present in the Nb-Fe corner of the phase diagram are presented in Fig. 4.16 (Shishov et al., 2007). Also, a simplified diagram is pre­sented in Fig. 4.17 (Garzarolli in Nikulina et al, 2006). Such phase diagrams

4.12

Modelling predictions for solute release to the matrix as a function of fluence for Zircaloy-2 (SPP size: 1R = 0.026 pm; 2R = 0.042 pm;

3R = 0.056 pm and Zircaloy-4 channel= 0.6 pm) irradiated near 300°C:

(a) Fe, (b) Cr and (c) Ni (Mahmood et al., 1997). Copyright 1997 by the American Nuclear Society, La Grange Park, Illinois.

are only approximate, and may vary because the heat treatments used may not achieve equilibrium conditions. Figure 4.16 points out that there is a region, not reported in Fig. 4.17 , where only the Laves phase exists. This tends to be for 1%Nb and 0.2-0.4 Fe; for example in the ‘normal’ E635 alloy Zr-1.2Sn-1.0Nb-0.4Fe.

None of the SPPs become amorphous at normal reactor temperatures, but at high burnup at 60°C (333K) the Laves phase does become at least partially amorphous. However, all of the SPPs undergo irradiation-induced dissolution. The PNb SPP loses Nb to the matrix, but the excess Nb then re-precipitates as a very fine PNb. The Laves phase Zr(Fe, Nb)2 transforms to a fine PNb SPP, with essentially all Fe ending up in the matrix (see Fig. 4.15). Behaviour of the T-phase (Zr, Nb)2Fe is more complicated, with Fe diffusing

to the matrix, and Nb and Sn concentrating in the outer shell of the SPP. The core remains a T-phase. Details can be obtained in the following references: Shishov et al. (2002, 2007) and Shishov (2011) and Doriot et al. (2004).

Most countries that generate nuclear power are in the process of develop­ing criteria, designs and sites for the permanent disposal of spent nuclear fuel, but they have yet to become licensed realities (Adamson et al, 2010). The most significant fuel related criteria for dry storage are compared in Table 5.1. Meanwhile the pools at the nuclear plant sites are filling up with spent fuel and the utilities are transferring the spent fuel from the pools to dry cask storage sites that are mostly located at the plant sites. Exceptions are the central, large intermediate pool facilities that serve all the plants in Sweden (CLAB facility) and all the plants in Finland (KPA-STORE). The lack of a licensed permanent fuel repository in any country has placed total reliance on intermediate storage. As a result the dry storage technology has become a major activity and business component of today’s back-end fuel strategies.

The key differences between dry storage and in-reactor performance of fuel are (Adamson et al., 2010): [4]

Dry storage criteria USA Germany Hungary S. Korea (CANDU only) Spain (follows USNRC) Switzerland Pellet clad °С, storage, drying, other 400° 570° 370° 370°, 100 hr 410° 410° None but maintain clad integrity 400° 570° None 570° Temperature cycling Max. no. of cycles and Max. ДТ (°С) per cycle 10/65° None None None 10/65° 10/65° BU, GWD/MT None 65 assembly avg. 49-50 None None None Clad hoop stress, MPa 90 (ifTclad > 400 °С) 120 @370° None None 90 Depends on fuel supplier*; typical values are 90 and 120 Clad strain limit None 1% circumferential None None 1% for HB 1% during storage Creep rupture limits None None None None Limited by °T and stress limits Cladding condition limits Oxide thickness for stress calculation Oxide thickness for stress calculation None None Limited by °T and stress limits None

Any assembly with defected rods

Yes in special container (ZIRAT11/IZNA6, Sect. 11.3.2)

0.95 normal 0.97 accidents

Yes, with actinides & fission products

{Continued)

• He gas pre-pressurization during fabrication

• fi ssion gases

• He from transmutation of B in burnable absorbers

• alpha decay of the Pu isotopes in the fuel during storage

and their pressure is further raised by the fuel decay heat. All of these pres­sure sources, except the pre-pressurization level, are burnup dependent.

The pressure outside the cladding in the cask is only slightly above atmo­spheric. Creep deformation of the cladding will occur at a relatively constant rate early in life, when the internal gas pressure, the cladding stresses and the cladding temperature are at their highest. As the decay heat decreases with time, the gas pressure and the cladding temperature both decrease. In addition, the internal free volume of the fuel rod increases as the cladding creeps outward, decreasing the gas pressure and cladding stresses further. All three of these factors eventually reduce the creep rate to a negligible value (Adamson et al, 2010).

