hold down spring failure analysis

17th International Conference on Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors August 9-13, 2015, Ottawa, Ontario, Canada

1

HOLD DOWN SPRING FAILURE ANALYSIS

Jacqueline N. Stevens1, Keith J. Leonard2, Maxim N. Gussev2, Gabriel Ilevbare3, J. Lawrence Nelson4

1AREVA Inc., 3315 Old Forest Rd., Lynchburg, Virginia, 24551, USA 2Oak Ridge National Laboratory, One Bethel Valley Rd. PO Box 2008, MS-6138, Oak Ridge,

Tennessee, 37831, USA 3Electric Power Research Institute, 3420 Hillview Ave., Palo Alto, California, 94304, USA

4JLN Consulting, 56 Bradford Farm Rd., Mills River, North Carolina, 28759, USA

ABSTRACT

Damaged cruciform-style hold down leaf springs were discovered in B&W units in 2007 and 2009. While high growth of host fuel assemblies has contributed to high stresses on the hold down spring assembly, a particular sensitivity in the affected manufacturing lots was suspected. The hold down leaf springs were manufactured from alloy 718 sheet material in accordance with accepted industry heat treatments. The springs undergo multiple manufacturing processes, including grinding, laser-cutting, machining, and forming. During operation, they are subjected to relatively high stresses.

Poolside post-irradiation examination of hold down springs irradiated for two 24-month cycles in a commercial reactor identified numerous instances of severe cracking within each hold down spring assembly. Two of the damaged irradiated springs have been evaluated at Oak Ridge National Laboratory. Detailed microscopy analyses have been combined with an evaluation of mechanical properties and residual strains to perform a failure analysis of the irradiated springs. Comparison of the microstructure, mechanical properties, and microhardness to archive samples allowed for the identification of material differences that may have led to the spring damage.

Archive samples were found to have differing levels of delta phase precipitation, resulting from allowable variation within the precipitation heat treatment. It appears that substantial delta phase precipitation on grain boundaries decreased the samples susceptibility to crack initiation. Potential mechanisms are discussed. It is concluded that the hold down spring failures are attributed to a mechanism combining high levels of stress and a microstructure susceptible to crack initiation under those stress conditions.

Keywords: Alloy 718, Failure Analysis, Delta Phase

1. BACKGROUND

The fuel assembly hold down spring (HDS) design for Babcock & Wilcox 177 units consists of cruciform style leaf springs manufactured from alloy 718 sheet material. Figure 1 shows an image of an individual HDS leaf that has undergone a number of different manufacturing steps prior to final assembly, including grinding, laser-cutting, machining, and roll-forming. The composition of the alloy 718 sheet material is in accordance with Table 1. The final leaf spring is precipitation heat treated according to [1] in a vacuum furnace prior to assembly.

During its operational life cycle, this HDS design encounters sustained tensile stress levels beyond the yield stress of the material.

In 2007 and 2009, damaged HDS were found at several US Pressurized Water Reactor (PWR) sites. Despite the high stress levels experienced by all HDS of this design type, the damaged leaf springs were confined to several manufacturing batches. With the support of Electric Power Research Institute, Exelon Corporation, and the US Department of Energy, two HDS packs were harvested from a PWR spent fuel pool at the end of 2012 and shipped to Oak Ridge National Laboratory (ORNL) for hot cell examination.


2

These harvested HDS experienced a dose of approximately 0.5 dpa, accumulated over two 2-year reactor cycles, in typical PWR water chemistry. Archive samples were also provided to ORNL for comparison to the irradiated samples. These archive samples were representative of batches of HDS that either failed during operation, or completed their operational life with no damage.

2. METHODOLOGY

Leaf springs were examined using an optical microscope (OM), a scanning electron microscope (SEM) including application of electron backscatter diffraction (EBSD), and transmission electron microscope (TEM) techniques. Mechanical properties and microhardness were also measured.

Samples were examined both unmounted and mounted in a standard epoxy material. Samples were examined through optical microscopy in the etched condition, using a solution of 5 g CuCl2, 100 mL HCl, 100 mL ethanol by swab method.

Samples for SEM were either examined in the as-received condition, or were polished for use in EBSD imaging. Polishing was performed through initial surface grinding using SiC and then diamond suspension of decreasing grit size. Following mechanical polishing, the samples were either electropolished in a Struers unit using standard A2 electrolyte at 30V DC for ~ 15 seconds or etched with a proprietary mixture. Samples were examined using either a Hitachi S-4700 cold field-emission gun equipped instrument with energy dispersive spectrometer (EDS) or a JEOL 6500F field-emission SEM with EDS and EBSD detectors.

