805

A VERSATILE APPROACH FOR THE DEVELOPMENT OF RELIABLE ULTRASONIC INSPECTION TECHNIQUES FOR SPENT NUCLEAR FUEL CANISTER INSERTS

Ulf Ronneteg, SKB (Swedish Nuclear Fuel and Waste Management Co), Oskarshamn, Sweden,

Thomas Grybäck, SKB, Robert RISBERG, Exova, Linköping, Sweden, Marija Bertovic and Mato Pavlovic, BAM (Federal Institute for Materials Research and Testing), Berlin, Germany

ABSTRACT. The Swedish KBS-3 design for the disposal of spent nuclear fuel is based on encapsulation of the fuel in canisters consisting of a nodular cast iron insert surrounded by an outer 5 cm thick shield of copper. The insert is the vital component to withstand the mechanical loads in the repository while the copper shell serves as a corrosion barrier. To make sure that the canisters will fulfil the requirements, an extensive programme for quality control is developed. In this programme the use of non-destructive testing (NDT) is vital and for the insert the use of ultrasonic inspection, primarily phased array techniques, are dominant. In order to develop a reliable inspection system a versatile approach is used in the development phase. This approach includes the use of a number of tools that have been used during the development of the ultrasonic inspection techniques. In order to evaluate the technical performance of the techniques, digital radiography and computed tomography have been used combined with destructive sectioning for defect characterization. Ultrasonic modelling have been used both for understanding of the used techniques and for definition of new techniques and probes. The procedures have been analysed with respect to human factors, with methods such as Failure Modes and Effects Analysis (FMEA) and with the use of the eye tracking methodology and the results have been used in the further development of reliable inspection procedures.

INTRODUCTION The Swedish Nuclear Waste and Management Company (SKB) is responsible to take care of all radioactive waste produced in Sweden. This includes first of all of the waste produced by the nuclear power plants but also from the traditional industries, research and from the medical field (see Figure 1). The spent nuclear fuel, that is the most critical portion of the waste, is stored for minimum 30 years in the central interim storage (Clab). The fuel is stored in water basins 30 metres below ground and the water works both as radiation shield and as coolant for the fuel. The operational waste is stored, surrounded by concrete, some tens of metres below ground in the final repository for operational waste (SFR). These two storages together with the specially build ship “Sigyn”, used for transportation of the fuel; have been running for more than 20 years.

What still needs to be built is an encapsulation plant where the fuel will be encapsulated and a final repository for the spent nuclear fuel. During 20 years SKB has been working on finding a place for the final repository. To begin with, feasibility studies were conducted in eight areas spread out in Sweden and from year 2002 site investigations were conducted at the areas of Forsmark and Oskarshamn. Results from these site investigations, were carefully compared. This resulted in that SKB in the beginning of June 2009 could decide, based on the long-term safety, to choose Forsmark as the site for the final repository. The next step in the program was to send in an application for building of the encapsulation plant in connection with Clab in Oskarshamn and the final repository in Forsmark. The application was sent to the authorities in spring 2011 and after approval from the authorities the construction work will start in 2017 and finally the deposition of canisters will start in 2025.

806

Figure 1. SKB’s system for managing the radioactive waste in Sweden.

SKB is developing a special method, called the KBS-3 method, for final disposal of the spent

nuclear fuel. The method is based on a three protective barriers, see Figure 2. The first barrier is the

canister composed of a nodular cast iron insert surrounded by a copper shell. The impermeable copper

canisters are then embedded in bentonite clay that serve as the second buffer and finally placed in the third

barrier, the crystalline basement rock at a depth of about 500 metres. Finally after the canisters have been

disposed, the tunnels and rock caverns will be sealed.

Figure 2. The KBS-3 system with the different barriers.

To be able to develop the techniques that will be used for building the deep repository and deposit

the canisters as well as make research on the properties of the bed-rock, SKB has built a full scale

laboratory (Äspö hard rock laboratory) in Oskarshamn with tunnels down to 450 metres below ground.

For developing the encapsulation technique, SKB has built the Canister Laboratory in Oskarshamn. In this

laboratory technique for welding and inspection of canisters are developed and tested in full-scale.

THE CANISTER

The copper canisters that will encapsulate the spent nuclear fuel are nearly five metres long and over one

metre in diameter (REF 1). The weight is between 25 and 27 tonnes including about 2 tons of spent

nuclear fuel. The canister, see Figure 3, is composed of a nodular cast iron insert and is manufactured in

two different types, the PWR type with four channels and the BWR type with twelve channels. The main

functions of the insert are to contain the fuel and withstand the mechanical loads in the repository. The

insert is surrounded by a five centimetre thick copper shell that serves as the corrosion barrier. The copper

807

shell consists of a lid and a base that is welded to the tube by means of friction stir welding (FSW).

