IAEA-TECDOC-1345

Fuel failure in water reactors:

Causes and mitigation
Proceedings of a Technical Meeting
held in Bratislava, Slovakia, 17–21 June 2002

March 2003
The originating Section of this publication in the IAEA was:
Nuclear Fuel Cycle and Materials Section
International Atomic Energy Agency
Wagramer Strasse 5
P.O. Box 100
A-1400 Vienna, Austria

FUEL FAILURE IN WATER REACTORS:

CAUSES AND MITIGATION
IAEA, VIENNA, 2003
IAEA-TECDOC-1345
ISBN 92–0–101703–0
ISSN 1011–4289
© IAEA, 2003
Printed by the IAEA in Austria
March 2003
 FOREWORD

At the invitation of the Government of Slovakia and in response to a proposal by the IAEA
Technical Working Group on Water Reactor Fuel Performance and Technology (TWGFPT),
the IAEA convened a Technical Meeting on Fuel Failure in Water Reactors: Causes and
Mitigation in Bratislava, Slovakia from 17 to 21 June 2002. The meeting was hosted by the
Nuclear Regulatory Authority and VUJE Trnava, Inc. Engineering, Design and Research
Organization.

For many years, the IAEA has been closely involved in the analysis of nuclear fuel
performance, basic fuel failure causes and failure mechanisms in water power reactors. The
IAEA conducted a Technical Committee Meeting on Fuel Failure in Normal Operation of
Water Reactors: Experience, Mechanisms and Management in 1992 in Dimitrovgrad, Russian
Federation (IAEA-TECDOC-709), conducted a study on fuel failures and published Technical
Reports Series No. 388, Review on Fuel Failures in Water Cooled Reactors, in 1998 and
conducted a survey on fuel failures in water cooled power reactors from 1995–1998
(CANDUs, BWRs, PWRs and WWERs) in 2000. The objective of this meeting was to
analyse and discuss utility and fuel vendor experience in fuel failure cause identification and
on implemented remedies to reduce the number of fuel failures and/or to mitigate fuel failure
impact on NPP operation.

Fifty-three specialists in fuel design, fabrication and operation from 18 countries took part in
the meeting in order to gather and discuss existing knowledge of the subject and identify the
need for further efforts. Twenty-six papers were presented in five sessions covering
experience with recent fuel failure events and their mitigation, and the current knowledge of
fuel failure mechanisms in light water cooled power reactors. The most frequently observed
events included grid-to-rod fretting failures in PWRs, severe secondary failures, especially the
long axial splits and circumferential fractures observed in BWRs, axial offset anomalies in
PWRs and some others. During recent years fuel performance in water-cooled power reactors
has improved significantly since most early problems have been solved. Fuel failure rates are
now at a low level (<10-5 or <10 ppm) and continue to be reduced. However, fuel failure has
remained a very important issue for NPP operation and economics.

The IAEA wishes to thank all the participants for their contribution to the meeting and to this
publication, especially V. Petenyi of VUJE Trnava, Inc. Engineering, Design and Research
Organization who co-ordinated the work of the local organizational committee. The IAEA
officer responsible for this publication was V. Onoufriev of the Division of Nuclear Fuel
Cycle and Waste Technology.
 EDITORIAL NOTE

This publication has been prepared from the original material as submitted by the authors. The
views expressed do not necessarily reflect those of the IAEA, the governments of the nominating
Member States or the nominating organizations.
The use of particular designations of countries or territories does not imply any judgement by
the publisher, the IAEA, as to the legal status of such countries or territories, of their authorities and
institutions or of the delimitation of their boundaries.
The mention of names of specific companies or products (whether or not indicated as
registered) does not imply any intention to infringe proprietary rights, nor should it be construed as
an endorsement or recommendation on the part of the IAEA.
The authors are responsible for having obtained the necessary permission for the IAEA to
reproduce, translate or use material from sources already protected by copyrights.
 CONTENTS

Summary .................................................................................................................................... 1

GENERAL OVERVIEW OF PRESENT STATUS OF EXPERIENCE (Session 1)

Experience and Reliability of Framatome ANP’s PWR and BWR fuel .................................. 21
W. Klinger, C. Petit, J. Willse
Spanish experience with LWR fuel: General overview ........................................................... 30
J.M. Conde López, M. García Leiva
Probabilistic-statistical analysis of WWER fuel element leaking causes and
comparative analysis of the fuel reliability indicator on NPPs with
WWER and PWR reactors .................................................................................................. 41
I. Chestakov
Fuel assembly chemical cleaning............................................................................................. 59
J. Schunk, M. Beier, F. Kovacs, S. Micó, P. Tilky,
H.-O. Bertholdt, I. Janzik, G. Marquardt
KNPP practices in assurance of reliable fuel operation for WWER-440 reactors................... 72
V. Tzotcheva
Ukrainian WWER-type NPP units. Results of cladding tightness inspection ......................... 77
N.Yu. Shumkova., O.V. Bykov, L.P. Belousova

MITIGATION OF FAILURES BY DESIGN AND MANUFACTURING (Session 2)

Mechanical design of the triplewave debris filter .................................................................... 87

A. Soderlund, R. Eklund, S. Helmersson
Fuel failures at ANGRA 1: Cause and mitigation.................................................................... 92
J.L.C. Chapot, R. Suano, N. do Couto, J.A. Perrotta, J.E.R. da Silva,
L.A.A. Terremoto, M. Castanheira
Operation experience of WWER-440 fuel assemblies and measures
to increase fuel reliability .................................................................................................. 117
V. Chirkov, V. Novikov, A. Sharikov
Fuel failure mitigation at the Ringhals plant .......................................................................... 123
T. Andersson

EXPERIMENTAL STUDIES OF FUEL FAILURE AND DEGRADATION

MECHANISMS (Session 3)

PWR fuel failure analysis due to hydriding based on PIE data ............................................. 137
Yong-Soo Kim
Outside-in failure of BWR segment rods during power ramp tests ....................................... 148
H. Hayashi, Y. Etoh, Y. Tsukuda, S. Shimada, H. Sakurai
Post-irradiation examinations of WWER-440 FA provided with
stainless steel spacer grids ................................................................................................. 164
A.V. Smirnov, V.P. Smirnov, D.V. Markov, V.S. Polenok, B.A. Kanashov, V. Shishin
Changes in geometry of claddings and fuel columns of spent WWER-440 and
WWER-1000 fuel rods under steady-state and transient operating conditions................. 171
B. Kanashov, S. Amosov, G. Lyadov, D. Markov, V. Ovchinnikov,
V. Polenok, A. Smirnov, A. Sukhikh, Ye. Bek,, A. Yenin,, V. Novikov
Failure root cause of a PCI suspect liner fuel rod .................................................................. 188
F. Groeschel, G. Bart, R. Montgomery, S.K. Yagnik
Model development of fuel failure in water reactors due to cladding hydrogenization ........ 203
E.Yu. Afanasieva, I. A. Evdokimov, V.V. Likhanskii, A.A. Sorokin, V.V. Novikov

MITIGATION OF FAILURES/DEGRADATION BY PLANT OPERATION (Session 4)

Development of fuel performance code FEMAXI-6 and analysis of

mechanical loading on cladding during power ramp for high burn-up fuel rod................ 217
M. Suzuk, H. Uetsuka
Operation and fuel design strategies to minimise degradation of failed BWR fuel............... 239
P. Rudling, T. Ingemansson, G. Wikmark

DETECTION AND MONITORING (Session 5)

Fuel reliability of Bohunice NPP ........................................................................................... 257

M. Kaþmar, J. BeĖa, I. Smieško
Failed rod diagnosis and primary circuit contamination level determination
thanks to the Diademe code............................................................................................... 265
D. Parrat, J.B. Genin, Y. Musante, C. Petit, A. Harrer
Disadvantages of means and methods of fuel failure detection ............................................. 277
O.O. Depenchuk
Fuel failures at Paks NPP ....................................................................................................... 284
A. Kerkápoly, N. Vajda, A. Csordás, Z. Hózer, T. Pintér
Defected fuel monitoring at Cernavoda nuclear power plant ................................................ 293
E. Gheorghiu, C. Gheorghiu
Fuel failure diagnostics in normal operation of nuclear power plants
with WWER-type reactors................................................................................................. 303
L. Lusanova, V. Miglo, P. Slavyagin
Summary of technical development on the on-line monitoring and fuel failure
evaluating system at the Temelin NPP .............................................................................. 315
M. Semmler, M. Martykan, J. Cizek, M. Valach, J. Hejna
Regulation of the fission product activity in the primary coolant and assessment
of defective fuel rod characteristics in steady-state WWER-type reactor operation......... 326
P. Slavyagin, L. Lusanova, V. Miglo

List of Participants ................................................................................................................. 339

 SUMMARY
INTRODUCTION

The objective of this technical meeting (TM) was to review the present knowledge of the
causes and mechanisms of fuel failure in water reactors during normal operational conditions.
Emphasis has been given to analysis of failure causes and their mitigation by means of design
as well as plant and core operation including strategies for operation with failed fuel. Some
information on detection techniques (on-line monitoring and diagnostics, flux tilting, sipping
techniques, etc) has also been presented.

This TM presented also the progress on the above-mentioned subjects since the last meeting
held in 1992 (Dimitrovgrad, Russian Federation).

The topics covered in the papers were as follows:

• Experience feedback on fuel reliability (8 papers)

• Strategies to avoid or mitigate fuel failures (4 papers)

• Experimental studies on fuel failures and degradation mechanisms (4 papers)

• Modelling of fuel failure mechanisms (3 papers)

• Detection and monitoring during operation or outage (4 papers)

• Modelling and assessment of fuel failures (3 papers)

SESSION I: GENERAL OVERVIEW OF PRESENT STATUS OF EXPERIENCE

Chairmen: P. Darilek (VUJE Trnava Inc., Slovakia and KAERI, Republic of Korea) and J.M.
Alonso Pacheco (ENUSA, Spain)

In this first session of TM a total of six papers were presented. The papers provided a
comprehensive summary of fuel performance experience and practices for enhancing fuel
reliability in different countries like Spain, Russian Federation, Bulgaria, Hungary and
Ukraine, or from fuel vendors like Framatome-ANP. The discussion at the end of the Session
was devoted to the identification of present major fuel failures root causes and remedies, and
to establish the guidelines for future research efforts that ensure maximum fuel reliability.

The following fuel failures root causes were identified:

• Debris
• Pellet-Cladding Interaction (PCI, particularly in BWRs)
• Manufacturing defects
• Flow Induced Vibrations, that includes different mechanisms and conditions like:
baffle jetting, enhanced cross flow at core periphery, enhanced cross flow at core
bottom (vessel flow anomaly), mixing vane distribution and orientation in the grids

1
 • Fuel Handling Incidents
• Poor resistance to secondary degradation (particularly in BWRs).

The following remedies were applied:

• Debris filter devises that retain and catch foreign particles preventing fuel rod damage.
This type of remedy, in conjunction with NPP’s efforts for minimising debris sources,
has proved efficiency since debris is not major fuel failure mechanism anymore
• Improving mechanical performance against flow induced vibrations. That it is being
accomplished by addressing two main parameters:
o Reduced vibration response (i.e., adequate overall fuel assembly mechanical
design, good grid design…)
o More resistance to fretting wear (contacting materials, grid design…)
• Barrier cladding has proved to be effective protection against PCI failures in BWR.
Iron addition to the barrier is the general accepted remedy for preventing secondary
degradation of failed barrier fuel rods.
• To improve manufacturing and inspection techniques, besides the implementation of
more robust and efficient quality systems as a way to eliminate manufacture related
defects.

• The following further research efforts were recommended:

• Fuel surveillance. Not only failed fuel needs to be investigated but characterisation of
key properties of representative fuel is required (corrosion, growth, fretting marks…).
• Minimise number of failed fuel cases where no inspection is accomplished.
• In case of PWR that needs to operate under load follow condition, PCI resistance
needs to be satisfactorily addressed.
• Improve understanding of Flow Induced Vibration mechanism and effects including:
o Extensive inspection of potentially affected fuel;
o Better analysis of reactor specifics and circumstances;
o Representative fuel assembly testing loops and procedures;
o More powerful analytical tools and models.
• Exchanging fuel failure experience more openly.

SESSION II: MITIGATION OF FAILURES BY DESIGN AND MANUFACTURING

Chairmen: W. Klinger (Framatome-ANP, Germany) and A. Bykov (NAEC, Ehergoatom,
Ukraine)
The session showed that fuel vendors face a task to provide a fuel design that would assist
avoiding main actually observed fuel failure mechanisms. Information on fuel behaviour from
operation, PIE, and modelling is the basis for taking a decision on modification of fuel rod
(FR) or fuel assembly (FA) design. Papers presented demonstrated that major fuel failure
causes are very similar for different reactor types. Of course, failure frequency for each cause
depends on specific reactor type, because of different fuel rod and assembly design, water

2
chemistry, coolant flow, thermal and mechanical loads, etc. Major modifications that have
been and are being done by fuel designers/producers are presented below.

Debris catchers/filters: The purpose of a filter is to catch foreign objects to avoid cladding
damage, but with only minor impact on pressure drop. Since the early 90ies, fuel vendors
have developed several debris filter designs. Therefore, most of current fuel deliveries include
a debris filter. With regard to Swedish presentation on Triple Wave debris filter, the
discussion showed that cladding damage by metallic debris is the most severe failure cause in
BWRs worldwide, making 40–50% of all failures. Debris filters were first used in PWR FAs
and demonstrated their efficiency drastically reducing the number of debris fretting related
failures. It is expected that advanced filters in BWRs will be very effective also. Debris filters
in WWER FAs are now under testing in commercial reactors.

Stiffening FA skeleton: To achieve an improved stiffness, the FA design should have an

optimized Guide Tube (GT) wall thickness (“as thick as possible”), it should have a thick-
walled dashpot, and should dispose of a stiff SG/GT (SG-Spacer Grid) connection (i.e. direct
welding or comparable technique). Such modified FA designs have been implemented during
the last years. In addition, new low-growth materials with reduced susceptibility to stress
relaxation have been introduced.

Fretting: Factors influencing rod/spring vibrations were understood and, to significant extent,
eliminated. Analysis of grid-to-rod fretting in Angra-I PWR in Brazil showed that deficiency
in FA design resulted in significant number of fuel rod fretting failures. During discussion of
the root failure cause, participants agreed upon the fuel vendor’s comment that the transfer of
so-called “proven for plant type A fuel design” to plant of type B was a fuel failure
prerequisite in Angra-I plant. Even minor changes in fuel design or operating conditions may
result in lowering fuel failure threshold, i.e. worsening fuel reliability. This factor has to be
taken into account.

PCI: PCI was the major failure mechanism in BWR fuel in the past, causing most of the fuel
failures until the mid 90ies. Introduction of cladding materials with increased PCI resistance
resulted in a drastic reduction of such failures.

Accelerated Corrosion: Fuel failures due to accelerated corrosion have been observed in few
PWR plants, mainly in combination with unfavorable coolant chemistry. To avoid such fuel
failures, advanced cladding types have and are been implemented with improved corrosion
resistance.

Dismountable Fas: Repair of failed fuel before reinsertion for further use reduces the activity
release into the coolant and is even required by authorities in some countries. Fuel vendors
developed fuel assemblies with dismountable top or/and bottom end pieces. Repair of such
fuel assemblies can be done by replacement of the failed rod by a dummy rod or a matching
uranium rod. Today, a large experience exists with handling, repair and reuse of failed
assemblies.

Mitigation of secondary failures: Formation of secondary defects results in increased

activity release into the coolant and has even caused early shutdown of some plants. Measures
to prevent the growth of primary defects and formation of secondary defects were discussed.
During the last years fuel vendors have introduced advanced cladding types with improved
resistance against secondary degradation of failure.

3
Reactors of different design: The operating experience in reactors of different generations
showed significant differences in fuel performance even for the same fuel type. For example,
reactor WWER-440 has several modifications (V-179, V-213, V-230, V-270) with different
operation features and, respectively, with different fuel failure rates.