Samples for TEM were prepared through focused ion beam (FIB) milling using either a Hitachi NB5000 or FEI Quanta 3D 200i instruments. FIB milling of TEM specimens allowed for the precise selection of the location for specimen extraction from the bulk material. Samples were examined on a FEI CM200 TEM, a Schottky 200 keV field emission instrument with scanning-TEM (STEM) capabilities equipped with an EDS and Gatan image filter for electron energy loss spectroscopy (EELS).

Due to some surface oxidation of the irradiated samples, after initial examinations, samples were subjected to chemical oxide removal to enable evaluation of the actual alloy 718 surface. The following steps were repeated as necessary and determined to not modify the alloy surface. The samples were first soaked for five minutes in a boiling solution of 30 g/L potassium permanganate and 100 g/L sodium hydroxide. This was followed by a five-minute soak in a boiling solution of 30g/L ammonium oxalate. The samples were finally rinsed ultrasonically in distilled water.

3-point bending tests were conducted as well, allowing for in-situ observation of the sample surface during straining, followed by SEM/EBSD examination to evaluate localized strain distributions. An MTS “Insight 2-52” screw-driven machine with a force capacity of 2 kN was used in conjunction with a specialized three-point bend test assembly for sub-sized specimens. The span distance was 3.77 mm with the top rod having a diameter of 1.23 mm. Bend tests were performed at room temperature with an estimated strain rate of ~ 10-3 s-1. The design of the specialized load frame allowed for optical light microscopy observations of the surface during straining. A Keyence VHK-1000 long-focal optic microscope running at 15 frames per second was used for real-time observations.

Miniature sheet tensile samples of the type SS-J3 were removed using electrical discharge machining from the arm of the Archive A and Archive B materials. Tensile sample orientations were sectioned in two directions normal to each other to examine any effect of texture on the materials mechanical properties. Microhardness data were collected on either the electropolished archive B material (no damage during operation) or on the etched archive A samples (failed during operation). The parameters used for Vickers microhardness data were experimented with prior to settling on a test load of 500 g with a 10 to 15 second dwell time.


3

Tensile specimens were tested on an MTS “Insight 2-52” screw-driven machine with a force capacity of 2 kN. Tensile experiments were conducted at room temperature (RT) at a strain rate of 0.001 s-1. For each data point, at least three identical specimens were tested.

Tensile experiments were conducted using optical non-contact measurements and digital image correlation algorithms. Optical strain measurements are based on comparing images of an object (tensile specimen) before and after deformation. An Allied Vision Technology GX3500 digital camera was used to obtain images during tensile experiments; the resolution was 10–15 m per pixel. To calculate strain fields and true stress-strain curves, VIC-2D commercial software and ORNL-developed software were used. For all tested specimens, true stress–true strain curves were calculated for the middle third part of the gauge.

In addition to the irradiated samples, archive samples were evaluated using similar techniques. The focus was primarily on two archive samples that were representative of batches of HDS, which either failed during operation (archive A), or completed their operational life with no damage (archive B).

3. EXAMINATION RESULTS

Thorough visual examinations were completed on all twelve leaf springs from the two harvested HDS packs, before and after pack disassembly. Damage was extensive on the samples with most leaf springs showing either well-developed crack networks or complete fracture of a portion of the spring. Higher magnification of the cracked locations showed evidence that cracks likely initiated at multiple locations before propagation formed into crack networks. Examples of the damage are shown in Figure 2.

Portions of the damaged leaf springs were sectioned for further evaluation with OM and SEM/EBSD techniques. The examination focused on determining the cause of crack initiation rather than propagation as the operational stresses are sufficiently high to readily propagate cracks through a stress corrosion cracking (SCC) mechanism once they have initiated.

Figure 3 shows typical fractographic results confirming that the cracks followed intergranular paths, characteristic of an SCC mechanism [2]. The images are from a HDS leaf, which contained a well-developed crack upon receipt at the hot cell; it was subsequently fully fractured to enable the fractography investigation. The images on the left of the figure show that the cracks followed intergranular paths. Also evident was a mud-crack morphology in the oxidized region indicative of precipitated salts from the reactor coolant. The in-cell fracture surface also appeared intergranular. This is attributed to the precipitation hardening of the matrix within the grain of the alloy 718 and not a true indication that the material had become brittle. The HDS operated at stress levels sufficient to promote crack propagation by an SCC mechanism.