Figure 3. The copper canister.

THE CANISTER LABORATORY

The Canister Laboratory (CL), see Figure 4, in Oskarshamn was initially, in 1998, built to develop

techniques for full size welding of the copper lid to the tube by electron beam welding, develop NDT

techniques for weld inspection and test the handling of the canister in a similar way as for the planned

encapsulation plant. The inspection of the welds was performed by means of digital radiography, by the

use of a linear accelerator and a linear detector array, and by phased array ultrasonics. In 2002 a system for

friction stir welding was installed and thereby the inspection techniques were adapted for this new type of

welding.

In 2005 it was decided to focus on the whole canister and not only the lid weld. Full scale

inspection systems was build to facilitate development of inspection techniques for the copper lid, base

and tube as well for the nodular cast iron insert. In addition, to get a better understanding of the

manufacturing processes, the CL also became the centre for development of the manufacturing techniques

for canister components.

Figure 4. The Canister Laboratory.

NDT METHODS

The development of NDT techniques for the canister components and welds is conducted in an iterative

way at the Canister Laboratory. This means that first the basic technique is tested and if the result is

promising the technique is further developed and tested in full-scale. After this the technique is evaluated

808

against the actual requirements and the need for optimization is identified together with possible needs for

complementary inspections.

NDT Equipment

Full-scale inspection systems for the canister components and welds have been built up at the Canister

Laboratory. This includes systems for digital radiography and ultrasonic testing of the friction stir welds

(Figure 5) as well as systems for ultrasonic testing of the canister components, i.e. the copper tube, lid and

base as well as the cast iron insert (see Figure 6 and 7). The manipulators used for ultrasonic inspections

have been built in a flexible way, meaning it’s not fixed to the inspection equipment to be used. This

means that the systems can be used in combination with various inspection instruments, i.e. different

ultrasonic instruments, eddy current instruments etc. To increase the reliability of the inspection of the

large components (copper tube and cast iron insert), the reference blocks are integrated in the inspection

system, giving the same geometry for the reference blocks as for the real components.

Figure 5. System for digital radiography (left) and ultrasonic inspection (right) for the friction stir welds.

Figure 6. Ultrasonic inspection system for the copper tube and cast iron insert.

809

Figure 7. Ultrasonic inspection system for the copper lid and base.

Preliminary NDT techniques for the cast iron insert

The nodular cast iron insert that shall withstand the mechanical loads in the repository has different

requirements for different areas of the volume. In general it can be stated that the requirements are tougher

close to the surface than in the interior part (REF 1). On the basis of this the insert has been divided into

three areas from inspection point of view according to Figure 8, where the lilac area represents the near-

surface, the green area the thick interior part and the yellow area the volume between the channels. For

these testing areas, the following reference methods are currently being tested:

• Angle incidence ultrasound testing, using contact probes according to Figure 9, of the near-surface

area of the insert (lilac) in order to detect volumetric and crack-like defects.

• Normal incidence ultrasound testing, using a linear array in local immersion according to Figure

9, of areas between the surface of the insert and the channel tubes (green) in order to detect volumetric

defects.

• Transmission ultrasound testing, using contact linear arrays according to Figure 9, of the area

between the channel tubes (yellow) in order to detect volumetric defects.

•

Figure 8. Inspection areas of the BWR-insert.

Angle testing of the near-surface area of the insert is carried out in four directions by means of the

so-called TRL technique (Transmitter Receiver Longitudinal, a double-crystal probe that generates

longitudinal sound waves) which is used for the testing of reactor vessels within the nuclear power

industry. Both normal testing and transmission testing are performed with phased array ultrasonics

developed from conventional technology that is normally used for the testing of various types of castings.

Figure 9. Angular ultrasonic inspection (left), phased array ultrasonic transmission inspection (middle) and

810

local immersion phased array ultrasonic inspection (right) of the BWR-insert.

Experiences from NDT of the cast iron insert.

During 2008-2011 some tens of cast iron inserts have been inspected with the developed preliminary NDT

techniques and have resulted in a lot of achieved experiences. Some of the experiences are general for all

techniques while others are specific.

General experiences

The results from the inspections have shown that the acoustic properties of the nodular cast iron is better

than expected which facilitates the use of higher ultrasonic frequency than initially applied. The results

from the inspections show that the presence of large defects giving high amplitudes is very limited while

the presence of low amplitude indications is quite frequent. The results also show big differences between

the amounts of indications from the different inspection techniques.