Mixed cores: Because of ongoing FA design modification, many plants are operating with
FAs from one vendor, but of slightly different design. Moreover, FAs supplied by different
vendors may operate in one core, so-called mixed core. This practices was described in
Swedish paper for Ringhals NPP where interaction between different type FAs had an impact
on fuel performance. The effect of mixed cores have to be taken into account in evaluation
and modelling of performance of lead test FAs (they operate in a core surrounded by FAs of
“old” design).

Recommendations for future work: From the utility standpoint of view, the operational
reliability of FAs is one of the most important aspects. Regarding future efforts to improve the
fuel reliability, the participants agreed on the following:
• Grid-to-rod fretting due to fluid induced vibration is still the major PWR fuel failure
cause that requires further R&D effort, both in FA design and structural material
improvement and also in modeling.
• Possible problems related to mixed cores should be taken into account in evaluation
and modelling. In addition the performance should be ensured by lead test FAs.
• The possible negative effects of fuel failures in NPPs (e.g. losses in case of early plant
shut-down) justify further investments into the improvement of rod/FA design, fuel and
structural materials.

SESSION III: EXPERIMENTAL STUDIES OF FUEL FAILURE AND

DEGRADATION MECHANISMS
Chairmen: Yong-Soo Kim (KAERI, Republic of Korea) and A.V. Smirnov (SSC RF RIAR,
Russian Federation)

The failures of nuclear fuel rods in nuclear power plants were reported and respective failure
causes have been investigated by using Post-Irradiation Examination (PIE) techniques
(Republic of Korea, Japan, Russian Federation, Switzerland-United States of America) and
modelling (Russian Federation). The consequences of such defects are fatal. The coolant can
let into the fuel rod, the coolant flashes into steam, and then complicated processes such as
steam oxidation of UO2, oxidation and hydriding of the cladding inner surface, restructuring
of UO2. Fuel failures led to increasing activity levels in the reactor coolant system and the
reactor prematurely shut down.

According to the published data, currently the causes of PWR fuel rod damage are as follows:

• Fretting of fuel rod claddings under the spacer grids (grid-rod fretting) — 40–45 % of
cases;
• Interaction between claddings and debris in the coolant flow (debris-fretting) — 40–45
of cases;

4
 • Causes conditioned by violation of the fuel production technology (primary hydriding,
welding defects, primary cladding tube defects) — less than 5%;
• Undetermined causes and — the rest.

Similar proportions take place in the WWER reactors. The difference consists that percent of
fuel rod failure due to grid-rod fretting is significantly lower. The interaction between fuel
column and cladding (PCI) are not the cause of WWER fuel rod failure under the design-basis
conditions. The failure thresholds of high burnup WWER and PWR fuel rods by PCI-
mechanism are situated above 400 W/cm.

BWR fuel rods have increased PCI resistance and provide large flexibility in reactor operation
by allowing faster power ascension rates. Nevertheless on same cases fuel rods failed after
3 cycles of operation and an approximate burnup of 26 MWd/kg U, after a power transient
following a control blade manoeuvre.

In that way two main causes are responsible for fuel failure, namely:

• Debris-fretting because of primary coolant contamination;

• And grid-to-rod fretting because of FA vibration.

Elimination of first cause is water purification on primary coolant and development of debris
filter. But the experience of primary coolant purification shows that the amount of fuel failure
cases after purification can be increased. It is necessary to develop the purification criteria and
specifications. To remove the second cause it is necessary to develop the FA design that will
resist to vibrations.

In any case there are the tasks of identification of failed fuel rod, finding the cause and repair
of FA. These tasks might be solved by using inspection and repairing stands.
TM marks importance of modelling of behaviour of the damaged fuel. Thus, it is necessary to
develop the following models:
• Primary defect formation at residual moisture in the fuel rod as a result of
manufacturing
• Degradation of cladding properties at high burnups.

SESSION IV: MITIGATION OF FAILURES/DEGRADATION BY PLANT

OPERATION
Chairmen: J.A. Perrotta (IPEN/CNEN-SP, Brazil) and H. Hayashi (NUPEC, Japan)

The title of this session implies two items. The first one relates to the plant operational
procedures to mitigate any fuel failure, and the second one relates to plant operational
procedures to avoid degradation of leaking fuels and to mitigate fuel pellet wash-out. This
session included two papers concerning modelling of BWR cladding mechanical behaviour at
power ramps and high burnup and analysis of development of secondary failures in BWR
claddings. These papers and following discussion allow to summarize the status in the area.

Mr. Suzuki presented the analysis of the deformation behaviour of BWR fuel segment rod
during power ramp test using fuel performance code FEMAXI-6 which incorporated fission

5
gas bubble swelling and pellet-cladding bonding models. Importance of the pellet and
cladding properties, especially gas bubble swelling and creep rate of cladding, and also the
effect of bonding to generate biaxial stress mode in the cladding was shown. It was estimated
that circumferential stress at RTP would be about 300 MPa and it is sufficient for crack
penetration after formation of radially oriented crack. It was also shown that FEMAXI-6 code
is capable to provide useful information on cladding mechanical loading at high burnup.

Mr. Rudling presented an overview on operation and design strategies to minimize

degradation of failed BWR fuel. Primary fuel failure causes were presented and scenarios for
secondary failure development were discussed including circumferential cracks (breaks) and
long axial cracks. Attention was paid to the tendency of primarily failed BWR rods to degrade
depending on fuel design and reactor operation. Recommendations were given for avoiding
load follow, fast power increase rates after power decreases, flux tilting and power suppress in
order to increase resistance to the degradation of failed rod. Also, a new model “BwrFuel
Release” for analysis of fuel failures which is capable to separate activity released from tramp
uranium from that released from the defect(s). Model works well for variety of water
chemistry regimes presently used in BWRs.

Plant operational procedures to mitigate fuel failure

Plant operation with impact on fuel failure mitigation is mainly related to power ramp and
load following. PCI fuel failure is the main mechanism associated to this operational mode.
Operational factors affecting PCI fuel failure during power ramps are: burnup accumulated
prior to the ramp; maximum rod power during the ramp; power increment beyond the pre-
irradiated power level; average power ramp rate; and dwell time at high power. Procedures for
power maneuvers have been established by fuel designers and vendors.

PCI failures in PWR have not been any more a problem for many years. Due to the use of
chemical reactivity control and less inserted control rods, local power perturbation in fuel rods
is minimized. Moreover, some plants are equipped with automatic control system for power
distribution, using input data from in-core instrumentation, and they can operate in power
ramp and load following with very little restriction. For plants without this automatic control
system, more restricted procedures are used, limiting power ramp rates above a threshold
value of the reactor power. These values vary from vendor to vendor and can be as low as (for
situations with no previous power conditioning) 20 to 40% full power as threshold power, and
3% full power per hour as power ramp rate. Studies have being done and changes in power
ramp rate and threshold power have being recommended, improving current plant start-up and
load following strategies, and plant economics.

BWRs are more subjected to PCI failures, when compared to PWR, due to the fact that
control blades are used for compensation of reactivity changes with burnup increase and for
power changes. This induces more distortions in power distribution of adjacent fuel rods.
Procedures are needed to impose limitations on the power increase rate during start-up after
refueling or after control rod sequence exchange and on the control rod withdrawal speed in
the high power region when the fuel is not preconditioned. Procedures are established by
vendors and are similar in nature but dependent on the design of the control blade driving
mechanism. These procedures have being important to the mitigation of PCI failures in
BWRs, although design changes, as the introduction of barrier cladding, has significantly
reduced the onset for PCI fuel failure.

6
Research efforts have been done for both PWR and BWR fuel in order to improve knowledge
and data for high burnup fuel performance under power ramp and power cycling. Data for
burnup higher than 40 MWD/kgU for BWR and 50 MWD/kgU for PWR have been published
recently. Threshold power for the onset of vulnerability to PCI failure can be obtained from
these data. The modeling of PCI during power ramp, considering the influence of fission gas
swelling on the observed cladding strain, has to be developed (see, for example, IAEA-
TECDOC-1179, IAEA-TECDOC-1233) and taken into account for establishing plant
operational procedures.

Plant operational procedures to avoid degradation of leaking fuels and to mitigate fuel
pellet wash-out

Normally, utilities do not reinsert failed fuel in reactor, although some utilities, e.g. EDF,
allow reinsertion of leaking fuel assemblies, subject to a sipping test criterion. Fuel failure can
affect plant operation. Limits for coolant or off-gas activity during normal operation must be
within technical specification for failed fuels. Continuous operation of failed fuels can lead to
a further fuel degradation and induce release of significant amounts of radioactive products to
the coolant.

After the instant of fuel rod cladding failure, coolant enters through the breach and begins to
oxidize the fuel pellets and the inside surface of the cladding. The production of hydrogen
within the fuel rod leads to hydriding of the zircaloy and degradation of the cladding.
Oxidation of the UO2 fuel increases the fuel temperature and diffusivity of fission gases and
volatile fission products. Continuous operation may culminate in large breaches of the
cladding (transversal break or axial split) and fuel particles escaping to the coolant system.
The mechanisms responsible for cladding breach occur at secondary locations away from the
primary defect. The extent of physical deterioration can vary depending on the time the
primary defect was formed, the type and location of the primary defect, the fuel rod design,
and the operating power history. Particularly for axial split formation it can initiate from a
heavily localized hydrided region at a certain distance away from the primary defect (debris
failure pattern) or propagate from a primary defect (PCI failure pattern).

Due to differences in coolant conditions (pressure, steam fraction) and cladding material
(composition, microstructure), the degradation of failed fuel is more severe in BWR rods than
PWR rods. There have been several incidents of BWR fuel failures using high purity Zr-liner
cladding that have resulted in high off-gas plant contamination. There is a consensus that the
high purity Zr-liner contributes heavily to the failed fuel degradation.

By understanding the physical and chemical processes of fuel rod degradation, it was possible
to evaluate potential mitigating actions, which could lessen the consequences of a leaking fuel
rod (mainly concerned to an axial split formation). Design and plant operational actions were
taken. Use of alloyed Zr-liner (with Sn or Fe) to reduce the rate of corrosion and the
concomitant hydriding of the cladding during post failure operation is one of the design
actions taken for advanced fuels. One plant operational action is to decrease the fuel rod
power level and to avoid power changes in order to minimize stresses imposed on embrittled
cladding material (PCMI) required to initiate and propagate an axial split.

Concerning the plant operational procedures, there are significant differences between PWRs
and BWRs. In PWR the power reduction is done throughout the core, whereas local power
suppression through control rod insertion is not possible. On the other hand, BWR operators
have the option of inserting control blades, which gives chance for addressing operational

7
guidelines. In general these guidelines consist of: identifying the suspect leaking fuel rod
location in the core using power suppression testing; decreasing the local power level to
minimize the rate of secondary degradation; limiting the extent and rate of power changes in
the suspect location. Based on these general guidelines, fuel vendors have established
recommendations keeping some specificity for each fuel and reactor design.

EPRI also performed a comprehensive program to investigate the phenomena associated to

fuel degradation, to propose operational guidelines for mitigating the effects of failed rods
during a reactor cycle, and to develop a model to help predict the behavior of failed fuel rods.
The “Defective Fuel Element Code-T (DEFECT)” code, that is capable of modeling the
complex physical and chemical processes involved in failed fuel rod degradation, has been
developed. The code includes models for the simulation of thermal, mechanical and chemical
processes within an operating LWR fuel rod that has steam into the pellet-cladding gap due to
cladding breach. It considers the coupling of fuel thermomechanical behavior models with an
axial gas transport model for hydrogen evolution in order to determine the formation of
secondary defects and propagation of axial cladding cracks. The models were based on
industry-wide knowledge gained from laboratory tests, hot cell PIE, and fuel experience with
BWR fuel failures. The code can be used to evaluate operating strategies, which mitigate the
degradation of a fuel rod, and the effectiveness of various fuel rod design modifications
proposed to overcome the problem.

Recommendations on future work:

• Further development of models for simulation of PCI/PCMI rod failures at high burnup in
order to verify the margin’s decrease related to the actual threshold and procedures for
power ramping and load follow is needed.
• Experimenters and modelers should be encouraged to develop models and codes for
simulating failed fuel rods degradation.

SESSION V: DETECTION AND MONITORING

Chairmen: J. Schunk (NPP Paks, Hungary) and V. Petenyi (VUJE, Slovakia)

Eight papers of this session were devoted to detection and monitoring of fuel failures in
reactor cores. Detection and monitoring are essential tools for fuel failure determination
during normal or transient operation and outages. The main aim is to measure and determine
different parameters (mainly from primary coolant) in order to calculate fuel failure
characteristics and other important parameters affecting the safe and reliable operation. Based
on these results decision are made to change or reload failed fuels.

The main findings resulted from paper presentations and discussion are as follows:

1. Fuel reliability at Jaslovske Bohunice NPP:

- long term (15 years) experience on fuel failures are given,

- the root cause of higher failure rate at V-230 than at V-213 is not found, but many
possible reasons are given (dummy assemblies implementation, in-core parameter
differences technological/physical differences, etc.),

8
2. Failed rods diagnostic and primary circuit contamination levels determination thanks to
the DIADEME code:

- correlation between primary activity concentration and fuel failure parameters are
found,
- characterization of failed fuel rods (quantity of transuranium, number of failed rods,
defect size, burn-up of leaking rods, UO2 or MOX discrimination) is given,
- extrapolation of alpha-activities to the end of cycle to prepare maintenance operations
can be done,
- predictions of primary activity levels due to reload of defective fuel assemblies can be
estimated by PROFIP code.

3. Disadvantages of means and methods of fuel failure detection:

- critical evaluation of present evaluation method was given.

4. Fuel failures at Paks NPP:

- fuel performance examination by evaluation of activity concentration in primary

coolant (during normal and transient operation) was carried out together with adopting
spiking model for WWER 440 and using micro/radio analytical examinations,
- the fuel failure rate evaluation was done by expert system and steady state model.

5. Defected fuel monitoring at Cernavoda NPP:

- summary of defect investigation at limit 1 by on-line detection and location system are
given based on gaseous fission product monitoring system (Xe-133, 135, Kr-88, I-131)
and delayed neutron system (I-137, Br-87), with a good correlation between them.

6. Regulation of the fission product activity in the primary coolant and assessment of
defective fuel rod characteristics in steady-state WWER-type reactor operation:

- determination of maximum permissible level of fuel rod failure and fission product
activity concentration in primary coolant was given,
- the reliability of the assessment was increased by using the TIMS code.

7. Summary of technical development on the on-line monitoring and fuel failure evaluation
system at Temelin NPP:

- an easy-to-use on-line gamma-spectrometry system is used for collection and

calculations of complex information on fuel performance data (number and type of
defects),
- different codes (PES, PEPA) need some fine-tuning to Temelin fuel.

8. Fuel failure diagnostics in normal operation of NPPs with WWER-type reactors:

- reliability of failed fuel monitoring and detection systems was considered and found to
be appropriate,
- but failure due to manufacturing reasons reduce the reliability of failure diagnostics.

9
Conclusions and recommendations for future work:

• Detailed PIE are required and highly recommended to establish to check irradiated
fuel conditions and characterize failures, find root causes in order to avoid future
failures and reduce fuel failure rate.

• A surveillance programme should be introduced and implemented during complete

refueling outages to ensure the requested cleanliness in primary systems.

• Different codes should be harmonized in order to be able to compare results gained

with them. Further improvements are also necessary to adjust these codes to multiple
failure description and characterization.

• Results from modelling and real statistical data should be clearly differentiated.

• Fuel design and manufacturing process upgrading is necessary to minimize failures

from manufacturing reasons.

PANEL SESSION
Chairmen: D. Parrat (CEN Cadarache, France) and P. Rudling (ANT, Sweden)

1. CURRENT SITUATION OF THE UTILITIES

In many countries, the production and distribution of electric power has experienced a dual
evolution over the last decade. Firstly, the construction and commissioning of new nuclear
power plants decrease dramatically. Secondly, the liberalisation and deregulation of the
electrical market has forced the nuclear utilities to become more competitive.

In order to improve effectiveness in the field of the fuel cycle economy, operators and fuel
vendors are considering a variety of means to enhance plant performance and to reduce costs
by introducing measures such as:

• fuel burnup extension,

• power uprates,
• more aggressive loading schemes,
• decrease trends for system materials Intergranular Stress-Corrosion Cracking (IGSCC),
• decrease in activity buildup,
• longer fuel cycles,
• plant life extension.