Figure 4 shows the clean grain boundaries of the archive A sample, which represents the same batch of HDS as the irradiated material. The few delta precipitates observed in the irradiated samples were located within the grains and were of the blocky morphology associated with formation during the high temperature annealing treatment [3]. The manufacturing precipitation heat treatment of these samples included a comparatively rapid cooldown of approximately 485 °C / hour between 718 °C and 621 °C.

The crack network on the surface of the irradiated sample after oxide removal is shown in Figure 5. The oxide removal step exposed extensive slip lines surrounding the well-developed crack networks. This indicates slip line oxidation occurred during operation. Also evident are secondary cracks, which appear to be intra-granular and perpendicular to the external stress direction. A full discussion of this crack network is available in Reference [4].

Figure 6 contains images of a bend test sample, machined from the irradiated material, after straining to approximately 15%. Slip lines first appeared at approximately 1600 MPa, well below the yield strength of the material. Cracks initiated in the material, at locations of inclusions and on grain boundaries, at


4

approximately 3000 MPa. However, in the irradiated samples, there was no evidence that cracks initiated at inclusions [4].

The effect of irradiation on mechanical strength of the material was determined through the use of bend tests. Figure 7 shows a comparison of the bend stress versus plastic deformation of the irradiated material compared to archive A material. Significant softening due to irradiation did not occur.

Tensile tests were performed on both archive materials. The average results are reported in Table 2. Archive A showed a definitive trend to higher strength and lower elongation. This corresponds well to the microstructural differences seen in the two samples and is discussed below.

Microhardness measurements were performed on both irradiated and archive samples. Table 3 reports average values. A range is provided for the irradiated sample as measurements were obtained both near crack networks as well as at a distance from the cracked regions. While a slight difference is seen between archive A and archive B material, the irradiated samples do not show a significant impact of irradiation.

Figure 8 shows the microstructure of archive B, for comparison with the microstructure of archive A samples shown in Figure 4. Archive B contains significant levels of delta precipitation on the grain boundaries. Archive B was subjected to a cool-down of 55 °C / hour between 718 °C and 621 °C during the manufacturing precipitation heat treatment. This is in contrast to the 485 C / hour cooling rate for the archive A material.

A higher magnification of the grain boundary is shown in Figure 9 depicting a TEM image of archive B, showing the precipitate free zone (PFZ) associated with a delta precipitate on the grain boundary. While small, approximately 50 nm, the PFZ is of a similar size to those previously reported [5].

Figure 10 compares EBSD scans of archive A and archive B samples following a 3.7% bulk strain of bend samples. Inverse Pole Figure (IPF) maps show the grain structure and orientation of the two archive samples. Grain Reference Orientation Deviation (GROD) maps provide an estimate of the amount of localized strain within individual grains, while the Kernel Average Misorientation (KAM) evaluate strain over a larger area of many grains, providing an estimate of the strain misorientation at the grain boundaries. The GROD maps show that the archive A sample contains more areas with high local misorientation (greater number of green areas). The KAM images show that the grain boundary strain localization is also much higher in the archive A sample.

4. DISCUSSION

Failed HDS batches were represented both by the irradiated samples and archive A material. Non-failed HDS batches were represented by archive B material. The primary difference identified between the archive samples was the presence of delta phase on most high angle grain boundaries found within the archive B sample.

Delta phase, Ni3Nb, forms at the expense of the γ” strengthening precipitate. When delta precipitates are present on the grain boundaries in semi-continuous or continuous distribution, an area adjacent to the grain boundary becomes depleted in γ” (Figure 9). This has been shown to increase the local plasticity of the material in this grain boundary adjacent region [6]. The result of the difference in delta precipitation and the PFZ between archive A and B samples is well demonstrated in the mechanical properties shown in Table 2. Archive B had such a significantly increased quantity of delta precipitation on the grain boundaries that local plasticity effects (and reduced strengthening in the PFZs) result in a reduction in bulk material strength and increased elongation. It is believed that these microstructural differences led to the observed differences in localized strain between the archive samples, as shown in Figure 10. This mechanism is illustrated in Figure 11, showing the impact of a PFZ, or area denuded in γ” strengthening precipitates. It is believed that once the slip lines interact with the PFZ, cross-slip occurs across the grain boundary, lowering the stress and activating multiple slip systems.