Creep-rupture is the most likely cladding failure mode during dry storage and there is general consensus on this mechanism among the industry and the regulators (Adamson et al., 2010). The parameters that determine the potential for creep rupture are the cladding stress level, the cladding tem­perature and the rate of decay heat decrease, all parameters that are burnup dependent.

The cladding temperature, currently one of the primary USNRC licens­ing criteria, is determined by the decay heat generated in the fuel, the heat transfer capability of the cask and the surface temperature of the cask in its storage environment (Adamson et al., 2010). The decay heat is generated primarily by absorption of the alpha decay either directly or indirectly from the plutonium (Pu) isotopes. Increasing burnup will increase the level of Pu isotopes formed by transformation of the 238U, and increase the cladding temperature in dry storage conditions. In comparison, MOX fuel will have significantly higher temperatures under the same conditions.

Several other potential failure mechanisms were considered, but elim­inated as highly unlikely (Rashid, 2006). They are summarized below (Adamson et al, 2010).

Stress corrosion cracking (SCC) is not a credible failure mechanism in dry storage because:

• There is insufficient elemental iodine present to cause SCC.

• At the stress and strain rates in dry storage, initiation of intergranular cracking is nearly impossible; the 180-200 MPa stresses needed for SCC are well above those for high burnup fuel rods.

• Hydrides, including radial hydrides, will not affect iodine induced SCC.

• The occurrence of all the conditions that cause DHC is highly unlikely, but cannot be ruled out. The initial conclusions are based on the follow­ing evaluation:

• Analyses indicated that at a hoop stress of 250 MPa (well above dry storage conditions) in a cladding wall thickness reduced by 100 pm oxide with an 83 pm crack size, the stress intensity factor is below that needed to initiate the DHC process,

• Hydride re-orientation that might assist crack propagation is intended to be minimized or prevented by current regulations and industry practices, but cannot be ruled out.

• In addition, propagation of a crack assisted by radial hydrides may not occur for many of the hydride morphologies.

Also under accident conditions during storage or subsequent transporta­tion, the fuel must remain subcritical and should be recoverable by nor­mal methods (Adamson et al, 2010). The hypothetical accident conditions that these criteria have to meet, as defined by the USNRC, are specified in 10CFR71.73 (NRC, Rules and Regulations, Title 10 Code of Federal Regulation, Chapter 71). Of all the accident scenarios the most limiting scenario is a free drop of the cask for a distance of 9 m (30 ft) onto a flat, unyielding horizontal surface, striking the surface in a position that would cause the maximum fuel damage.

Radial hydrides in zirconium alloy cladding are undesirable because they reduce the critical stress intensity required to propagate a radial crack through the wall of the cladding during handling or transportation (Adamson et al, 2010). The objectives of the dry storage regulations are to limit the conditions that could result in hydride re-orientation.

A certain fraction of the hydrogen (H) picked up during the oxidation reaction is soluble in the zirconium matrix and the remainder forms zirco­nium hydrides (Adamson et al, 2010). The solubility of the H is a function of temperature, alloy composition and microstructure. Solubility is also a function of irradiation history, heating or cooling rates during service. The orientation of the hydrides formed during normal reactor operation are generally circumferential near the cooler cladding OD and remain so dur­ing wet storage of the spent fuel.

The hydrides can reorient in the radial direction if they are precipi­tated from solid solution by cooling the alloy from a higher temperature under a tensile or hoop stress (Adamson et al, 2010). The hydrides will align themselves in the direction perpendicular to the tensile stress. This can occur during reactor operation although it is generally unlikely. It could occur during dry storage if the internally pressurized cladding is at a high temperature, holds sufficient hydrogen in solution and is then cooled while under the hoop stress. The hydrides in solution will precipitate in the radial orientation (provided the hoop stresses are large enough), while the hydrides that did not dissolve will remain in their original circumferen­tial orientation. This is most likely to occur during rapid cool-down from high temperatures after cask drying or evacuation procedures rather than during storage when the rate of temperature and pressure reduction that control the stress levels are extremely slow.

In summary, the factors that affect hydride re-orientation in irradiated cladding are (Adamson et al, 2010): [5]

• Microstructure features such as grain size and shape, amount of CW, and perhaps others.

• Texture.

The radial hydrides can be present in a wide variety of sizes and distribu­tions as well as fractions of the total hydrides present and each type of struc­ture can have a different effect on mechanical properties. This emphasizes the importance of characterizing the structures when they are related to the mechanical properties measured.