This trend has taken place in the last several years, and is likely to continue in the long term.

As a consequence of this development, fuel isotopic composition of the irradiated fuel has
strongly evolved, with a presence of fission gases, alpha and long half-life fission products
specific activities more important at the end of life. On the other hand, larger cladding stresses
result from fuel-clad interaction, external corrosion, or grid-rod interaction. This changed
situation could impact the fuel reliability and, if necessary, the associated plant operational
surveillance or the fuel management.

10
In this context, fuel failures become an important point for the nuclear utilities, which play an
important role as a driving force for the R&D efforts related to fuel performance. They are
now facing numerous inputs, from a technical or an organizational point of view, and have to
deliver two main outputs:

• sustained safety margins,

• a preserved place on a deregulated market, involving a stronger competition.

This situation is shown on the diagram, see Fig. 1. One can identify following parameters
including inputs and outputs:

1.1. Inputs

• Regulatory bodies, which can formulate specific requests, or define new operating
technical specifications. An important point concerns the fuel performance in accidental
conditions.
• Political environment: the current situation of the nuclear industry leads to consider the
“political risk” at least as important as the technical risk (suspension of plant operation for
example if the plant is operated inappropriately).
• Economy: this parameter is now become a key-point, which presents many incentives.
• Usage of MOX fuel assemblies in some countries, which is accompanied by a specific
management and operational surveillance.
• Intermediate spent fuel storage, which necessitates on-site facilities.
• External scientific support, coming from R and D institutions, universities, independent
experts.
• The drainage of competence that has been seen over the last decade in the nuclear area.
This situation has occurred both at the utilities as well as at the fuel vendors due to that
few new young engineers have the confidence to start to work in the nuclear field while
the experts are retiring. To ensure safe and economical operation of the plants, it is crucial
that staff both at nuclear utilities and fuel vendors participate in teaching and training
classes.

1.2. Outputs

• Maintaining safety margins during normal operation and accidental conditions, by means
of:
o new fuel designs,
o new fuel management schemes,
o R and D on fuel performance in accidental situations (RIA, LOCA…) to assess
the mechanisms involved
o new codes and methodologies (including the treatment of uncertainties).*

• Maintaining competitiveness on a deregulated market by:

o continuous improvements in the fuel design and fuel operation that lead to less
fuel malfunctions -> less cost for inspection, repair or reconstitution,
o higher discharge burn-up -> lower back-end cost,
o longer cycles and shorter outages -> lower outage costs and easier staff
management,

11
 o power up-rates and more aggressive loadings (which may lead to higher
peaking factors and more local boiling in PWRs) -> improved neutron
economy,
o load follow and remote control -> better price for electricity if it is needed,
o improved chemistry : Noble Metal Chemistry (NMCA) and Hydrogen Water
Chemistry (HWC) in BWRs, and Zn injection in PWRs -> less cracking in
piping and less inspection and repair costs,
o less fuel failures and degradation, {Zn, Fe, O} dosage in BWRs, Zn dosage in
PWRs, increased LiOH content in PWRs -> lower activity build-up, with
several favourable consequences in the plant management (lower surveillance
and repair cost, lower cost for wastes disposal, better achievement of the
ALARA principle,…).

2. CURRENT ISSUES FOR FUEL FAILURES IN NORMAL OPERATION

These evolutions or improvements in the fuel or in the plant management have involved a
new repartition of fuel failures

2.1. “Non failure” situation

It is worthwhile to notice that some external causes could affect the fuel reliability, and shall
be taken into consideration. These external causes are:

• fuel handling damages,

• fuel assembly bowing,
• Axial Offset Anomalies (AOA),
• new water chemistry regimes, such as Noble Metal Chemical Addition (NMCA) or
LiOH increase.

2.2. Primary failures

The meeting put the stress on the following root causes:

2.2.1. Fretting

• due to debris in the coolant,

• due to grid-rod interaction (baffle jetting, cross flow, manufacturing defects, improper
design towards the high neutronic fluence).

2.2.2. Pellet-Cladding Interaction (PCI)

• manufacturing defects (missing pellet surface in liner or non-liner fuel, fuel chip),
• improper fuel surveillance codes.

2.2.3. Hydride assisted cracking

• a new failure mechanism was noted where a crack initiated at a massive hydride layer
at the clad outer surface propagated through the whole cladding thickness during ramp
testing. This failure mechanism may potentially limit high burnup operation
specifically in BWRs since hydriding will become more pronounced at higher burnups

12
 and simultaneously the pellet-cladding gap will become smaller. During power
ramping, e.g. by pulling a control rod, PCMI may result in fuel failures.

2.2.4. Other manufacturing defects

• primary hydriding,
• weld crack or incomplete plug welding,
• clad flaw or defect.

2.2.5. Crud Induced Local Corrosion Failures (see § 6.4)

2.3. Secondary failures and fuel degradation

The main consequence of a fuel degradation is the release of fissile material into the coolant
and an uranium core contamination. Experience feedback has shown that the natural removal
of this contamination (spent fuel assemblies discharge, cleaning of the primary circuit walls)
takes several years. Another consequence is a potential problem to extract the failed rod,
which can breaks if the clad is severely hydrided.

3. TOOLS TO MANAGE THE PRESENCE OF FUEL FAILURES

Participants agree that the annual fuel rod failure rate is the most suitable and the mostly
applied quantity to evaluate the PWR fuel reliability by the utilities. An improvement could
be to define a normalization between ”failed rods” and ”failed rods with degradation”. For
BWRs the situation is not comparable since there is a vast variation in activity release from a
failed rod. If severe degradation occurs, the plant may have to shut down just to remove one
leaking rod. For BWRs, the most relevant measure is the total off gas activity and uranium
contamination.

Some scientific progresses to do, and experience feedback or devices to use, have been
highlighted during the meeting, due to their interest to mitigate the consequences of failures
on plant operation:

3.1. R&D support

Some topics coming from the R and D support shall be investigated, in order to have in hands
a more reliable predicting of the failure evolution during the current cycle:

• better mechanistic understanding of the root cause development or of the degradation

mechanism,
• better assessment codes,
• better out-of-pile tests to verify behaviour of fuel and clad under given conditions.

3.2. On-line assessment

Determination of the failure root cause is an important point to identify and anticipate a
potential generic problem on the fuel. Several tools are commonly used in plants for that, and
some of them shall be improved or have to be more extensively used. It is notably the case for
determination of failed fuel characteristics under operation, by means of:

13
 • on-line primary gamma activities measurements,
• more accurate and reproducible sampling methods,
• more accurate assessment and diagnosis methods and codes.

3.3. Outage inspection

On the other hand, complementary information are gained due to inspection methods to
assess failures during outage:

• in-mast or in-cell qualitative sipping test,

• quantitative sipping test in some countries,
• individual inspection of failed rod, after extraction,
• ultrasonic testing and Eddy Current testing if rods are individually reachable.

3.4. Doubtful cases

For a doubtful case, or when importance or potential consequences justify it, it is useful to
send the failed rod to a Hot Cell Laboratory for complementary non-destructive and
destructive testing (see § 6.2).

4. REMEDIES TO PREVENT FAILURE OCCURRENCE

Some remedies or methodologies have been identified during the meeting to prevent failures:

• more robust fuel designs (e.g. spacers),

• plant modifications (e.g. elimination of debris, flow conversion, change to Ti-
condensers),
• better manufacturing control,
• better verification of new fuel designs : the first point is of course to verify that a
specific fuel design is suitable for the considered purpose. But a second point, which
should be as important as the first one, is to assess the limits of the new design in all
respects related to the operating conditions.

5. POTENTIAL ISSUES LINKED TO NEW FUEL DESIGNS AND NEW MATERIALS

UTILIZATION

New fuel designs or new materials utilization may potentially lead to some operating
problems:

• As mentioned in § 4, incomplete verification of margins related to new fuel designs

may cause problems in non-typical operating conditions.

• The fuel assembly integral behaviour should be verified, in order to avoid interaction
between two parameters, apparently not linked after a first analysis.
• New problems could emerge, even if they were not real problems in the old designs.

14
6. INSPECTION, RESEARCH AND DEVELOPMENT EFFORTS NEEDED TO
SOLVE POTENTIAL ISSUES

The current status of the failures in water reactors, and the trends observed in the fuel
management, lead to formulate some recommendations:

6.1. On-site inspection needs

It is highly recommended to perform on-site inspections in order to:

• identify failed fuel assemblies,

• identify number of failed rods and of defects,
• identify the root cause(s) for the defect(s),
• eliminate as soon as possible certain primary failure causes (e.g. debris),
• prevent potential generic root causes.

For these purposes, several tools are available on-site: in-mast sipping, in-cell sipping,
underwater visual inspection (assembly and individual rod), ultrasonic testing and Eddy
Current testing (if assembly design permits it).

It has to be underlined that dismountable assemblies permit individual rod extraction and
inspection, and facilitate the assembly repair.

Moreover, some participants underlined the interest to develop a non-destructive technique

for quantitative analysis of hydrogen in fuel rods.

6.2. Hot Cells post-irradiation examination needs

In some cases, complementary examination in hot cells permits a more complete assessment:

• definite failure root cause determination or confirmation,

• irradiated materials properties verification (e.g. behaviour of defective fuel or clad in
specific situations or in accidental conditions).

6.3. Technology and experience feedback transfer between different reactor types

During the meeting, it has been remarked that some problems have been solved in one type of
reactor, and that remedies could be usefully considered in another type still facing a similar
problem. Some examples have been highlighted:

• cladding liner technology developed for BWRs: interest for PWRs and WWERs,
• Zr/Nb material behaviour in WWERs experiences: interest for PWRs,
• material behaviour experiences in more PWRs demanding conditions (higher burn-up,
longer cycles, mixed cores,…): interest for WWERs.

Moreover, participants guess to exchange experience feedback on “how to” avoid primary
failure and fuel degradation (other subjects can be also discussed).

15
6.4. Mechanistic understanding

A need for a mechanistic understanding of the failed fuel rod behaviour has been also
underlined, to develop or adapt models. There are numerous cases were the fuel vendor has
made a “quick-fix” of a problem by essentially making some change in their manufacturing
process but without any mechanistic knowledge behind the change. Therefore a small change
in operation, applied water chemistry regimes or material manufacturing process will make
the problem to reoccur.

The crud induced corrosion problems both in PWRs and BWRs is a good example. These
types of failures occurred already 20 years ago and they still reoccur. The only way to resolve
this problem is to get a mechanistic understanding of the crud induced corrosion process. As
long as we have not this understanding, this type of failure will reoccur.

The following material related issues clearly needs to mechanistically understood to resolve
the issue:

• crack initiation and propagation in the cladding,

• release of fission products out of defective fuel during transients,
• internal volumes of a failed rod description.

6.5. Early integration of R&D results

Intensive R and D programmes often accompany new improvements on fuel or assembly

design. These programmes concern generally only a very specific topic (e.g. clad external
corrosion, grid-rod fretting, pellet-cladding interaction, fission gases retention in the fuel
matrix…). It is important that results gained thanks to these programmes can be early
presented to other designers (reactor water chemists, material people, fuel people), to integrate
them and avoid supplementary R&D tests.

6.6. R&D needs for experiments

New designs behaviour (fuel, clad, assembly...) shall be assessed thanks to two main types of
experiments, either out of pile, or in pile:

6.6.1. Separate effects experiments

These experiments permit to:

• enhance the database in addition to the one coming from surveillance programs (other
parameters, other operating conditions)
• verify the influence of changes in the design of in the operating conditions,
• assess the behaviour during extreme conditions, to find the operating limits of the
design (existence of a possible cliff-edge effect ?),
• anticipate possible industrial generic problems.

16
6.6.2. Global (or integral) experiments

A few experiments of this type are useful to:

• assess the performance of the global design thanks to a final verification taking into
account numerous parameters and a bundle geometry as representative as possible,
• check the whole code prediction (benchmark).

6.7. Better assessment of experiments or on-site inspection results

In some cases, non-typical fuel or rod behaviour observed through in-pile experiments or on-
site inspection, are simply classified as “anomaly” or “non-representative operating
conditions”. These cases should be analysed more attentively, because they may enable the
utility to catch an emerging potential problem.

17
18

FIG. 1. Current issues, tools, remedies and R&D needed to solve potential issues.
GENERAL OVERVIEW OF PRESENT STATUS OF EXPERIENCE
(Session 1)
EXPERIENCE AND RELIABILITY OF FRAMATOME ANP’S
PWR AND BWR FUEL

W. KLINGER*, C. PETIT**, J. WILLSE***

* Framatome ANP GmbH, Erlangen, Germany
** Framatome ANP, Lyon, France
***Framatome ANP Inc, Lynchburg, United States of America

Abstract

Based on three decades of fuel supply to 169 PWR and BWR plants on four continents, Framatome
ANP has a very large database from operating experience feedback. The performance of Framatome
PWR and BWR fuel is discussed for the period 1992–2001 with special emphasis on fuel failures,
countermeasures and their effectiveness. While PWR fuel performance in most reactors has been
good, the performance in some years did suffer from special circumstances that caused grid-to-rod
fretting failures in few PWRs. After solving this problem, fuel of all types showed high reliability
again. Especially the current PWR fuel products AFA 3G, HTP, Mark B and Mark BW showed a very
good operating performance. Fuel reliability of Framatome ANP BWR fuel has been excellent over
the last decade with average annual fuel rod failure rates under 1 × 10-5 since 1991. More than 40% of
all BWR fuel failures in the 1992-2001 decade were caused by debris fretting. The debris problem has
been remedied with the FUELGUARDTM lower tie plate, and by reactor operators’ efforts to control
the sources of debris. PCI, the main failure mechanism in former periods, affected only 10 rods. All of
these rods had non-liner cladding.

1. INTRODUCTION
The irradiation performance of the Framatome ANP nuclear fuel products during the period
1992-2001 is presented with emphasis on fuel failures and the performance of current fuel
products. Results of failed fuel examination are presented together with counter measures
taken and the efficiency of measures.

2. OVERALL IRRADIATION EXPERIENCE

By December 2001, nuclear fuel fabricated by Framatome ANP in Belgium, France,
Germany, and in the USA had been irradiated in 169 commercial power reactors on four
continents. This fuel included more than 140,000 fuel assemblies containing over 25 million
fuel rods. The maximum assembly burnups are 65 GWd/tU in a PWR and 71 GWd/tU in a
BWR.
The burnup distribution of individual fuel assemblies with burnups above 40 GWd/tU is given
in Figure 1. Although a large part of the fuel currently in core (which is scheduled to reach
higher discharge burnups than earlier fuel designs) is still operating at low burnup, an
increasing fraction of Framatome ANP fuel has already achieved burnup values beyond the
former burnup targets.

21
 6000

5000

BWR
Number of Fuel Assemblies 4000

3000
PWR

2000

1000

0
40- 42- 44- 46- 48- 50- 52- 54- 56- 58- 60- 62- 64- 66- 68- 70-
41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71

Burnup (GWd/tU)

FIG. 1. Burnup distribution of Framatome ANP fuel assemblies (as of Dec. 2001).

3. FUEL FAILURE EXPERIENCE

Failure experience is presented in terms of annual fuel rod failure rate, i.e. number of failed
rods during a year divided by the number of operating rods in this year. For many years, the
industry-wide target to achieve was a failure rate below 10-5. Although some setback did
occur recently, this target has been achieved with both, BWR and PWR fuel.

3.1. PWR fuel

Framatome ANP PWR fuel of all types had already shown high operating reliability with
mean annual fuel rod failure rates around 4 × 10-5 at the end of the 80ies and a continuing
downward trend.
This positive trend was interrupted in the mid 90ies, when fuel in some reactors suffered from
special situations that led to a number of grid-to-rod failures, mainly in 16 × 16 fuel
assemblies. Recovering from this instance, the fuel performance showed a positive trend
again and the mean failure rates were less than 1 × 10-5 in the years 1999 and 2000. This was
the best annual result achieved with Framatome ANP PWR assemblies (Fig. 2).
The results for 2001 took an unexpected turn due to a number of grid-to-rod fretting failures
as reported hereafter, which disrupted the positive trend.
Fuel examination revealed that grid-to-rod fretting, debris fretting and fabrication deficiencies
caused most failures in this period (Fig. 3).