5

Therefore, while high stresses, beyond the material yield strength, are present for all HDS during operation, the microstructure of certain batches of springs, when similar to archive B, is able to better accommodate these stresses without initiating cracks where the slip bands impinged on the grain boundaries.

The microstructural differences observed between the archive samples developed during the precipitation heat treatment process. The difference between the heat treatments of the archive samples was the cool-down rate between 718 °C and 621 °C. Archive A was cooled at 485 °C / hour while archive B was cooled at 55 °C / hour. A slower cooling rate results in material held for a longer time in the temperature range necessary to form delta precipitates on the grain boundaries.

5. CONCLUSION

Thorough metallurgical exams using OM, SEM, TEM, as well as evaluation of strain distributions in plastically deformed samples have been used to characterize both the irradiated HDS as well as archive samples. It is speculated that the cracks found in the irradiated HDS were initiated through a complicated mechanism of contributing factors. The primary contributing factors were the high level of stress experienced by these HDS as well as a specific microstructure that was unable to accommodate that stress.

It is believed that in the absence of a continuous or near-continuous distribution of delta phase precipitates on the high angle grain boundaries and the corresponding PFZs, the material is unable to relieve the local stresses formed at grain boundaries by heavy slip bands. Under high plastic deformation, dislocation slip bands built up on the grain boundaries, eventually leading to crack initiation to relieve stresses. Once initiated, the cracks appear to have propagated through an irradiation-assisted stress corrosion cracking mechanism.

REFERENCES

[1] AMS 5596, “Nickel Alloy, Corrosion and Heat-Resistant Sheet, Strip, Foil and Plate. 52.5Ni-19Cr-3.0Mo-5.1Cb(Nb)-0.90Ti-0.50Al-18Fe. Consumable Electrode Remelted or Vacuum Induction Melted 1775 °F (968 °C) Solution Heat Treated”

[2] Miglin, B.P., et. al. “Stress Corrosion Cracking of Commercial-Grade Alloy 718 in Pressurized Water Reactor Primary Environment,” 4th International Conference of Environmental Degradation, 1989

[3] Deleume, J., et. al. “Influence of δ phase precipitation on the stress corrosion cracking resistance of alloy 718 in PWR primary water,” Journal of Nuclear Materials 382 (2008) 70-75.

[4] Leonard, K.J., et. al. “Characterization of Materials Properties and Crack Propagation Mechanisms in Damaged Alloy 718 Leaf Springs Following Commercial Reactor Exposure,” 17th International Conference on Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors, 2015.

[5] Burke, M. G., et. al. “Stress Corrosion Cracking of Chemistry Variants of Alloy 718 Part 2: Microstructural Characterization,” 6th International Symposium on Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors, 1993.

[6] Zhang, Yun, et al. “Delta Phase and Deformation Fracture Behavior of Inconel 718,” Superalloys 718, 625, 706 and Various Derivatives, 1997.

NOMENCLATURE

EBSD Electron Backscatter Diffraction

EDS Energy Dispersive Spectrometer


6

EELS Electron Energy Loss Spectroscopy

FIB Focused‐Ion Beam

GROD Grain Reference Orientation Deviation

HAADF High Angle Annular Dark Field

HDS Hold Down Spring

IPF Inverse Pole Figure

KAM Kernel Average Misorientation

OM Optical Microscope

ORNL Oak Ridge National Laboratory

PFZ PWR RT SCC

Precipitate Free Zone Pressurized Water Reactor Room Temperature Stress Corrosion Cracking

SEM Scanning Electron Microscope

STEM Scanning Transmission Electron Microscope

TEM Transmission Electron Microscope

TABLES

Table 1: Typical chemical composition for precipitation heat treated alloy 718, representative of the spring leaves in this study

Composition

Min (%) Max (%)

C ‐‐ 0.08

Mn ‐‐ 0.35

Si ‐‐ 0.35

P ‐‐ 0.015

Si ‐‐ 0.015

Cr 17.00 21.00

N 50.00 55.00

Mo 2.80 3.30

Nb 4.75 5.50

Ti 0.65 1.15

Al 0.20 0.80

Co ‐‐ 1.00

Ta ‐‐ 0.05

B ‐‐ 0.01

Cu ‐‐ 0.30

Fe remainder


7

Table 2. Average measured mechanical properties of archive samples

Avg. Yield Strength

(MPa)