Experiences from TRL ultrasonic inspections

The TRL inspections have resulted in numerous indications in many of the inspected inserts, see example

of amount in Figure 10. Among the indications two types of low amplitude indications have been

identified; wide spread and point-source. The wide spread indications was assumed to originate from some

kind of cluster of defects while the point-source indications was assumed to originate from small

individual indications and due to the large amount of indications suspicions were entertained if they all

were relevant. Another experience from the TRL inspection was that the evaluation of data was very

complicated and time consuming due to the fact that echoes from several wave modes were received in the

inspection range, see Figure 11.

Figure 10. Number of TRL indications in one BWR-insert.

811

Figure 11. Ultrasonic modelling of the TRL inspection. Echoes originating from the channel tubes.

Experiences from transmission ultrasonic inspections

The use of the transmission ultrasonic inspection technique met a number of obstacles. To begin with, the

temporary fixture used was too weak giving the effect that it was sensitive to miss-alignment of the two

transducers. The most interesting finding was that for a number of BWR-inserts the transmitted wave got

totally lost for most of the insert length. This was first seen as a problem with the inspection technique, but

after more careful investigation it was found out that the channel tubes, which the sound should pass

between, were bent and that there were no clear sound path between them. This finding was then used as

an input to develop the casting process in order to prevent this effect. Regarding the inspection technique

it was found that it was not possible to achieve a good focus at larger depth due to the relative short near-

field generated by the phased array transducers used.

Experiences from local immersion phased array ultrasonic inspections

The local immersion phased array inspections have resulted in few indications and in combination with the

results from the TRL inspections some questions were raised:

• Are there ”no” defects at larger depths?

• Is TRL70º really that much better than the used phased array method?

Based on these observations some conclusions were drawn. First of all, the inspection cover did not

the area close to the surface where the larger TRL indications were found and secondly, POD studies

showed that the detection capabilities (POD) did decrease at larger depths, see Figure 12, and this could be

explained by the use of a not optimal ultrasonic beam.

In addition to defect detection the technique has, since the findings with the transmission technique

that the channel tubes could be bend, successfully been used to measure the positions of the outer corner

of the fuel channels and thereby verify their straightness.

812

Figure 12. Ultrasonic detection capabilities (POD) decrease at larger depths

DEVELOPMENT OF RELIABLE NDT METHODS

Based on the experiences above a number of “tools” have been used to identify the weaknesses and the

possibilities to improve the inspection techniques:

• NDT Knowhow

• Computed tomography

• Destructive sectioning

• Ultrasonic modelling

• POD studies

• Human factor studies.

Technical development

As mentioned in the previous chapter a lot of indications were found during the TRL inspections and in

order to understand their origin a number of different tests were performed. For example in areas with the

small point-like indications, samples were cut out, x-rayed and finally examined by metallographic

inspection and the results showed defects in the range of 1mm and that these could be judged to be

irrelevant.

The wide spread indications were examined in several steps. First a more sensitive local immersion

phased array technique was developed in order to accurate position the defects. Samples were then cut out

in these areas, the samples were examined by computed tomography, both as large samples (50mm) and as

small samples (15mm) and finally after tensile testing the fracture zones were examined by metallographic

inspection, see Figure 13.

813

Figure 13. Characterization of ultrasonic indications. Ultrasonic results to the left showing the A-, B-, C-

and D-scan of an indication. In the C-scan the yellow capture frame shows the area where the test sample

is cut out. The image in the middle show the CT results with a cross-section of the indicated porosity on

top. To the right a metallographic image show one of the individual pores (size ~1mm).

As mentioned in the previous chapter the evaluation of the TRL data was complicated. Therefore an

extensive modelling study was initiated to understand the behaviour of the TRL and to investigate the

possibilities to replace the TRL-probes with shear waves and achieve the same or better sensitivity. The

results showed that this easily could be achieved by the use of a phased array shear wave probe and as can

be seen in Figure 14 and Table 1, the expected sensitivity will clearly be increased.

Figure 14. Modelling of phased array shear probe (top) and TRL-probe (bottom).

Ultrasonic modelling has also been used as a tool to develop new technique for the inspection

between the channels where two large linear arrays are used to inspect the whole volume by both pulse-

echo and transmission technique, see Figure 15.

814

Ligament [mm] Circle [6 mm] Elliptical [18x3

mm]

SDH [2 mm]

New Probe 15 16.7 20.1 26.6 TRL-probe 15 3.1 12.9 22.1 New Probe 30 18.7 24.1 31.9 TRL-probe 30 0 6.8 18.2 New Probe 45 11.7 20.2 26.8 TRL-probe 45 0 6.3 16.7

Table 1. Signal to noise ratio for TRL-probe and phased array shear probe.