22
 10

Rod failure rate (x10-5)

8

0
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Year

FIG. 2. PWR rod failure rates in 1992-2001.

Fuel irradiated in Germany has recovered from grid-to-rod fretting failures due to spacer
spring breakage in the mid 90ies. The cause of failure was traced back to the combination of
high stress and high stress corrosion cracking susceptibility in spacer springs, caused by
improperly heated Inconel springs and the use of this spacer at the lowermost spacer position
which is below the active length in the affected plants. This problem was solved by use of
proven Inconel spacers outside the active region.

Cases of grid-to-rod fretting occurred at the core periphery in this period. These failures were
often but not always due to baffle jetting. In a German plant for instance, grid-to-rod fretting
failures have repeatedly been observed at the same core positions at the core periphery with
fuel using conventional spring-and-dimple spacers. Since HTP assemblies have been loaded
at these core positions in 1997, such failures were not observed again.

Undetermined
Crud + handling Fretting
Manufacture spring breakage

Debris

Fretting at the
core periphery

Fretting
(other causes)

FIG. 3. PWR rod failure causes in 1992-2001.

23
Grid-to-rod fretting due to fuel rod vibration remains as a significant cause of fuel failure and
is responsible for one quarter of PWR fuel failures. Such failures were observed last year. 28
AFA 2G fuel assemblies developed leaking fuel rods during cycle 8 of Cattenom 3 (a French
nuclear plant). After the outage in early 2001, fuel examinations were performed and it was
determined that the failures were caused by spacer grid fretting wear under the lower grid.
Studies and hydraulic testing are still under way to fully understand the root cause. This grid-
to-rod fretting is partially due to Cattenom 3 high cross flow and partially to the long
residence time of some of the fuel assemblies in the most demanding positions. Cattenom 3 is
a 4 loop 14 foot core plant utilizing long (18 month) irradiation cycles with a load follow and
frequency control operation. Our design and manufacturing departments are working hard to
deliver reinforced fuel assemblies by the end of this year.
Besides grid-to-rod fretting, debris-induced fretting caused most of the failures (Fig. 4). The
TRAPPERTM, FUELGUARDTM and IDF anti-debris devices proved their filtering
efficiency with a drastic decrease in the number of failures due to debris. To date there have
been no debris failures in any PWR fuel using one of these debris retaining bottom nozzles.

FIG. 4. Anti debris devices have proven effective.

Fuel examinations revealed that primary hydriding due to contamination by hydrogenous

compounds was a frequent fabrication cause of leaks. One typical leak of this kind occurred at
the end of 1994 in a French reactor. After two weeks of operation, high fission product
activity was measured and the reactor was shut down. Ten fuel rods in two fresh assemblies
were leaking. All of the failed rods had blisters just above the bottom end plug. The
information gathered from the hot cell examinations and fabrication investigation (root cause
analysis) demonstrated that the cause of the failures was pollution of some pellets by
hydrogenous compounds. Measures have been taken in the fabrication plant to prevent
repeated occurrences of pollution.
Some fuel assemblies suffered from specific problems like fabrication (end plug welding),
and primary coolant chemistry (crud). A typical example of a failure caused by welding is
described below. In 1999, one failed fuel rod was a rod that had been repaired. The seal weld
cycle proceeded incorrectly and it was repaired by re-welding. The investigation proved that

24
the root cause of the failure was this specific repair. This seal weld repair procedure has been
eliminated.
There has been one incidence of crud related failures, which occurred in the United States. In
a 1995 refuelling, nine failures and close to one hundred degraded rods were discovered. Low
pH and high boron concentrations early in the cycle led to the formation of a thick layer of
crud on the peripheral rods of some of the high powered fuel assemblies. It is believed that
steam blankets formed into the crud layer, significantly increasing the cladding temperature
and causing accelerated corrosion. The fuel rods failed as a result of the accelerated corrosion.

3.2. BWR fuel

In the 70ies and 80ies, the rod failure rates for BWR fuel assemblies fluctuated more widely,
mainly due to instances with enhanced number of PCI (pellet clad interaction) failures. As
discussed below, this situation has changed completely.

5
Rod failure rate (x10-5)

0
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Year

FIG. 5. Annual BWR fuel rod failure rates in 1992-2001.

In the last ten years, the performance of Framatome ANP BWR fuel has been excellent.
Average annual fuel rod failure rates showed a generally decreasing tendency with values
below 2 × 10-5 since 1991, and the average for the period even dropping as low as 0.7 × 10-5.
Poolside examination revealed three failure mechanisms that caused most of the BWR fuel
failures.
Fretting by metallic debris was the leading failure mechanism in BWR fuel in this period,
causing more than 40% of all fuel rod failures. From this, a further reduction of debris fretting
failures is the best measure for a further improved BWR fuel performance. Same as in PWR
fuel, the debris problem has been remedied with the implementation of debris resistant lower
tie plates, and by reactor operators’ efforts to control the sources of debris.

25
 Undetermined
Debris

Manufacture
PCI

FIG. 6. BWR rod failure causes in 1992-2001.

PCI, the main failure mechanism in former periods, affected only 10 rods in the time period
since 1992. All of these rods had non-liner cladding. The countermeasures initiated as a
consequence of the PCI damage in 8x8 and 9x9-1 fuel (short term: administratively by
operating recommendations, medium-term: improved fuel quality and clad materials) were
generally successful. Just 3 failed rods confirm this within the last 5 years. Two major
contributing factors in this area have been the introduction of new cladding materials and
improvements in fuel pellet quality. Thanks to these advances as well as the excellent
capabilities of today's core monitoring systems, the probability of fuel rod failures being
caused by pellet-clad interaction (PCI) has been further reduced.
Few failures in 1-cycle fuel rods had to be attributed to fabrication deficiencies. Poolside
examination revealed primary hydriding due to hydrogen containing material within the rods
as the most likely cause of failure. Measures were taken in fuel fabrication to prevent further
failures or at least reduce such failures.
Based on this situation, the most effective measure to achieve a further reduction of BWR fuel
failure rates is the reduction of debris failures. The commercial implementation of debris
filters in BWR fuel started with the ATRIUM 10 design.
4. CURRENT FUEL DESIGNS
Today’s fuel performance depends on the operating behaviour of current advanced fuel
designs. Such advanced PWR fuel products are supplied as AFA 3G and HTP assemblies in
Europe and as Mark B/BW and HTP in the US. In case of BWR fuel, this is the ATRIUM 10
concept.
4.1. HTP
HTP fuel assemblies have shown an excellent operating behaviour over a 14-year period
during which altogether 3,340 HTP assemblies have been in service.
With a total of seven failed rods, five of which occurred in one plant, the annual fuel rod
failure rate for all HTP fuel remains as low as 3 × 10-6. Fuel exam revealed, that 5 failures
were caused by grid-to-rod fretting in the outermost bi-metallic spacers, which in the
meantime has been replaced for most fuel delivery batches. One rod failed due to debris, the
remaining failed assembly has not yet been inspected.

26
Most of the operating experience with HTP fuel assemblies was gained with assemblies
having a FUELGUARDTM anti debris filter. Up to now, no debris fretting failure was
observed in these assemblies.
4.2 AFA-3G
Since 1997, more than 5,000 AFA 3G fuel assemblies have been irradiated in France and in
13 PWRs in others countries. All are showing very good behaviour.
Only seven AFA 3G fuel assemblies have developed failures. Three of the failures occurred in
Germany as a result of spacer spring failure. Another two were due to a fabrication problem.
Both problems have been addressed. The two remaining failed assemblies will be inspected in
the near future. The resulting average annual fuel rod failure rate is 6 × 10-6.
All AFA 3G fuel assemblies are configured with the TRAPPER™ bottom nozzle. It consists
of a perforated plate welded to the top of a ribbed supporting structure. Its effectiveness in
stopping debris larger than 3.3 mm is 100%. Moreover, the cavity formed by the internal ribs
prevents debris from migrating to the perimeter of the nozzle. The TRAPPER™ bottom
nozzle design is based on the vast experience with the AFA 2G anti debris device. Its
effectiveness is superior to the AFA 2G anti debris device. There have been no debris failures
in the AFA 3G fuel assemblies.

FIG. 7. Trapper anti debris device.

4.3 Mark BW and Mark B

The Mark BW, which is a 17 × 17 replacement for Westinghouse designed PWRs, was first
produced on a batch basis in 1991. Since then, 2,587 fuel assemblies have been manufactured.
In the first couple of years debris was the dominant failure mechanism. After the first two
years of production and the introduction of the debris resistant lower end fitting, there have
been no debris failures. That is no debris failures in the last 555,000 fuel rods manufactured.

27
Other failure mechanisms have occurred very infrequently. The failure rate for the Mark BW
for the last five years, for all causes, has been 3 × 10-6.
The Mark B fuel assembly, a 15 × 15 fuel assembly designed for B&W reactors, has been in
production since 1971. Since then, 8,916 fuel assemblies have been manufactured. The most
common failure mechanism of the early designs was debris fretting. Since the introduction of
a debris resistant design, there has been only one debris failure in the last 670,000 fuel rods
manufactured.

50 100

45 90

Percentage of reactor cycles with

40 80
Discharge burnup (GWj/t)

35 70

30 60

fuel failure
25 50

20 40

15 30

10 20

5 10

0 0
1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
Year

FIG. 8. Increasing discharge burnup and high operating reliability: no contradiction.

4.4. ATRIUM 10
Current BWR fuel deliveries are based on the ATRIUM 10TM fuel design. BWR cladding,
which has been optimised for waterside corrosion resistance, is supplied either as Zircaloy-2
through-wall cladding or, optionally, as Fe-enhanced Zr liner cladding, which has high PCI
resistance in combination with high resistance to secondary degradation in the event of
primary failure from any cause.
So far, ATRIUM 10TM fuel assemblies have been supplied to a total of 19 plants in Europe,
Asia, and the US giving Framatome ANP operating experience with nearly 4,000 fuel
assemblies of this type and, thanks to ongoing supply contracts, the experience will continue
to be enhanced. The ATRIUM 10TM fuel assemblies showed a very good operational
performance with only six failed rods in total. Fuel examination confirmed debris fretting as
the primary cause of failure in three of these rods. For the other three rods the cause of failure
remains undetermined.

5. CONCLUSION
Feedback from the operational behaviour of more than 25 million irradiated Framatome fuel
rods have provided the knowledge to deliver fuel with high reliability even under today’s

28
demanding operating conditions. Each failed fuel rod is a setback for fuel reliability.
Framatome ANP follows a firm procedure to clarify the root cause of failure as a starting
point for the development and implementation of effective counter measures. This proceeding
resulted in improved fuel reliability in the past and will enable Framatome to reach even
better fuel performance in the coming years.

29
SPANISH EXPERIENCE WITH LWR FUEL: GENERAL OVERVIEW

J.M. CONDE LÓPEZ, M. GARCÍA LEIVA

Consejo de Seguridad Nuclear,
Madrid, Spain

Abstract

The operating experience in Spain regarding fuel failures is described in this paper. The operational
strategies followed by the NPPs, their fuel failure history, the root causes and the remedies
implemented in the form of fuel design changes, operational actions and analysis methods changes
too, are discussed. The operational strategies now followed include the operation of longer cycles,
nuclear power uprates, mixed cores, and a tendency to apply best estimate analysis methods. Changes
associated to this strategy are the increase of the fuel discharge burnup, the use of higher enrichments
and burnable absorbers, and the operation of the fuel in a more aggressive environment in general
terms. These changes have an impact on the fuel safety and reliability aspects, and have led to fuel
design changes like the utilization of advanced cladding materials, pellets, etc. All those changes have
also obligated to upgrade the thermal mechanical models. The presentation shows the fuel failure
history, including all the failures founded, both at PWR and BWR NPPs, the main root causes (crud
induced localized corrosion (CILC), debris, baffle jetting, hydriding, etc.), and the design and
operation solutions implemented by both the vendors and the NPPs (design, chemical control,
condenser tubes change, upflow conversion, etc.).

1. INTRODUCTION

The different vintages and technologies of its nuclear power plants (NPP) characterize the
Spanish nuclear park. Fuel failure history, or more strictly speaking, the phenomena
associated with fuel failure, is then related to individual plants in general, because of their
different nuclear characteristics (nuclear power, different technology, different core designs,
etc), and with the exclusion of those plants, PWR, from second and third generation and
United States (US) technology.

Generally, the second and third generation plants from US technology, have applied different
operational strategies that affect core environment, and have influence on fuel behaviour, and
fuel failure history.

The strategies followed by the different SNPPs are the extension of fuel cycles, the increase
of nuclear power, the utilization of mixed cores, (not only due to fuel design changes, but also
to new fuel designs), the increase in discharge burnup, and also the use of low leakage loading
patterns.

One of the ways the SNPPs are working involves changes in design and analysis
methodologies, oriented to a statistical approach. These “best-estimate” models also imply
uncertainty analysis.

From nine units Spain has, seven have increased their cycle length. This change affect
specially, the second and third generation SNPPs with US technology, and follow the way
open by the American ones.

30
 Table 1. Spanish Nuclear Park
Plant (number of reactors) Model
Design origin
Nuclear/Electric Power (Date of construction)
C.N. Jose Cabrera (1) a
1 loop PWR
510/160 MW Westinghouse (US)
(1968)
C. N. Sta. M. De Garoña (1) a
BWR-3
1381/460 MW General Electric (US)
(1971)
C. N. Almaraz (2) b
PWR
2696/930 MW Westinghouse (US)
(1981-1983))
C. N. Ascó (2) b
PWR
2696/930 MW Westinghouse (US)
(1983-1985)
C. N. Cofrentes (1) b
BWR-6
2952/994 MW General Electric (US)
(1984)
C. N. Vandellós 2 (1)c
PWR
2775/992 MW Westinghouse (US)
(1987)
C. N. Trillo (1) c
PWR
3010/1066 MW KWU (Germany)
(1988)
a
1st Generation ; b
2nd Generation; c 3rd Generation

Table 2. Current cycle lengths and uprating process of different SNPP

1st 2nd Next
Cycle length 3rd uprate Current
NPP uprate uprate future
(months) % status
% % %
GAROÑA 24 (a) - - - 100 % -

JOSE CABRERA 12 - - - 100 % -

COFRENTES 18(a) +2 +2.2 (1998) +5.8% 110 % -

(1988) (2002)

ASCO 1 18(a) +8 108 % +1.5

(2000) (2003)

ASCO 2 18(a) +8 108% +1.5

(1999) (2002)

ALMARAZ 1 18(a) - - - 100 % +1.4

(2003)

ALMARAZ 2 18(a) - - - 100% +1.4

(2004)

VANDELLOS 18(a) +4.5 +1.5 (2002) - 106% -

(1999)

TRILLO 12 - - - 100 % -

(a) Change of cycle length.

31
Concerning the uprating strategy, the SNPPs have followed the international trend to increase
their nuclear power. We can distinguish between two kinds of uprates in Spain until the
current situation. Those which implies strong charge of licensing and analysis, that are usually
greater than 2%, and those less than 2%, called mini-uprates, that are usually related to the
improvement of flow measurement on the feedwater system.

The evaluation of the uprating, by the CSN, is different depending on the type of uprate we
are dealing with. Anyway, the SNPPs have followed an uprating process in different stages
(see Table 2).

After the deregulation of the electricity markets in Europe, the Spanish NPPs are intended to
reduce operating costs. One of the strategies followed by the NPPs is to have different fuel
suppliers.

Moreover, the operational strategies have implied fuel design changes. The SNPP have mixed
cores (see Table 3), with fuels of a same supplier but different designs, or with fuel form
different suppliers too.

The safety limits are associated to the fuel design. Thus, different fuels have different safety
limits. If the utilities do not change the supplier, the various analyses are internally coherent;
as long as the supplier design and monitoring methods are approved, no additional action is
needed.

Table 3. Mixed cores

Licensed Maximum Burnup
DESIG FUEL FUEL DEM
NPP Enrichme enrichment Limit
N VENDORS TYPE O
nt reload MWd/Kg
COFRENTES GE GENUSA GE 11 No Fuel type 4.5% f.t.d.
(BWR-6) GE-12 dependent
ABB ATOM SVEA 96+ (f.t.d.)