A material experiences transitions in mechanisms when the applied stress or the test temperature is varied.

3.5.1 The Bird-Mukherjee-Dorn equation4

As discussed earlier, for the same temperature and stress combinations, a material can creep via different mechanisms if the grain size is different. In fact, as Equation [3.25] would suggest, a material creeps with higher strain rates for smaller grain sizes. For relatively smaller grain sizes, creep could occur by diffusion of vacancies through the grain boundaries. But larger grain sized materials, under the same stress and temperature conditions, could creep by dislocation-based processes or by lattice diffusion processes.

In order to illustrate the effect of stress and temperature on transitions in mechanisms it is necessary to suitably modify the creep equation. Sherby analyzed steady-state creep-rate results using strain rate compensated by diffusivity versus stress normalized by temperature-dependent modulus of elasticity16

[3.42]

so that different materials can be compared with each other. While this equation seems to work well, it would be more appropriate to use dimen­sionless strain rate as well, and Dorn and co-workers4 proposed a dimen­sionless equation that can appropriately describe the effect of changes in stress, temperature and microstructure on mechanisms of creep. This equa­tion known as the Bird-Mukherjee-Dorn (BMD) equation is given by

[3.43]

As shown in Fig. 3.16a, changes in stress and temperature for a given con­stant microstructure of the material can reveal changes in the stress expo­nent value.60 At low normalized stress values, the deformation mechanism appears to proceed with a stress exponent value of 1. At intermediate stress values a stress exponent value of 2 corresponding to GBS is obtained. At the highest normalized stress values, the mechanism of deformation oper­ates with a stress exponent value of 5 corresponding to power-law creep. The diffusivity value utilized for constructing the plot corresponds to the lattice diffusion activation energy of titanium, and thus data at different temperatures follow different curves in the GBS and viscous creep regimes where the appropriate activation energy is that for grain boundary diffu­sion. On the contrary, if one chooses to use the activation energy for grain boundary diffusion, the data at high stresses will lead to different lines for different temperatures. The BMD plot thus allows an easy understanding of the transitions in creep mechanisms following changes in stress and temper­ature. Such an analysis was found to be very useful in delineating various creep mechanisms in Zr-based alloys as depicted in Fig. 3.16b.73 Moreover, such plots made for different materials would show the material behaviors at equivalent loading conditions.4

Enhanced corrosion of zirconium alloys may occur when the corroding surface is close to, or in contact with, certain other metallic components. The shape of the component is often reproduced in the shape of an area of enhanced corrosion, suggestive of a shadow cast by the component on the zirconium alloy surface. The term ‘shadow corrosion’ is therefore often used to describe the phenomenon. Observations of shadow corrosion on water reactor components have been noted for many years. In 1974, Johnson et al. (1974), reported enhanced corrosion in Zircaloy coupons located near, but not touching, small pieces of platinum in the advanced test reactor (ATR). Also, Trowse et al. (1977) reported enhanced fuel rod corrosion beneath steel spacers in steam generating heavy water reactors (SGHWRs).

Most commonly observed are the control blade shadows on BWR chan­nel surfaces adjacent to the control blade handles, such as shown in Fig. 4.53. In this case the stainless steel control blade handle is imaged as a black shadow on a light background, but the reverse is sometimes also observed. It was shown (Chen & Adamson, 1994) that the handle image is faithfully reproduced on the channel surface, but is larger than the actual handle, shown schematically in Fig. 4.54. Hot cell examination of a similar channel shows that oxide thickness within the shadow area can be much higher than outside the shadow (Adamson et al., 2000 ).

Channel

Control blade handle Control blade

CHANNEL *91695

4.53

Control blade shadow on a BWR channel. Upper diagram shows the relative arrangement of blade handle and shadow (Chen & Adamson, 1994). Copyright 1994 by the American Nuclear Society, La Grange Park, Illinois.

Zry-2 channel, 43 MWd/kgU

Oxide

20 ..m

Away from shadow area 20 .m, 300 ppm H2

(b)

Oxide

20 .m

In shadow area 12 .m, 150 ppm H2

4.55 Oxide thickness (a) away from and (b) within a control blade handle shadow (Adamson et al., 2000).

Figure 4.55 illustrates a 6x difference in oxide thickness within and with­out of the shadow. It is noteworthy, however, that in this case, hydrogen content in the shadow is 150 ppm, while outside the shadow it is 300 ppm. Since temperature gradients are small in a channel wall, redistribution of hydrogen by thermal-gradient driven diffusion is also small, indicating in this case a much reduced hydrogen pickup rate in the shadow. More recent data indicate variability in pickup fraction (HPUF), as (Mahmood et al., 2010) reported normal HPUF. It is also observed that control blade shadows preferentially form during the first cycle of operation and that the thickness tends to saturate with burnup.