Avg. Ultimate Tensile

Strength (MPa)

Avg. Uniform Elongation

(%)

Avg. Total Elongation

(%)

Archive A 1402 1598 15.4 24.9

Archive B 1297 1514 16.7 25.9

Table 3. Average Vickers microhardness measurements of archive and irradiated samples. 500-g load, 15 second dwell time

HV500

(kg/mm2)

Archive A 495

Archive B 470

Irradiated 450 - 500

FIGURES

(a) (b)

Figure 1. Pictures of the HDS showing (a) individual alloy 718 cruciform style leaf spring prior to assembly into (b) a HDS pack. The region within the dashed box of (a) corresponds to the relative location on a leaf spring of the images shown in Figure 2.


8

(a) (b)

(c)

Figure 2. Select still images (a, b) from a video of the damaged HDS packs. The images shown are representative of the severity of damage found throughout the examined samples. The areas depicted in these images correspond to the region within the dashed box of Figure 1; multiple leaves within the HDS pack are shown. (c) Higher magnification optical micrograph of a highly cracked location showing multiple crack initiation sites.


9

Figure 3. SEM micrographs from a location examined on the fracture surface. An oxidized fracture surface is present along the cracked leaf spring exposed during in-service operation. The fractographs on the left show a typical intergranular fracture surface. However, due to precipitate hardening of the matrix phase, fracture paths follow grain boundary pathways. The mud crack morphology in the oxidized region is likely dry precipitated salt from the reactor coolant.

Oxidized Region

Major Crack near Surface

Transition between pre-existing crack and in-cell fractured material


10

(a) (b)

(c)

Figure 4. TEM micrographs of (a) a grain boundary and (b) tri-junction observed in the archive A material. Note the lack of grain boundary precipitation. The archive A sample represents the same batch of HDS as the irradiated samples. Image (c) shows a collection of precipitate phases thought to be blocky delta precipitates that are not located on grain boundaries.


11

Figure 5. SEM micrographs showing examples of secondary crack networks near large cracked structures in the irradiated material following oxide removal. These secondary cracks (red arrows) are believed to be transgranular. As a rule, they are perpendicular to the expected external stress direction.


12

Figure 6. Cracks at the surface of the irradiated material following bend testing. The cracks shown initiated at grain boundaries.

Figure 7. Engineering bend curves showing bend stress vs. plastic deformation. A comparison of archive A to the irradiated material.

Archive A Sample

Irradiated Sample


13

(a) (b)

Figure 8. TEM images correlating (a) bright field and (b) High Angle Annular Dark Field (HAADF) images of a near continuous delta phase development along a grain boundary in the archive B material. The precipitate phase terminates at the intersection of a twin boundary. (b) The HAADF image was taken under scanning tunneling electron microscopy (STEM) conditions.

Figure 9. Dark field TEM micrograph using the γ” reflections off of the g=200γ two beam condition showing the presence of a denuded zone near the delta precipitation along a high angle grain boundary. The width of the PFZ is approximately 50 nm.

Delta phase

Twin Boundary


14

(a)

(b)

(c)

Figure 10. EBSD images from scans (500X magnifications, 0.5 µm step) for the deformed surface of both archive A and archive B samples. (a) IPF maps, (b) GROD maps, (c) KAM maps. Color scales are the same for both archives (0 to 25° and 0 to 5° for GROD and KAM maps, respectively). The white dashed oval shows an area in the archive B material with low near-grain boundary misorientation level. GROD and KAM maps show the archive A sample contains greater levels of misorientation compared to the archive B sample.

A B

A

A

B

B


15

(a) (b)

(c) (d)

Figure 11. Schematic diagrams explaining the proposed mechanism for crack initiation at intersection of slip lines and grain boundaries in the absence of a precipitate free zone (PFZ). Relative local material yield strength across a grain boundary for a material which (a) has no delta phase on the grain boundary and thus no PFZ and (b) which contains a delta phase film on the grain boundary and associated PFZ. (c) shows a schematic representation of dislocations on a slip line initiating a crack where stress associated with the dislocations builds at the grain boundary. (d) shows the lower strength PFZ permitting cross-slip of the dislocations, resulting in a lower stress within the material and no crack initiation.

Dislocations / Slip Line

Dislocations /

Slip Line

Grain Boundary Grain Boundary with Delta Phase Film

Crack

Initiation

Dislocations / Slip Line

PFZ

hold down spring failure analysis

Documents