GAROÑA GE GENUSA GE-11 No Fuel type - f.t.d.

(BWR-3) GE14 dependent
(f.t.d.)
VANDELLOS W ENUSA W-MAEF No 4.90 % 4.7% 60 [1]
(PWR-3 loops) W-AEF
W-OFA

ASCO 1 & 2 W ENUSA W-MAEF No 4.90 4.7% 60 [1]

(PWR-3 loops) W-AEF

ALMARAZ W ENUSA W-MAEF, AEF Yes 4.5 4.5% 60 [1]

1&2 [Framatome] [AFA (demos)]
(PWR-3 loops)

TRILLO KWU KWU AH76 Yes 4.3% 4.5%

(PWR-3 loops) [ENUSA] AH116 AH196
FOCUS AH216
FOCUS AH266
ENUSA (Demos)

JOSE W ENUSA 14x14 HIPAR No 4.0 % - 60 [1]

CABRERA 14X14
(PWR-1 loop)) LOLOPAR

[1] average burnup per fuel pin rod (45-48 MWd/Kg U per fuel assembly: f.t.d.)

32
However, if more than one fuel vendor is involved, the utility must take appropriate action to
ensure that the different methods and correlations do not carry over any inconsistencies or
mismatches.

On the other hand, we can also see in table 3 that the enrichment is in some cases near the 5%
in weight, and that the discharge burnup is near the licensing limit. The value of this limit is
62 GWd/Tn U.

The continuous increments of enrichment in PWR make it necessary to use burnable poisons
at "beginning of cycle" (BOC). Initially, this need was realized by the chemical control of
coolant, but now control is taken by the use of WABAS or pellets with burnable poisons
mixed with fuel. This strategy has had some impact on the analytical methods used for fuel
rod performance analysis.

The methods and models used for fuel rod performance analysis, and used as input for
accident analysis, have suffered various changes and have become (or evolved to) statistical
or best estimate cases. This is due to the better understanding of the behaviour of new
materials, new designs, and the data compiled in post Irradiation examinations (PIE).

The Spanish NPPs have been introducing core designs with rods containing burnable poison
of varying (different) contents in weight. This technique allows to improve the Chemical
coolant programme and reduces the risk of fuel failure (The SNPP used a modified coolant
chemistry programme).

However, the use of burnable pellets implies a penalty over the enrichment, and this affects
the core design too.

The use of burnable poisons mixed into the fuel pellet reduces the conductivity. The penalty
associated to this fact implies problems with the core design. The use of a new conductivity
correlation for Gd pellets with a concentration up to 2%, as a best-estimate model, eliminates
the penalty on the enrichment, and permits more elements containing low concentration of
burnable poison. This allows more flexible core designs.

In the table 4, we can appreciate the near future plans. We can see that all the strategies
discussed before are not finished yet. Mini upratings, new design elements, higher
enrichment, higher discharge burnup, etc.

From a regulatory point of view, these strategies have an effect on safety limits. The use of
mixed cores, the higher enrichment, and the more and more aggressive operational conditions,
combined with the use of statistical or best estimate methods, all affect the traditional
approach to fuel limits. Variables like DNBR or CPR, shutdown margin, reactivity
coefficients, must be taken into account as a whole, because their behaviour has a synergistic
interaction.

The uprating, for example, increases the linear heat generation rate (LHGR). This affects the
following safety related criteria: the DNBR/CPR, the reactivity coefficient, the shutdown
margin, the internal gas pressure, PCI, amongst others in normal operation as well as
anticipated transients or postulated accidents.

33
Table 4. Near future plans
Higher Other
New Higher New analysis New management
NPP discharge aspects
fuels enrichment methodologies strategies
burnup
COFRENTES
(BWR-6) X X X X X X

GAROÑA
(BWR-3) - - - X - -

VANDELLOS
(PWR-3 loops) - X - - - X

ASCO 1 & 2
(PWR-3 loops) - X - - X X

ALMARAZ 1 & 2
(PWR-3 loops) - X X - X X

TRILLO
(PWR-3 loops) - X X X - X

JOSE CABRERA
(PWR-I loop)) - - - - - -

Table 5: PWR failure history during the last 10 years

PRIMARY UNKNOWN/
DEBRIS FRETTING
HYDRIDING TOTAL
YEAR NON INSPECTED
E KWU E KWU E KWU E KWU E KWU ALL
93 2 0 0 0 0 0 0/3 0/1 5 1 6
94 3 0 0 0 0 0 1/4 0/0 8 0 8
95 0 0 0 0 0 0 0/6 0/0 6 0 6
96 0 1 0 0 0 0 0/3 0/0 3 1 4
97 9 0 0 0 0 4 0/3 0/0 12 4 16
98 0 0 0 0 0 1 1/3 0/0 4 1 5
99 1 0 1 0 0 0 1/0 0/0 3 0 3
00 6 0 0 0 0 0 0/0 0/0 6 0 6
01 0 0 0 0 0 0/1 0/0 1 0 1
02 2 0 0 0 0 0 0/0 0/0 2 0 2

2. FUEL HISTORY: CAUSES AND ACTIONS

Let’s look now the fuel failure history of PWR. We can see in this transparency that I have
made a distinction between American design NPPs and German ones, because of the
differences between their fuel element designs. We have to bear (or keep) in mind that there
are six American design NPPs and only one of German design in Spain.

The table 5 shows data over the last ten years. If we’d add the data of fuel failure from the
years before 1992, we could observe a trend of the fuel failure. Before seeing the evolution
graph, let’s consider this table in more detail. The main cause of fuel failure is the existence
of Debris in coolant. The second, due to Primary hydriding, has affected only the German
design plant.

34
 PWR - ENUSA

DEBRIS
50
40 FRETTING
30
993 20
1 UNKNOWN
96 10
19 0 NON
99 INSPECTED
19 TO
UN TA TOTAL
02 DE KN L
20 BR O UNK/N.I.
IS W
N Accum.

FIG. 1. Fuel failure statistics-causes for PWR-ENUSA plant.

PWR-KWU

1993 4
3
1996 DEBRIS
2
1999 1 PRYMARY
2002 0 HYDRIDING
NOT
PR IS

NO ARY

TO
DE

YM

TA
BR

T
IN HYD

INSPECTED
L
SP
EC

TOTAL
TE ING
RI

D
D

FIG. 2. Fuel failure statistics-causes for PWR-KWU plant.

35
Analysing the fuel failure graph, we can see a reduction in the frequency of fuel failure over
the last four years. If we see the evolution per failure cause in Fig. 1, we observe a reduction
on DEBRIS, hydriding and fretting causes, but we observe too the existence of not inspected
and unknown failed elements, so we cannot conclude that the mentioned causes have been
controlled.

For the KWU PWR plant, see Fig. 2, the primary hydriding has been considered the main
cause on fuel failure, but the remedies seem to be sufficient to eliminate this cause.

• Concerning the main failure causes, as I mentioned before, the appearance of DEBRIS
into the coolant has produced fuel failure by fretting, generally underneath the first spacer.
The corrective actions taken, were to recuperate the elements by replacing the failed rods
with stainless steel ones, and to make improvements on the plant maintenance activities to
reduce the probability of appearance of DEBRIS and the change of fuel design in order to
incorporate a DEBRIS filter in the bottom plate.

It should be mentioned that the maximum value of failed elements, if we split these values
by each plant, has happened after major modifications on plant systems or actions on the
primary circuit. In any case, the evolution of this failure cause indicates to us that the
problem doesn’t seem to have been solved.

In the past, when a susceptibility cause has been encountered, safety studies have been
carried out to prevent fuel damage due to DEBRIS (for example, the fracture of top nozzle
spring screws)

• The fuel induced vibrations fretting is identified like the cause of failed elements in the
SNPPs. We can distinguish various mechanisms which can be considered in this set of
failures, like manufacturing related fretting, or baffle jetting,, operational induced fretting,
or vortex induced fretting. The failed element due to this cause is identified like
manufacturing related cause.

Attending to the vortex induced fretting, the existence of turbulence and flow
inhomogeneities at the core bottom lead to the existence of fretting modes between spacer
grids and rods, and finally to rod damage. This failure cause was discovered in the United
States, and it isn’t a cause of failure in Spain. The root cause analysis concluded that two
facts affect this: the specific design of spacer grids, and the specific flow the elements are
submitted to.

However, the preventive actions taken by the SNPP were to use burnable poison type
WABA on pheripherical configuration, to reduce the resonance frequency and eliminate
breaking risks, and a modification on the assembly process into the element fabrication.
This problem has been corrected by new grid design.

• The failure by baffle jetting cause affected only one plant in Spain. The rod vibration by
cross flow, due to the differential pressure existing between the ascending flow of coolant
through the core, and the descending flow providing internal cooling for the baffles, was
corrected:

Firstly by the substitution of both failed and damaged rods by stainless steel ones,
including the introduction of fresh fuel with stainless steel rods in the susceptible
positions, and the performance of mechanical peening on the baffle joints to reduce the

36
 gap size between segments and thus reduce the effective flow which can cause rod
vibration.

And finally by the modification on flux sense from down-flow to up-flow in order to
reduce the differential pressure with the ascending flow of coolant through the core.

• Hydriding was observed in one plant. The root cause analysis confirms the cause of this
failure to be the existence of contamination during the fabrication process. The
improvements on the manufacturing process, seem to be sufficient to avoid this failure
cause.

• Although they aren't the direct cause of fuel failure, there are other problems which could
be a problem for safety related limits. Apart from mechanical problems, like rod bowing
or rod growth, the corrosion is, possibly, the only one that has an implication on the
different safety criteria.

Some inspections carried out in 1997 at a PWR plant observed the existence of a corrosion
layer above the design limit This fact forced the NPP to select other material for the
cladding. The original material was improved Zr-4, and the new material selected by the
PWR-NPPs is ZIRLO.

As a consequence of this inspection, the various NPPs at risk of the same behaviour
developed a common inspection program PIC, together with the fuel supplier, to evaluate
the new aspects of new design fuel.

Now, we’ll move on BWR fuel failure history. In the same way than PWR treatment, we can
see in table 6, the main causes of fuel failure over the last ten years.

This aside, the number of fuel failures, is low and seems to be stable As you can see, I have
include the CILC phenomena into the table, although there hasn’t been any failed elements
over these years, because this failure cause was one of the main fuel failure cause in the past.

Table 6. BWR failure history during the last 10 years

Unknown /
YEAR CILC PCI Fabrication Hydriding TOTAL
Not Inspected
88-91 21 1 2 1/2 0 27
92 0 0 1 0/0 0 0
93 0 0 0 2/0 0 1
94 0 0 0 0/0 0 2
95 0 0 0 0/0 0 0
96 0 0 0 0/0 0 0
97 0 0 0 1/0 0 0
98 0 0 0 0/0 0 1
99 0 0 0 0/2 0 2
00 0 0 0 0/1 0 0
01 0 0 0 0/0 0 1
02 0 0 0 0/3 0 3

37
 LAST 10 YEAR BWR FUEL FAILURE
HISTORY / 100,000 rods BWR

14

12

10

0
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

FIG. 3. BWR fuel failure history.

We can appreciate the stability of these values (see Fig. 3). Again, the inspection policy of the
NPPs, prevent us to understanding the main cause of the last fuel failures, especially when the
fuel reached its discharge burnup.

• Although Crud Induced Localized Corrosion (CILC) seems to have been solved, it was
the main fuel failure cause in the past in one plant. The actions taken by the NPP to solve
it were oriented to prevent the high concentrations of Copper found in primary coolant
being the main reason obtained from the root cause analysis.

The action taken was to change demineralisation process with the use of one more filter,
and the physical modification in order to augment the filter surface. With this procedure,
CILC, as a failure cause, seemed to be solved.

However, recently, the NPP has implemented a design modification, in order to change
the material of condenser tubes from Copper to Titanium. With this definitive solution,
the origin of Copper in the coolant system disappears.

• Over the last ten years, the fabrication cause has been the most important cause of fuel
failure. The solutions have been, in general, to improve the manufacturing and the
inspection processes (UT, ET,. VT, etc.)

As we can see on the table, the last failed elements haven’t been inspected, and we put
them on the unknown column, so we can’t guarantee that the different failures have been
solved.

• Other causes, like PCI or Hydriding were considered as one time events. The operational
strategies (PCIOMR), and the introduction of barriers in the inner side of cladding seem to
be sufficient to prevent them.

38
• As regards recent indications of shadow corrosion detected under the spacer, the studies
performed seem to conclude the level is stable under normal operational condition.
Nevertheless, there is an ongoing surveillance programme for GE-12 and GE-14 cladding
material.

Concerning the operation of the plant, the strategies to operate with failed elements are in
accordance with the international ones.

Thus, control is ensured through the follow up of coolant or off-gas activities. The Iodine and
Xe activity level measurements serve to find or confirm the existence, type and size of fuel
failure. On the other hand, the relation between Caesium isotopes serves to predict the burnup
level, and then, where the probable area is, so the failed element can be located.

Apart from the limits established in Technical specifications, no other fuel safety criteria on
coolant activity exist.

In the case of BWR plants, there is the possibility of screening the failed element, in order to
continue with the operation and prevent the risks of secondary degradations.

In terms of fuel failure history, in conclusion, we can see that fuel failure rate of the SNPPs
has been maintained over the last ten years in PWR and BWR, but the level is lower than in
previous years.

The main failure cause in PWR NPPs is DEBRIS (Westinghouse plants), and the high values
encountered coincided with major design modification or activities on the primary circuit, and
PRIMARY HYDRIDING (KWU plant). These problems appear to have been solved, but we
don’t have the information on the last failures in the Westinghouse plants. This is due to the
NPPs inspection policy on fuel elements with high burnup.

On the other hand, for BWR plants, the main causes were CILC and fabrication. The last year
data seem to demonstrate that both of them have been solved, specially CILC problem.

3. CONCERNS AND NEEDS

With the advent of advance fuel and core design, the adoption of more aggressive fuel
management modes and the implementation of more accurate (statistical or best estimate)
design and analysis methods, the SNPPs have increased the level of reliability, and more
aggressive conditions on the fuel, Fig. 4. Both of them affect the safety limits. The concerns
about the normal operation problems seem to be solved, but the aggressive conditions, rise us
doubts about the fuel behaviour under RIA and LOCA conditions.

Spain participates in most of the fuel research international programs, like HALDEN, ALPS,
CABRI, Robust Fuel Programme, which intend to demonstrate the availability and safety of
high burnup fuel.

These investigation programmes aren’t focused on normal operation, and I won’t discuss
them. However, SNPPs and ENUSA as a fuel supplier, are developing a fuel inspection
programme (PIC), devoted to the behaviour of the fuel under normal operation.

39
 NEW FUEL Design NEW STRATEGIES

More Reliability Aggressive conditions

SAFETY LIMITS

DOUBTS ABOUT THE BEHAVIOUR UNDER RIA AND LOCA

EXPERIMENTAL INSPECTION
PROGRAMMES PROGRAMMES

FIG. 4. Schematics of the fuel concerns, needs and programmes.

The fuel inspection programme was carried out because of high corrosion levels encountered
during inspection of elements with Zr-4 cladding material in PWR plants. The values obtained
were higher than the design limits. This, combined with the design changes due to rod
bowing, induced the NPPs to develop an inspection programme to corroborate the data used
to license the fuel design. This programme covers different operational conditions.
The programme is currently ongoing, and the first results, over fuel elements with hard
operational conditions show a better behaviour of ZIRLO versus Zircaloy-4 for corrosion
growth. These results also show the dependence on the operational conditions.

4. CONCLUSIONS.
An overview about the different operational strategies carried out in Spain, and the fuel failure
history for both PWR and BWR NPPs, and the ways followed to eliminate or control them.
We can conclude that all the operational strategies carried out imply more aggressive
environmental conditions on the fuel elements.
• Fuel failure rate has exhibited a declining trend over the years in Spain, however some
occurrence spikes were observed during last decade. On the other hand, existing fuel
inspection policy does not guarantee investigation of all fuel failure events, in particular
when failures affect to relatively highly burnt assemblies. So, a continued fuel failure root
cause analysis programme is needed for in the future.
• The high burnup fuel behaviour is not well understood, so there is a need to continue with
the experimental programmes.
• The data obtained from the fuel inspection programme reveal a dependence on operational
conditions. So the inspection programmes to follow the fuel behaviour in different
operational conditions have to be maintained.