Shadows on fuelled or non-fuelled rods have been observed under Zircaloy spacers with Inconel springs or under all-Inconel spacers (Fig. 4.56). Again,

this illustrates the trend for saturation of the oxide thickness with fluence or burnup. In most cases, shadows have not caused fuel performance problems. The upper curve in Fig. 4.50, is an exception for a specific cladding condition, called Enhanced Spacer Shadow Corrosion (Zwicky et al, 2000).

A number of experiments have been conducted to elucidate the details of the shadow corrosion mechanism. Combined with the commercial reactor observations, these experiments reveal:

1 A variety of metals are observed to cause shadows on Zircaloy. These are:

(a) Stainless steel (many)

(b) Pt (Shimada et al, 2002; Johnson et al, 1974)

(c) Hf (Shimada et al, 2002)

(d) Inconel X750, X718 (many)

(e) Inconel 600 (Adamson et al, 2000)

(f) Welded regions on Zircaloy (Chen & Adamson, 1994; Shimada et al, 2002).

2 Nitronic 32 was observed not to cause shadows (Andersson et al, 2002).

3 Resistance to shadow formation depends upon inherent corrosion resis­tance (Andersson et al, 2002; Garzarolli et al, 2002; Shimada et al, 2002).

4 The distance between components is critical. Oxide thickness is a func­tion of distance. There is a maximum distance above which there is no effect (a few mm) but there is no minimum distance, including touching (Chen & Adamson, 1994; Lysell et al, 2001; Andersson et al, 2002).

5 In general, the shadow forms at low fluence or exposure and thickness of the shadow saturates with fluence (Fukuya et al, 1994; Hoffmann & Manzel, 1999; Andersson, 2000). In unusual cases, perhaps due to spe­cial microstructure and water chemistry, accelerated shadow corrosion begins at high fluence or burnup (Zwicky et al, 2000; Wikmark & Cox, 2001 ; Andersson et al., 2002 ).

6 Shadow formation requires a nuclear reactor environment and it has not been possible to reproduce it in laboratory autoclaves (Andersson, 2000; Garzarolli et al;, 2001). However, use of ultraviolet light in the labora­tory has been shown to increase electro-chemical potentials between common components, believed to be related to shadow formation (Kim et al., 2010 ).

7 No reports of shadow formation have been made for PWRs or high — hydrogen cases. BWR hydrogen water chemistry conditions, however, do allow shadows (Lefebvre & Lemaignan, 1997; Adamson et al, 2000).

8 Thick oxide shadows do not necessarily result in proportionally high hydrogen pickup and can result in unusually low hydrogen pickup (Adamson et al, 2000) or normal pickup (Mahmood et al, 2010).

9 Pre-oxidation autoclaving of Zircaloy does not prevent shadows, but applying a zirconia layer to Inconel does (Andersson et al, 2002).

10 Shadow corrosion has been observed when the two metals are not in contact physically and are nominally electrically insulated from each other. However, in making such observations it has been assumed that the radiation field has no effect on the conductivity of the insulating medium.

11 Shadow formation has been reported (Chatelain et al, 2000; Andersson et al. , 2002) in a reactor position outside and downstream of the MIT test reactor core where the neutron and gamma fluxes are reported to be near zero. On the other hand, no shadows were reported (Lysell et al., 2001) in a reactor position outside and upstream of the R2 test reactor core where the neutron flux and gamma power (flux) were also near zero. However, there does appear to be some uncertainty in the actual gamma intensities in the MIT experiment, so there may need to be a re-evaluation of the out-of-core results.

12 Shadow formation can be prevented if electrical connection between the two components is prevented (Lysell et al., 2005 ).

The early thoughts on the mechanism of shadow corrosion centred on it being a form of galvanic corrosion. As such, the mechanism required a path for electron transfer between the cathodic shadower (the material that causes the shadow) and the anodic component (the Zircaloy or zirconium alloy component on which the enhanced corrosion occurs) and a conduc­tive path through the water separating the two parts. But since the shadow phenomenon could not be reproduced in the laboratory, it was clear that some sort of radiation effect was also required. Problems with the galvanic hypothesis included lack of evidence, in some cases, of any electrical con­nection between the shadower and component (although for commercial reactor components an obscure path can always be suspected) and concern that the zirconium oxide, which is always present on component surfaces, was not conductive enough to allow the postulated conductive paths to operate.