Figure 15. Modelling of the inspection between the fuel channels. The two images to the left show

the field from different transmitting focal laws while the image to the left shows the principles for the

receiving aperture.

Human factor perspective

It was realized during the evaluation of the inspection techniques that, even though the manual inspections

had been replaced with mechanized inspections, a number of manual operations could affect the reliability

of the inspection. Based on this, various studies of the human factors were initiated.

One of these studies included a risk identification method Failure Modes and Effects Analysis

(FMEA), customized for the identification of errors committed by human operators, rather than by the

technical equipment, and it was performed in several steps, focusing both on the physical inspection and

the data evaluation. Using this method, a group of NDT experts identified possible errors, as well as the

consequences (should an error occur) and the risk for an error to happen. Finally, existing and possible

barriers, used to prevent the errors, were identified and further analysed. These barriers were then used in

the further development of the inspection procedures. Possible errors were identified in four different

phases: preparation, sensitivity settings, inspection and data evaluation and the most severe errors were

connected to the sensitivity setting and the data evaluation (REF 3, 4).

It was also identified that even if the techniques are totally reliable, there still is no guarantee for

reliable inspections, as the human still has to follow a procedure. Therefore, a study on the reliability of

the procedures was initiated. The study was performed both in a theoretical and experimental way. The

theoretical way was to study the PANI 3 study performed within the Health and Safety Executive, UK

(REF 2). One of the PANI 3 study’s aims was to study the current procedures and make suggestions for

815

their improvement. Some of the human factors experts’ suggestions for the improvement of the procedure

include having a short form procedure, involving only the most relevant steps, written in a user-friendly

stepwise manner, as an addition to the long form. The specific instructions should be written in active

voice, only one action per step and the steps should be written in the order they are to be performed.

The experimental part was performed using the eye tracking methodology during data evaluation.

The eye tracker follows eye movements on the computer screen during any task and records useful metrics

on where the operator is looking, how long or how frequent which can then be used to determine weak

spots in the procedure, as well as in the data evaluation task. The experiments were conducted by four

operators, whose task was to evaluate a set of data for ~1.5 hour according to one inspection procedure

available on the screen. During the experiments, the operators were observed by two persons, one focused

on the technical issues (Have all indications been found? Have all steps been performed according to the

procedure? Have all the steps been followed? Etc.) and one ensuring that all the eye movements data have

been correctly collected and stored. Afterwards, the reported indications were compared and, by the help

of the recorded data, the causes for errors were identified. During the observations, additional errors were

identified and these were also analyzed in order to determine the possible consequences. Typical errors

included defects being missed or wrongly interpreted and example of causes of errors was that the wrong

settings were used or that the procedure was misunderstood. After “all” errors have been analyzed,

possible ways for prevention have been identified which should set the base for the further development of

the inspection procedures (REF 3).

CONCLUSIONS

The development of inspection techniques has started by building up resources for inspection and in the

first stage develop preliminary ultrasonic inspection techniques. Based on the results from full-scale

inspections of a number of inserts, a number of questions have been identified. By the help of various

tools, like ultrasonic modelling, POD calculations, computed tomography and destructive sectioning a

deeper understanding of the developed techniques has been reached. And in the same way these tools have

been used to further develop the techniques to achieve more sensitive and more reliable inspections.

Finally the way of doing the inspection and data evaluation has been studied from a human factor

perspective by the use of a customized FMEA, PANI 3 recommendations and eye-tracker experiments to

further develop reliable inspection procedures.

REFERENCES

1) SKB (2010). Design, production and initial state of the canister, SKB Report TR-10-14.

2) McGrath, B. A. (2008). Programme for the Assessment of NDT in Industry. PANI 3. Prepared by

Serco Assurance for the Health and Safety Executive. Research Report RR617.

3) Bertovic, M., Fahlbruch, B., Müller, C. Pitkänen, J., Ronneteg, U. (2012). Human Factors Approach to

the Acquisition and Evaluation of NDT Data. Proceedings of the 18th WCNDT, 16-20 April 2012,

Durban, South Africa.

4) Bertovic, M., Fahlbruch, B., Müller, C. Pitkänen, J., Ronneteg, U., Gaal M., Kanzler, D. & Ewert, U.

(2010). Human Factors Approach to the Reliability of NDT in Nuclear Waste Management in Sweden

and Finland. Proceedings of the 8th International Conference on NDE in Relation to Structural

Integrity for Nuclear and Pressurized Components, 29th Sep- 1

st Oct 2010, Berlin, Germany.