40
PROBABILISTIC-STATISTICAL ANALYSIS OF WWER FUEL ELEMENT
LEAKING CAUSES AND COMPARATIVE ANALYSIS OF THE FUEL
RELIABILITY INDICATOR ON NPPs WITH WWER AND PWR REACTORS

I. CHESTAKOV
All-Russian Research Institute of Nuclear Power Plants (VNIIAES),
Moscow, Russian Federation

Abstract

The results of a comparative analysis of the fuel reliability indicator on NPPs with WWER and PWR
reactors are presented in this report. As an exponent for a comparative reliability analysis of the
WANO fuel reliability indicator is used. The fuel reliability indicator provides a general measure of
the extent to which the reactor coolant activity is increased as a result of fuel damage. The analysis of
fuel reliability indicator values during 1991-2001 at NPPs with WWER-1000 and WWER-440
reactors (Russia, Ukraine, Bulgaria, Czech Republic, Slovakia, Hungary, Finland) is carried out. The
high level of WWER fuel reliability is scored except for cases of fuel failures in separate operating
periods of some units.

1. INTRODUCTION

The results of probabilistic-statistical methods of fuel rod (FR) leaking causes analysis
application are presented in the report. A probabilistic-statistical method of nuclear FR
leaking causes analysis was developed by VNIIAES. These methods were originally used for
the analysis of RBMK FR leaking causes. The methods allowed identifying and classifying
FR leaking causes for channel-type reactors. The causes identified are design-, fabrication-
and operation-related ones [1].

Over the past years, this approach has been effectively used for the analysis of WWER FR
leaking events [2]. Using this method is particularly effective for the analysis of fuel element
leaking causes when using the expensive examination of fuel assemblies (FAs) in shielded
"hot" cells is not possible. Due to this method, it was possible to find and to explain FR
leaking causes due to the excursions of water chemistry regime and appearance of slime,
resins and dispersed corrosion product particles in coolant. The method was successively used
in analysis of occurrences happened in Unit-2 of Kola NPP in 1991-1992 [1,3,4,5], in Loviisa
NPP in 1994-1995, in Paks NPP in 1997, in Unit-4 of Novovoronez NPP in 1998.

Due to the use of probabilistic-statistical analysis of fuel element leaking causes, the specific
mechanism of FR leaking was analyzed and described. This mechanism is dealt with mass
transfer in the coolant turbulent flow and deposition of the dispersed particles (particles with
specific inert and gravitational characteristics) on FR surface. Such particles might be
corrosion products of the primary circuit, particles resulted from detraction and destruction of
foreign objects in the primary circuit (a tear away mechanical details, welding "hail",
accidentally forgotten wooden and organic details during the repair works, filter resin
disintegration particles), as well as particles forming during chemical cleaning and not
removed from coolant. Alongside with local pit corrosion, the largest dispersed particles can
result in fuel rod cladding debris-type damage.

41
2. INPUT DATA FOR STATISTICAL ANALYSIS

Information not only on FAs with leaking FRs, but also on all similar FAs in a core (which
were fabricated and operated at the same time as leaky FAs) is used as input data. Input data
for the analysis are as following:

• Results of cladding tightness inspection in special canister (stand test in the cooling
pool) for discharged FAs;
• Data on the isotope composition of the primary coolant during the cycle before fuel
discharge;
• Data on the location of all analyzed FAs in the core for the whole time of their
operation, including all reshuffles and coolant flow restrictions;
• Data on power density distribution and fuel burnup in the core for the whole time of
operation for all analyzed FAs;
• Data on operation of similar, with regard to the date and technology of fabrication, FAs
at other units of the same NPP and other NPPs with similar reactors;
• Information on the water chemistry regimes for the whole time of operation of all
analyzed FAs, including detailed information on water chemistry during the first cycle;
• Information on reactor main parameters for the whole time of FA operation in the core
and parameters specific to the analyzed assembly (in comparison with specifications), as
well as pressure drop in the reactor, in main circulation pumps and steam generators,
outlet coolant temperature on output from FA during the whole time of operation;
• Passport data on all analyzed FAs (serial numbers, date of fabrication, enrichment and
etc).

3. ESSENCE OF PROBABILISTIC-STATISTICAL ANALYSIS METHOD OF FUEL

ROD LEAKING CAUSES

The method is based on using probabilistic-statistical system structural-dynamic approach to

the FR leaking causes analysis for WWER type reactors. Earlier this approach was used for
the analysis of RBMK FR leaking causes and demonstrated its high efficiency.

The method takes into account that fuel element depressurization process is multistage and
includes the following stages:

• Appearance of the FR cladding defect;

• Propagation of the defect till the opposite cladding surface;
• Transformation of the initial microdefect into the macrodefect;
• Appearance and growth of the secondary wall-through defects;
• Quick growth of the secondary defects till cladding failure.

Herewith, each stage of the FR leaking process is governed by specific physical mechanism
and, accordingly, has own prevailing factors. New causes-factors, stipulating in the whole FA
failure, are added on each new stage a cladding leaking. Root causes of a FR leaking define
processes of generation and wall-through defect propagation. This is why several functions
characterizing different stages of cladding defect growth are used for the analysis of fuel
element leaking causes, including:

• Probability of FA operation without failure, intensities of failures and flows of FA failures

on leaking functions (4 and 5 stages);

42
• Probability of FA operation without fuel element leakage, intensities of finding FAs with
leaking FRs and flows of finding FAs with leaking FRs (2 and 3 stage);
• Normalized steady-state primary coolant specific activities of iodine radionuclides
corresponding to different types of cladding defects (gas leaking - 2-3 stage, direct contact
of fuel and coolant - 4-5 stage);
• Probabilities of FR leaking in loaded FAs by cycles (1 stage).

Systematic approach is used in this method, including analysis of the following systems:

• FA with its structural components, which changes define the design-related causes of
FA damage;
• A technological system defining initial properties of as-fabricated FA and associated
failure causes;
• A reactor, its elements (including FAs) and subsystems (including system of interaction of
personnel and reactor influencing upon operational conditions) defining operation causes
of FR leaking.

Interaction of these tree above-mentioned systems defines FA in-core performance.

At the analysis of system influence on the fuel element leaking processes, the impact of
structural relatively stable factors (defined by the stability of elements and subsystems), and
dynamic variable-factors (defined by the links between elements and subsystems) is analyzed.

Parameters defining FR leaking causes are determined by investigating correlations between

response functions (effects) with structural relatively stable and dynamically changeable
variable parameters including initial properties and FR operational conditions.

For revealing of the factors corresponding to direct and root causes of fuel element leaking is
produce collation of parameters, defining causes of fuel element leaking, with the kit of
parameters, defining known mechanisms of fuel element leaking:

(a) In accordance with design:

• Creep down of FR claddings and densification of fuel pellets;

• Fretting-corrosion of FR;
• Growth and bow of fuel rods in fuel assemblies;

(b) In accordance with fabrication:

• Internal local hydriding;

• Improper enrichment and rod filling with pellets;
• Welding defects;

(c) In accordance with operation:

• Waterside corrosion of fuel element claddings;

• Waterside corrosion of claddings accelerated by a corrosion product crud on fuel
elements;
• Debris-fretting;

43
• A mechanical interaction of fuel with the cladding aggravated by the inner face cladding
interaction with fuel fission aggressive products.

Mechanisms of fuel element leaking dealt with operation conditions are defined by all variety
of factors, e.g. operation parameters, fuel element fabrication parameters, and design features.
Analysis allows to define prevailing factors.

4. LEAKING MECHANISM DEALT WITH THE PRIMARY CIRCUIT WATER

CHEMISTRY

Very specific fuel failure mechanism was revealed and described thanks to the use of the
probabilistic-statistical analysis of fuel element leaking causes. This mechanism is connected
with mass transfer in the coolant turbulent flow and deposition of the dispersed particles on
fuel elements. Such particles might be corrosion products of the primary circuit, particles
resulted from detraction and destruction of foreign objects in the primary circuit (a tear away
mechanical details, welding "hail", accidentally forgotten wooden and organic details during
the repair works, filter resin disintegration particles), as well as particles forming during
chemical cleaning and not removed from coolant. In WWER reactors dispersed corrosion
product may appear in the core as result of crud (on steam generator side) washout due to the
pH change.

Dispersed particles due to their gravitation and inert characteristics may be redistributed in a
primary coolant flow depending on its parameters. This redistribution is particularly essential
at the entering to the reactor vessel, on the bottom of the vessel and in inlet to the fuel
assemblies, when horizontal and vertical forming coolant velocities might be significantly
changed. This is why a concentration of dispersed particles inside FAs depends on their
location in a core.

Calculations have shown that, due to the specifics of WWER core design, ingress of such
particles into FAs situated in the core center and periphery is of low probability. At the same
time, these particles are concentrated in the ring between core center and periphery. Analysis
of the location of leaking FAs in a core during the first cycle has confirmed this point of view.
The ring may be a complete one, if there is a global violation of water chemistry regime, or
only a part of it if particles enter into a core only from the specific steam generators or main
circulation pumps. (See example on Fig.1).

The crud deposition theory tells that dispersed particles are to deposit on fuel element
cladding surfaces in the areas with the increased coolant flow turbulization, namely in FA
inlet and in the area of spacer grids. For depositing on cladding surface, particles should
overpass wall buffer and laminar frontier coolant layers. In areas with the increased flow
turbulence, thickness of frontier layers is below, and additional transverse forming flow
turbulence that defining transverse velocity of dispersed particles is higher.

As far as dispersed particles have dimension of about 10-30 microns, when precipitating on a
develop fuel element surfaces they cause some cladding raggedness. As a result of significant
increase of a hydraulic resistance in the above-mentioned parts of fuel assemblies. In turn, this
may result in redistribution of coolant flows in a core.

Dispersed particles are absorbed, to a significant extent, on surfaces of freshly loaded fuel
elements. This is due to the fact that defensive oxide film on cladding surface has not yet
reached its equilibrium thickness. Deposition of such particles may result in local cladding

44
corrosion. This is bound as with reduction of defensive characteristics an oxide film, so and
with formation of the porous crud resulted in so-called "wick" effect. This effect consists of
concentration of aggressive water admixtures (chloride ions, alkali, etc.) in pores. This is why
special attention should be paid to the maintaining a good water chemistry regime mode at
start-up periods after loading new fuel assemblies in a reactor core. The biggest dispersed
particles may cause debris-type failure.

FIG. 1. WWER-440 core cartogram for the 21st fuel cycle.

45
It is noteworthy that duration of corrosion process till formation of through-wall holes varies
usually from one to three and more years depending on amount of dispersed particles in
water. This is why the root causes of fuel element leaking and leakage itself, with release of
fission products into coolant, are separated in time. At the same time, debris-fretting failure
develops usually during one fuel cycle.

5. THE FUEL RELIABILITY INDICATOR FOR NPPS WITH WWER, PWR AND
PHWR REACTORS

WANO fuel reliability indicator is used for comparison of fuel reliability at different
reactors/units. The purpose of the fuel reliability indicator is to monitor industry progress in
achieving and maintaining high fuel integrity, and to foster a healthy respect for preservation
of fuel integrity. Failed fuel represents a breach in the initial barrier preventing off-site release
of fission products.

The fuel reliability indicator (FRI) is inferred from fission product activities present in the
reactor coolant. For PWRs, PHWRs and WWERs, the indicator is defined as the steady-state
primary coolant iodine-131 activity (Bq/g), corrected for the tramp uranium contribution and
power level, and normalized to a common purification rate.

Steady state is defined as continuous operation for at least three days prior to data collection
for at a power level that does not vary more than ±5 percent. In order to obtain an indicator
value for a month, the steady state power at which data is collected must be 85 percent or
greater. This ensures appropriate indicator accuracy. For months where no period of steady
state power was 85 percent or greater, the highest steady state power achieved should be
reported. Tramp contribution is caused by fissionable material that has been deposited on
reactor core internals from previous defective fuel elements or is present on the surface of fuel
elements from the manufacturing process. The fuel reliability indicator for periods longer than
a month is determined as the average of the most recent operating quarter monthly values.

The monthly value of the PWR, WWER and PHWR indicator is calculated as the following
equations:

FRIP = [(A131)N – k (A134)N] * [(Ln/LHGR) * (100 / Po)]1.5 (1)

131
Where: (A131)N is the average steady-state activity of I in the coolant normalized to a
common purification rate and expressed in Bq/g;

k is the tramp correction coefficient (a constant with a value of 0.0318). This coefficient is
based on a tramp material composition of 30 percent uranium and 70 percent plutonium;
134
(A134)N is the average steady-state activity of I in the coolant normalized to a common
purification rate and expressed in Bq/g;

Ln is the linear heat generation rate used as basis for normalization (18.0 kilowatts per
meter);
LHGR is the average linear heat generation rate at 100 percent power (kilowatts per meter)
for the unit;

46
Po is the average reactor power (percent) at the time activities are measured.

If a calculated monthly indicator value for a unit is less than 3.7*10-2 Bq/g, it is replaced by the
value 3.7*10-2 Bq/g.

The average steady-state activity of 131I and 134

I in the coolant normalized to a common
purification rate is calculated as follows:

AN(i)= AM(i) * (Oi + Ba) / (Oi + Bn) (2)

Where: AN(i) is the average steady-state isotopic activity of “i” nuclide in the coolant
normalized to a common purification rate and expressed in Bq/g;

AM(i) is the average measured isotopic activity of “i” nuclide in the coolant (Bq/g);

Oi is the decay constant of the “i” nuclide (seconds-1);

Bn is the purification rate constant equal 2*10-5 seconds-1 and taken assumed for unity of
normalization;

Ba is the actual purification rate constant (seconds-1) defined below:

a Gi 1 (3)
Ba = ¦ • (1  )
i =1 M Ki

Gi is the letdown flow rate (kg/sec) of the “i” system filters corrected to normal reactor
coolant system operating temperature

M is the reactor coolant mass at normal operating temperature (excluding the pressurizer) and
expressed in kg;

Ki = Ai / Aoi is the clearing coefficient for iodine nuclides in “i” system filters

Ai is the iodine radionuclide specific activity upstream of the “i” system filters in
Becquerels/gram;

Aoi is the iodine radionuclide specific activity downstream of the “i” system filters in Bq/g.

The results of WANO fuel reliability indicator (FRI) calculation of arithmetic mean values on
the indicated procedure for units with WWER-440, WWER-1000, PWR and PHWR reactors
are represented in Figure 2. These results are grounded on the information of the WANO
Atlanta Center database. It is necessary to mark, that in the given report the results of
calculation of arithmetic mean values of fuel reliability indicator, instead of median, used
WANO are represented.

47
 1000

FRI, Becquerels per gram

900
800
700
600
500
400
300
200
100
0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

VVER-440 VVER-1000 PWR

VVER & PWR PHWR One defects

FIG. 2. Average values of the WANO fuel reliability indicator at NPPs with WWER, PWR and
PHWR reactors.

For units with WWER-440, PWR and PHWR reactors the fuel reliability augmentation
tendency on an FRI in 1991-2001 is scored. For units with WWER-1000 reactors, the FRI
value in 1991-2001 is a little bit worse. At the same time, for all numbered types of water-
cooled reactors the considerable differences of FRI average value per miscellaneous years are
scored.

A reactor core containing one or more fuel rod defects is likely to produce indicator values
(under steady-state conditions) greater than 19 Bq/g. The FRI average values in miscellaneous
countries in 2001 and share of units in them (%) having leaking fuel rods (FRI > 19 Bq/g) are
represented in Figure 3.