Another hypothesis arose when Chen and Adamson (1994) noted that the range in water of beta particles from Mn-56 and Zr-97 (originating in the shadower material) could explain the shape and size of observed shadows if a beta-damage mode could be found. Lemaignan (1992) proposed that extra radiolysis caused by the imposed local beta flux could result in accelerated corrosion. However, additional calculations by Andersson et al. (2002) and Shimada et al. (2002) indicated that the extra beta flux from the shadower does not make a significant change to the overall beta flux in the reactor, so this hypothesis has been discounted. Also, as noted above, it has been shown that the alloy Nitronic 32 does not cause shadows, even though the flux of beta particles from activated Mn-56 from that alloy is much higher than for the known shadowers Inconel and stainless steel.

The latest view of the mechanism is that it is indeed a form of irradiation — assisted galvanic corrosion. Points which support this hypothesis include:

1 It is known that there is a corrosion potential difference between, for instance, stainless steel or Inconel and Zircaloy in non-hydrogenated water (BWR type) (Table 4.9). Also, this potential difference is enhanced in-reactor (Lysell et al., 2001) and by ultraviolet light outside the reactor (Kim et al., 2010 ).

2 The observed relationship between component separation distance and shadow oxide thickness is roughly as expected (Adamson et al., 2000 ; Lysell et al., 2001; Andersson et al., 2002).

3 A true stainless steel/Zircaloy galvanic couple in-reactor produced thick oxide and low hydrogen pickup in Zircaloy, similar to that observed for a control blade handle/Zircaloy shadow (Adamson et al, 2000; Lysell et al, 2005).

4 A radiation enhancement of electrical conductivity of oxides has been reported. Electrical conductivity of oxide films on Zircaloy markedly

increased during electron irradiation (Howla et al., 1999), gamma irra­diation (Kang et al, 1994) and in-reactor (Shannon, 1962). Also, the con­ductivities of various ceramics, including Al2O3 were reported to increase dramatically under proton or x-ray irradiation (Hobbs et al, 1994). So perhaps a way to allow closing of the galvanic circuit is indicated. Note that Al2 O3 was used as insulation in the MIT experiment (Andersson et al, 2002) and ZrO2 as insulation in others (Chen & Adamson, 1994; Shimada et al;, 2002). Some distinction must be made between surface conduction and bulk conduction in thick ceramics, but little is known.

5 Shadows were not formed when the components were confirmed to be electrically insulated (Lysell et al, 2005).

Points which bring doubt to the irradiation-assisted galvanic mechanism hypothesis include:

6 It is not certain that an electrically-conducting path truly exists between the shadower and component.

7 Conventional galvanic reactions are inhibited when the cathode is small and the anode is large, as is often the case with observed shadows.

8 The MIT experiments (Andersson et al, 2002) appear to produce shad­ows in a region of very low radiation.

It is also quite possible that several different mechanisms contribute to shadow corrosion depending on specific conditions.

It is obvious that the mechanism of shadow corrosion is not precisely defined, although the number of observations which help to provide a work­ing hypothesis are many. It is clear that enhanced corrosion can and will occur in cases where Zircaloy and a variety of other metals or alloys are in close contact. Potential problems include loss of strength and integrity due to wall thinning and oxide spallation. In most cases, the enhanced corro­sion does not cause serious operational problems. The oxide thicknesses are moderate, hydrogen absorption is not increased by straightforward shadows and spalling usually does not occur. The severe problem of enhanced spacer shadow corrosion (ESSC)) can apparently be mollified by control of both Zircaloy microstructure and by reactor water chemistry. No harmful effects have been reported or been related to the use of Pt in noble water chem­istry treatments in BWRs, but further studies are underway. BWR channel bowing has been attributed to manifestations of shadow corrosion, and has produced major in-reactor problems to be discussed later in this chapter, but further study is required to understand the details (Mahmood et al, 2007; Blavius et al, 2008; Munch et al, 2008). There does not appear to be an immediate remedy for shadow corrosion — for instance an oxide prefilm on Zircaloy does not prevent shadows. Coatings on the shadower (such as Inconel springs) would appear to be promising, but may not be practicable.