100 100
1.0E+05 100
83
1.0E+04
FRI, Becquerels per gram

69 69 80
1.0E+03
% Unit with defects

1.0E+02 19 50 60
43
1.0E+01

1.0E+00 40
25
1.0E-01 16 17 17
8 20
1.0E-02 4 4
0 0 0 0 0 0 0 0 0 0
1.0E-03 0
United Kingdom
Slovenia

Czech Republic
Netherlands
South Africa

Finland
Spain

Slovakia
Switzerland
Japan

China
Sweden

Hungary
USA
France

Germany
Korea

Russia
Ukraine

Armenia
Belgium

Taiwan

Bulgaria
Brazil

FRI One defects % Unit with defects

Fig. 3. Average values of the WANO fuel reliability indicator in 2001 for countries with
WWER, PWR and PHWR reactors and share of units with leaking fuel rods (FRI > 19 Bq/g).

48
An attempt to understand considerable differences in FRI values in different countries
operating WWER type reactors has been undertaken abd it is described in this paper.

1st and 2nd units with the WWER-440 of B-230 design and 3rd and 4th units with the WWER-
440 of B-213 design are in operation at Bohunice NPP (Slovakia). For units 3 and 4 FRI
values were lower than 19 Bq/g (Fig.4). It confirms absence of leaking fuel rods. In too time
on units 1 and 2 with B-230 design reactors in batches occurred leaking fuel rods. On unit 1
the fuel rod leakage has given in FRI magnification above than 19 Bq/g in 1996-1997 (1996 -
219 Bq/g) and in 1999 (68 Bq/g). On unit 2 the fuel rod leakage is marked in 1991-1993
(maximum of 389 Bq/g in 1992) and in 1996-1998 (maximum of 269 Bq/g in 1997).

450
FRI, Becquerels per gram

400
350
300
250
200
150
100
50
0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-230) Unit 2 (ȼ-230) Unit 3 (ȼ-213)

Unit 4 (ȼ-213) BOHUNICE One defects

FIG. 4. Fuel reliability indicator values for Bohunice NPP (Slovakia) with WWER-440
reactors of ȼ-230 and ȼ-213 designs.

Four units of the WWER-440 of B-213 design are in operation at Dukovany NPP (Czech
Republic). The FRI values more than 19 Bq/g speaking about availability leaking fuel rods
was scored as given below (Fig.5):

• Unit 1 - in 1994-1995 (maximum of 114 Bq/g in 1995);

• Unit 2 - in 1993-1994 (maximum of 109 Bq/g in 1993);
• Unit 3 - in 1994 (24 Bq/g);
• Unit 4 - in 1993-1994 (maximum of 69 Bq/g in 1994).

Thus, the worst FRI values for Dukovany NPP (114 Bq/g, Unit 1 in 1995) was significantly
lower than for Bohunice NPP (389 Bq/g, Unit 2 in1992 and 219 Bq/g, Unit 1 in 1996), that
testifies to smaller amount of leaking fuel rods.

Four units of the WWER-440 of B-213 design are in operation in Paks, Hungary. The FRI
values, higher than 19 Bq/g, justifying availability of leaking fuel rods, was scored as given
below (Fig.6):

• Unit 1 - in 1992 (30 Bq/g);

• Unit 2 - in 1997 (20 Bq/g);
• Unit 3 - in 2001 (69 Bq/g);
• Unit 4 – in 1991 (60 Bq/g) and in 1994-1995 (maximum of 85 Bq/g in 1994).

49
 120

FRI, Becquerels per gram

100

80

60

40

20

0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-213) Unit 2 (ȼ-213) Unit 3 (ȼ-213)

Unit 4 (ȼ-213) DUKOVANY One defects

FIG. 5. Fuel reliability indicator values on Dukovany NPP (Czech Republic) with WWER-440
reactorsȼ-213 design.

90
FRI, Becquerels per gram

80
70
60
50
40
30
20
10
0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-213) Unit 2 (ȼ-213) Unit 3 (ȼ-213)

Unit 4 (ȼ-213) PAKS One defects

FIG. 6. Fuel reliability indicator values for Paks NPP (Hungary) with WWER-440 reactors of
ȼ-213 design.

At the beginning of the 14-th fuel cycle in September 1997, at the Unit 2 of NPP Paks during
power increase the temperature measurements at the outlet of the fuel assemblies (FA)
indicated anomaly of temperature distribution. Outlet temperature near loops No.2 and 3 was
rather higher than in the remaining area. The reactor could not work on a rated power, as
technical specifications limitative value of temperature in an outlet of most weighted fuel
assemblies was reached at power 95%, and the anomaly developed the tendency to sluggish
propagation. The reason of asymmetry was clogging of several fuel assemblies by an
accumulated precipitation of corrosion products. Erosion caused by a foreign material in
system and chemical impact during a decontamination have caused a damage of defensive
oxide layer on interior surfaces of a primary loop, and then high concentration of dissolved
metal ions in the coolant. The deposition of these ions has resulted in accumulation of slimes
in fuel assemblies and partial clogging.

50
After necessary preparation, the unit was shut down for refueling outage 5 months before the
schedule. All fuel assemblies in the core were discharged and cleaning of the primary circuit
was carried out. Reactor started to operate with replaced fuel. These measures resulted in the
further operation without fuel failure.

The FRI value on NPP Paks indicates high fuel rod reliability, with appearance of only a few
rods with gas untightness.

700
FRI, Becquerels per gram

600

500

400

300

200

100

0
1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-213) Unit 2 (ȼ-213) LOVIISA One defects

FIG. 7. Fuel reliability indicator values on Loviisa NPP (Finland) with WWER-440 reactors
of ȼ-213 design.

At Loviisa NPP (Finland) with two units of the WWER-440 of B-213 design in 1991-2001
the FRI values higher than 19 Bq/g were noticed only in 1995 (Fig.7) at the unit 2 (677 Bq/g).
The cause of fuel rod leaking was bound in this case to the corrosion products remained in the
coolant after a decontamination of a primary loop in 1994. Decontamination was made using
the CORD-method (Chemical Oxidating Reducing Decontamination) developed by Siemens
AG. The total surface area involved in the process was about 17 000 m2, and totally 292
kilograms of iron, chromium and nickel were removed during the four cycles of the CORD-
process. This case, as well as NPP Paks case, has resulted in sags of the coolant flow and
temperature rise at output of fuel assemblies.

Two units (Units 1 and 2) of the WWER-440 of B-230 design and two units (Units 3 and 4)
of the WWER-440 of B-213 design are in operation at Kola NPP (Russia). For units 3 and 4
FRI values in 1991-2001 were lower than 19 Bq/g. The values lower than 100 Bq/g were
scored (Fig.8):

• Unit 3 in 1992-1995 (maximal in 1995 - 795 Bq/g) and in 1999 (201 Bq/g);
• Unit 4 in 1992-1995 (maximal in 1993 - 334 Bq/g).

As for other years, FRI values for Units 3 and 4 varied from 40 to 92 Bq/g, that evidenced
availability of a very few rods with gas untightness and on high fuel operational reliability at
these Units. At the same time FRI values for Units 1 and 2 varied from 321 to 15900 Bq/g. At
these units FRI values periodically raised higher 1000 Bq/g:

• Unit 1 in 1992 (15900 Bq/g), in 1995-1997 (6600 Bq/g in 1995) and in 1999 (1220
Bq/g);
• Unit 2 in 1992-1995 (maximal in 1993 - 6820 Bq/g), in 1998-1999 (2820 Bq/g in 1998)
and in 2001 (2670 Bq/g).

51
 8000

FRI, Becquerels per gram

7000

6000

5000

4000

3000

2000

1000

0
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-230) Unit 2 (ȼ-230) Unit 3 (ȼ-213)

Unit 4 (ȼ-213) KOLA One defects

FIG. 8. Fuel reliability indicator values for Kola NPP (Russia) with WWER-440 reactors.

It indicates the availability of fuel rods with direct contact of fuel and coolant in a core. The
analysis showed that activity rising was due to the appearance of foreign objects in the Unit 2
core. Two fuel failure mechanisms were noticed:

• Local pit corrosion under crud deposits;

• Debris-fretting initiated by foreign particles.

At Novovoronezh NPP (Russia) two units with WWER-440 reactors of the B-179 design
(units 3 and 4) and one unit with a WWER-1000 reactor of the B-187 design (unit 5) are now
in operation. The FRI value higher than 19 Bq/g was scored per following years (Fig.9):

400
FRI, Becquerels per gram

350
300
250
200
150
100
50
0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 3 (VVER-440, ȼ-179) Unit 4 (VVER-440, ȼ-179)

NOVOVORONEZH (VVER-440) Unit 5 (VVER-1000, ȼ-187)
One defects

FIG. 9. Fuel reliability indicator values for Novovoronezh NPP (Russia) with WWER-440
reactors of ȼ-179 design and WWER-1000 reactor of ȼ-187 design.

52
• Unit 3 in 1993 (162 Bq/g), in 1995 (361 Bq/g) and in 1998-2001 (maximum 90 Bq/g in
2001);
• Unit 4 in 1993-1996 (maximum of 90 Bq/g in 1995) and in 1999-2001 (maximum of
127 Bq/g in 2000);
• Unit 5 in 1993-1996 (maximum of 340 Bq/g in 1996) and in1999 (134 Bq/g).

At Kozloduy NPP (Bulgaria) four units with the WWER-440 of B-230 design (units 1-4) and
two units with the WWER-1000 of B-320 design (units 5 and 6) are now in operation. The
FRI values higher than 19 Bq/g were scored per following years (Fig.10):

2500 3880
FRI, Becquerels per gram

2000

1500

1000

500

0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (VVER-440, ȼ-230) Unit 2 (VVER-440, ȼ-230)

Unit 3 (VVER-440, ȼ-230) Unit 4 (VVER-440, ȼ-230)
KO ZLO DUY (VVER-440) Unit 5 (VVER-1000, ȼ-320)
Unit 6 (VVER-1000, ȼ-320) O ne de fects

FIG. 10. Fuel reliability indicator values for Kozloduy NPP (Bulgaria) with WWER-440
reactors of ȼ-230 design and WWER-1000 reactors of ȼ-320 designs.

• Unit 1 in 1994 (361 Bq/g), in 1996 (126 Bq/g) and in 1998-2000 (maximum 471 Bq/g
in 1999);
• Unit 2 in 1993-1994 (maximum of 204 Bq/g in 1993), in 1996 (107 Bq/g) and in 1998-
2000 (maximum of 1790 Bq/g in 1998);
• Unit 3 in 1992-2001 (maximums in 1992 - 461 Bq/g, in1996 - 2180 Bq/g and in 2001 -
130 Bq/g);
• Unit 4 in 1992-1993 (maximum of 170 Bq/g in 1992), in 1996 (380 Bq/g), in 1998-1999
(maximum of 817 Bq/g in 1998) and in 2001 (574 Bq/g)
• Unit 5 in 1992-1994 (maximum of 416 Bq/g in 1992) and in 1999-2001 (maximum of
2180 Bq/g in 1999);
• Unit 6 in 1994 (3880 Bq/g) and in 2000-2001 (maximum of 682 Bq/g in 2001).

At Rovno NPP (Ukraine) two units with WWER-440 reactors of the B-213 design (units 1
and 2) and one unit with a WWER-1000 reactor of the B-320 design (unit 3) are in operation.
The FRI values higher than 19 Bq/g were scored per following years (Fig.11):

53
 12000

FRI, Becquerels per gram

10000

8000

6000

4000

2000

0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (VVER-440, ȼ-213) Unit 2 (VVER-440, ȼ-213)

ROVNO (VVER-440) Unit 3 (VVER-1000, ȼ-320)
One defects

FIG. 11. Fuel reliability indicator values for Rovno NPP (Ukraine) with WWER-440 reactors
of ȼ-213 design and WWER-1000 reactor of ȼ-320 design.

• Unit 1 in 1991-1996 (maximum of 446 Bq/g in 1994) and in 1991-1996 (maximum of

297 Bq/g in 2001);
• Unit 2 in 1994 (30 Bq/g) and in 1997-2001 (maximum of 1400 Bq/g in 1999);
• Unit 3 in 1991-2001 (maximums of 2260 Bq/g in 1992 and 1994, 8740 Bq/g in 1997
and 11417 Bq/g in 2000).

The case of maximum average value of FRI at Unit 3 of Rovno NPP might be explained by
leaving in the core for this cycle the assembly with defect of fuel rod cladding of the type
“direct contact of fuel and coolant”.

At Khmelnitski NPP (Ukraine) one unit with WWER-1000 reactor of the B-320 design is in
operation. For this unit, FRI values higher than 19 Bq/g were scored per following years
(Fig.12): in 1992-1993 (32 Bq/g in 1992); in 1995-1996 (3200 Bq/g in 1996); in 1998-2001
(maximum of 3110 Bq/g in 1999).

3500
FRI, Becquerels per gram

3000

2500

2000

1500

1000

500

0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-320) One defects

FIG. 12. Fuel reliability indicator values for Khmelnitski NPP (Ukraine) with WWER-1000.

54
At South Ukrainian NPP (Ukraine) three units with WWER-1000 reactors of the
miscellaneous designs ȼ-302 (Unit 1), ȼ-338 (Unit 2) and ȼ-320 (Unit 3) are in operation.
The FRI values higher than 19 Bq/g were scored per following years (Fig.13):

• Unit 1 in 1992 (114 Bq/g), in 1994 (21 Bq/g), in 1996-1997 (169 Bq/g in 1997ɝ.) and in
1999-2000 (803 Bq/g in 2000);
• Unit 2 in 1995-2001 (maximum of 2930 Bq/g in 1998);
• Unit 3 in 1994 (116 Bq/g) and in 2000-2001 (maximum of 972 Bq/g in 2000).

At Zaporozhye NPP (Ukraine) six units with WWER-1000 reactors of the B-320 design are in
operation. The FRI value higher than 19 Bq/g were scored per following years (Fig.14):

• Unit 1 in 1992-1996 (maximum of 2490 Bq/g in 1995 and 1996) and in 2001 (41 Bq/g);
• Unit 2 in 1992 (630 Bq/g), in 1994-1996 (maximum of 46 Bq/g in 1995 and 1996) and
in 2000 (1352 Bq/g);
• Unit 3 in 1992-1996 (maximum of 896 Bq/g in 1994) and in 2001 (160 Bq/g);
• Unit 4 in 1992-1993 (630 Bq/g in 1992) and in 2001 (187 Bq/g);
• Unit 5 in 1996 (169 Bq/g) and in 1998-1999 (maximum of 78 Bq/g in 1998);
• Unit 6 in 1998 (1250 Bq/g) and in 2000-2001 (maximum of 287 Bq/g in 2000).

3000
2930

2500
FRI, Becquerels per gram

2000

1500

1000

500

0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-302) Unit 2 (ȼ-338) Unit 3 (ȼ-320)

SOUTH UKRAINE One defects

FIG. 13. Fuel reliability indicator values for South Ukraine NPP (Ukraine) with WWER-1000
reactors of B-302, B-338 and ȼ-320 designs.

55
 2490 2490
1400

FRI, Becquerels per gram

1200

1000

800

600

400

200

0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-320) Unit 2 (ȼ-320) Unit 3 (ȼ-320) Unit 4 (ȼ-320)

ZAPO RO ZHYE Unit 5 (ȼ-320) Unit 6 (ȼ-320) O ne defe cts

FIG. 14. Fuel reliability indicator values for Zaporozhye NPP (Ukraine) with WWER-1000
reactors of ȼ-320 design.

900
FRI, Becquerels per gram

800
700
600
500
400
300
200
100
0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-320) Unit 2 (ȼ-320) Unit 3 (ȼ-320)

Unit 4 (ȼ-320) BALAKOVO One defects

FIG. 15. Fuel reliability indicator values for Balakovo NPP (Russia) with WWER-1000
reactors of ȼ-320 design.

At Balakovo NPP (Russia) four units with WWER-1000 reactors of the B-320 design are in
operation. The FRI value higher than 19 Bq/g were scored per following years (Fig.15):

• Unit 1 in 1993-1997 (maximum of 789 Bq/g in 1997) and in 2000 (32 Bq/g);
• Unit 2 in 1992-1993 (260 Bq/g in 1993) and in 1996-2001 (maximums of 426 Bq/g in
1996 and 791 Bq/g in 1999);
• Unit 3 in 1992-1993 (maximum of 581 Bq/g in 1992) and in 1996-2000 (maximum of
409 Bq/g in 1997);
• Unit 4 in 1996-2001 (maximums of 166 Bq/g in 1996 and 590 Bq/g in 2000).

At Kalinin NPP (Russia) two units with WWER-1000 reactors of the B-338 designs are in
operation. The FRI value higher than 19 Bq/g was scored per following years (Fig.16):

56
• Unit 1 in 1991-1994 (maximum of 229 Bq/g in 1992) and in 1999-2000 (118 Bq/g in
2000);
• Unit 2 in 1993-1996 (maximum of 372 Bq/g in 1993), in 1999 (23 Bq/g) and in 2001
(72 Bq/g).

400

350
FRI, Becquerels per gram

300

250

200

150

100

50

0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Unit 1 (ȼ-338) Unit 2 (ȼ-338) KALININ One defects

FIG. 16. Fuel reliability indicator values on Kalinin NPP (Russia) with WWER-1000 reactors
of ȼ-338 design.

6. CONCLUSIONS

The high level of WWER fuel reliability is noticed except of cases of heightened fuel failures
in specific operating periods of some units. The analysis of these cases has shown, that they
are caused by appearance of foreign particles in a reactor core. Two mechanisms of fuel
failure were noticed:

• Local pit corrosion under crud deposits;

• Debris-fretting by foreign particles.

More often cases of increased fuel rod failure rates are scored for WWER-440 reactors of the
old designs (ȼ-179, ȼ-230). Interior vessel surfaces of these reactors are made from carbon
steel and do not have protective coating. It increases the probability of ingress of corrosion
product ‘slimes’ into a primary coolant.

Prevention measures are recommended as following:

• Implementation of measures to improve the safety culture during maintenance and

decontamination and better control of waterchemistry norms.
• Implementation of monitoring systems to control the size and concentration of dispersed
impurity particles and corrosion products in the coolant.
• Installation of debris filter on fuel assemblies.

57
• Development and usage of means to clean primary coolant clearing from foreign objects
and corrosion product ‘slimes’.

The developed probabilistic-statistical method of analysis of fuel element leaking causes has
demonstrated its high effectiveness, especially when PIE of failed assemblies is impossible.
The use of the probabilistic - statistical analysis for identification of fuel rod failure may save
time, money and other utility’s resources.

REFERENCES

[1] E.A. PANOV, YU. M.SHESTAKOV, V.N. MIGLO, “Analysis of Trends in Fuel Rod
Depressurization and Determination of "Gas Leak" and "Pellet-Water Interaction" Type
Failures Using Radiation Monitoring Techniques of Fuel Rod Leak-Tightness”, IAEA
Technical Committee Meeting on Fuel Failure in Normal Operation of Water Reactors:
Experience, Mechanisms and Management. 26-29 May 1992, Dimitrovgrad, Russian
Federation, IAEA-TECDOC-709, IAEA, Vienna (1993) 263.
[2] Yu. SHESTAKOV, M. NANIAS, “Positive Practice Using Probabilistic-Statistical
Analysis of Fuel Element Leaking Causes”, WANO Good Practice Annual Report – 1998.
GP-MOW-98-001.
[3] Yu.M. SHESTAKOV, E.A. PANOV, “Peculiarities of Operation and Radiation
Monitoring of Water-Cooled Reactors with Faulty Fuel Rods”, IAEA Technical
Committee Meeting on Fuel Failure in Normal Operation of Water Reactors: Experience,
Mechanisms and Management, 26-29 May 1992, Dimitrovgrad, Russian Federation,
IAEA-TECDOC-709, IAEA, Vienna (1993) 272.
[4] V.P. VELYUKHANOV, L.M. VORONIN, A.G. IOLTUKHOVSKY, A.I. KANATOV,
V.K CHISTYAKOVA, “Peculiarities of operating WWER reactor core containing leaking
fuel rods”, IAEA Technical Committee Meeting on Fuel Failure in Normal Operation of
Water Reactors: Experience, Mechanisms and Management. 26-29 May 1992,
Dimitrovgrad, Russian Federation, IAEA-TECDOC-709, IAEA, Vienna (1993) 152.
[5] V.A. YURMANOV, V.A. MAMET, Yu.M. SHESTAKOV, M.M. AMOSOV, “Water
chemistry in WWER reactors”, Influence of Water Chemistry on Fuel Cladding Behavior,
IAEA Technical Committee Meeting, 4-8 October 1993, Czech Republic, IAEA-
TECDOC-927, IAEA, Vienna (1997) 433.

58
FUEL ASSEMBLY CHEMICAL CLEANING

J. SCHUNK, M. BEIER, F. KOVACS, S. MICÓ, P. TILKY

Paks NPP, Hungary

H.-O. BERTHOLDT, I. JANZIK, G. MARQUARDT

Framatome-ANP GmbH, Germany

Abstract

One year ago Framatome Advanced Nuclear Power GmbH performed chemical cleaning of 170 fuel
assemblies and conducted qualification measures for reuse of the fuel assemblies at the Hungarian
Nuclear Power Plant Paks Unit 2 (WWER, 440 MW). The paper presents this experience.

1. INTRODUCTION

NPP Paks found a thermal-hydraulic anomaly in the reactor core during cycle 14 that was
caused by corrosion product deposits on fuel assemblies (FAs) that increased the hydraulic
resistance of the FAs. Consequently, the coolant flow through the FAs was insufficient
resulting in a temperature asymmetry inside the reactor core. Based on this fact NPP Paks
performed differential pressure measurements of all fuel assemblies in order to determine the
hydraulic resistance and subsequently the limit values for the hydraulic acceptance of FAs to
be used. Based on the hydraulic investigations a total number of 170 FAs was selected for
cleaning.

The necessity for cleaning the FAs was explained by the fact that the FAs were subjected to a
short term usage in the reactor core only maximum of 1,5 years and had still a capacity for
additional 2 fuel cycles.

The realization of cleaning has to be cheaper than to buy new FAs

'p of cleaned FAs have to meet the hydraulic acceptance criterion

The cleaning concept has to include qualification measures for reuse of FAs:

– The confirmation of the cleanness of FAs with 'p

– The evidence of the integrity of FAs after cleaning

The prerequisites of the customer concerning the performance of cleaning were:

2. BOUNDARY CONDITIONS AND OVERALL CONCEPT

Essentially in context with the customer targets was the development of the overall concept.
Certain boundary conditions had to be considered for this resulting in special requirements
which had to be fulfilled, Fig. 1. The overall concept developed to meet the requirements is
summarized briefly in Fig. 1.

Regarding the kind of deposits, which were corrosion products with a different quantity, the
range of blocking of the FAs was between 10–65 %, a well-proven cleaning technology for
safe dissolution of corrosion products was necessary. It also had to ensure that it would not

59
cause additional damage to the FAs. The chemical cleaning process HP/CORD® UV was
applied as cleaning technology.

The neutron flux and the high radiation of the FAs required consideration of the neutron
reaction and of the high shielded environment of the FAs, respectively. Neutron reactions
were taken into account by using boric acid as a cleaning solution. The shielded environment
was realized by performing the chemical cleaning 14 m below borated water in pool no. 1.

With regard to waste treatment, the usage of NPP disposal systems had to be avoided.
Therefore the minimization of waste was very important and no flushing water had to be used
as a consequence. The waste was minimized by means of a bypass purification system
(resins) and in situ decomposition of the chemicals.

Boundary Conditions Requirements for the concept Overall concept

Corrosion products on FAs Well-proven cleaning technology

Chemical cleaning process
Degree of blocking 10–65 % Safely dissolution of corrosion HP/CORD UV
170 FAs russian design products; no damage of FA material

Neutron flux of FAs Consideration of neutron reaction Borated cleaning solution (boric acid)

Performance in pool No. 1

High radiation of FAs High Shielded environment 14 m under borated water

Bypass cleaning with resin column

No usage of NPP disposal systems Minimization of waste; no flushing water In situ decomposition of chemicals

Cleaning of 7 FAs simultaneously

Cleaning and qualification
Short time-frame in a special designed tank which
measures of several FAs in parallel
Limited space includes arrangements for 'p and
in the same equipment
fission product measurements

FIG. 1. Development of the overall concept.

Two additional points were of particular importance for consideration in the overall concept.
Only a short time frame was available for cleaning the FAs, because NPP Paks planned
reusing the FAs in the reactor core at the next outage.

Further the limited space in pool no. 1 that was available for installation of the equipment did
not permit placement of separate measuring equipment in parallel to the cleaning equipment
and the time frame for a post-measurement was insufficient. Therefore it was required to
conduct the cleaning and qualification measures of several FAs in parallel by using the same
equipment. Finally 7 FAs were simultaneously cleaned in a special designed tank in which the
arrangements for differential pressure ('p) and fission product measurements were included.
Furthermore NPP Paks requested the performance of qualification tests for:

Cleaning process Preliminary lab tests for material compatibility

Overall concept Test cleaning of original FAs with original process engineering

60
The harmlessness of the chemical cleaning process HP/CORD“ UV regarding the base
material was not only proven by a multitude of laboratory tests but also by the practical long-
term experience of more than 15 years of decon applications in NPPs. This has qualified the
HP/CORD“ UV process for component and system up to full system decontamination.

Independent of these facts, preliminary lab tests for material compatibility were carried out
for the chemical cleaning of FAs at the NPP Paks. The results showed that the structural
materials were compatible to the CORD process and that the HP/CORD“ UV process would
be the suitable process to dissolve the corrosion products from the FAs.

With regard to the overall concept a test cleaning of original FAs with original process
engineering had to be performed to demonstrate that by applying of the overall concept, the
fuel assemblies could be safely and reproducible cleaned and qualified.

Of particular importance was the approval of the concept and results obtained from the
pretests, the test cleaning and subsequently the industrial cleaning by:

Russian FA manufacturer MSZ

Hungarian Atomic Energy Authority

3. PERFORMANCE OF CLEANING

3.1. Arrangement of cleaning equipment

The cleaning process of the FAs was carried out in the reactor hall of Unit 2 on the reactor
podium.

The mobile decontamination system AMDA® (Automated Modular/Mobile Decontamination

Appliance) represents the decontamination equipment used for cleaning the FAs.

Refuelling machine
AMDA equipment

Pool No. 1 with

Cleaning Tank

FA Storage Pool

FIG. 2. Arrangement of cleaning equipment (AMDA) in the reactor hall of unit 2.

61
Fig. 2 gives an impression of the in situ situation for the performance of the cleaning. The
AMDA equipment was positioned nearby the FA storage pool close to pool no. 1 in which the
cleaning tank was inserted and connected with hoses to the AMDA. The controls of the
AMDA were positioned one level below of the reactor podium in order to minimize personnel
dose exposures.

The cleaning tank inside the pool was placed such that the individual tank positions could be
approached by the refueling machine in automatic mode.

Process lines

Cleaning tank cover

7 FA positions

Cleaning tank

Bottom of pool

FIG. 3. Cleaning tank temporarily installed in pool no. 1.

Fig. 3 shows the cleaning tank temporarily installed at the bottom of the pool. Framatome
ANP designed a special cleaning tank with 7 FAs positions to insert FAs of special Russian
design (fig. 4). The cleaning tank was connected to the AMDA with process lines to feed the
chemicals into the cleaning loop and consequently in the FAs. The cleaning tank remains
inside the pool when opening and closing the tank cover. The prerequisite to open the tank
inside the pool was that the water quality after the cleaning is comparable to the quality before
cleaning.

The russian design of FAs had

to be considered for the
construction of the cleaning
tank. FA channel
The russian design of FAs for
pressurized water reactors Bypass holes
have a FA channel and bypass
holes.

FIG. 4. Design of russian fuel assemblies.

62
3.2. Processes engineering

Fig. 5 shows schematically the cleaning tank positioned inside the pool and the interfacing
cleaning circuit. The flow path inside the tank is illustrated for one FA. However, the tank
permitted simultaneous cleaning of 7 fuel assemblies. Therefore, the tank has 7 inlet lines and
one common return line.

Furthermore, the scheme illustrates the dimension of the tank in relation to the pool. The tank
has a height of 4 m and is positioned 14 m under water.

FIG. 5. Implementation of arrangements for qualification measurements.

A point of particular interest was the feasibility of qualification measurements inside the
cleaning equipment.

Continuous monitoring of Kr85 was conducted in order to prove the integrity of the FAs. The
fission product Kr85 was measured by using the exhaust air exiting the cleaning circuit in the
outlet line.

The most important qualification measure was the performance of differential pressure
measurements in order to prove the cleanness of FAs. Both measures were realized by means
of the cleaning circuit just downstream of the cleaning tank.

3.3. Chemical cleaning process HP/CORD“ UV

The cleaning process of the blocked FAs was carried out by using the well known and proven
technology of HP/CORD“ UV of Framatome ANP. The synonym HP/CORD UV stands for:
Permanganic acid (H) used for the pre-oxidation step of the Chemical Oxidation Reduction
Decontamination process and UltraViolet light utilized for in situ decomposition of the decon
chemicals, subsequently. The principle of the process is illustrated in Fig. 6.

63
 FIG. 6. Principle of the HP/CORD® UV process.

Fig. 7. Flow diagram of the chemical cleaning process.

Dissolution of the fuel assembly corrosion products is done with oxalic acid. Dissolved
activity and corrosion products (Fe, Cr, Ni) are continuously removed during the entire
cleaning step by a bypass cleanup path through ion exchange resins.

At the end of each cleaning step, the chemicals are decomposed to water and CO2 with a
ultraviolet light. Therefore the water quality after the cleaning is comparable to the one
before.

64
Up to three CORD UV cycles were performed for the most blocked FAs, less blocked FAs
could be cleaned with only one cycle. The pre-oxidation step was only applied during the
second and third cleaning cycle, respectively.

Fig. 7 represents the flow diagram of the closed loop of the chemical cleaning process. The
process consists of the main circuit containing the heater and the UV skid and a bypass circuit
with the ion exchange resins.

The circuit is filled with borated water and heated to 92°C. Then the CORD chemicals are
injected into the system. The chemical cleaning solution flows into the cleaning tank,
dissolves the corrosion products from the FAs and returns from the tank into the UV skid for
decomposition of the chemicals at the end of the cleaning step. Bypass cleanup through ion
exchange resins takes place during the entire cleaning step.

Typical curves obtained during the chemical cleaning with the HP/CORD“ UV process are
shown in Fig. 8.

oxidation chemical cleaning

purification
decomposition

Iron [g]

Fe Sol + IX
Fe in Sol
Fe on IX

0 5 Time [h] 9

FIG. 8. Iron removal curves during the cleaning step.

The diagram represents 3 iron removal curves in dependency of the cleaning duration. The
lower curve represents the iron amount presently dissolved in the cleaning solution. The green
marked curve indicates the total amount of iron which is already absorbed by the ion
exchange resins and the red marked curve is the total of iron still in solution plus iron already
on ion exchange resins. No iron removal takes place during the oxidation step.

The overall trend of the 3 curves show that the total amount if iron removed from the FAs is
fixed onto ion exchange resins after the cleaning step.

65
The most important advantages of the HP/CORD“ UV technology with regard to cleaning
FAs are:

Very effective for desolve oxids for all reactor types and
all types of water chemistry

Produce reliable and reproductible results

Regenerative process
Entire cleaning is done with only one fill of water

Waste volumes are very low

Complete oxidative, in-situ decomposition of the cleaning acid to CO2

at the end of the cleaning process
No chelates in waste

3.4. Performance of qualification measurements

3.4.1. Kr85-Measurements

To prove the tightness of the FAs continuous on-line monitoring of the Kr85 isotope was
conducted during the entire chemical cleaning process. The fission product Krypton 85 would
penetrate into the cleaning solvent in case of a fuel rod defect and could thus be detected in
the non-pressurized surge tank of the AMDA cleaning system. A typical curve produced by
Kr85 monitoring of a FA batch is shown in Fig. 9. No peaks were detected, that means no
damage occured on the FAs.

25

countrate in cps

20
cleaning tank
heating on cover open
15
start cycle 1 finish cycle 1
(25th batch) (25th batch)

10

cleaning tank heating off

cover closed
5

0
05.01. 05.01. 05.01. 05.01. 05.01. 06.01. 06.01.
11:00 14:00 17:00 20:00 23:00 02:00 05:00

FIG. 9. Continuos on-line monitoring of Kr85.

66
3.4.2. Differential pressure measurements

For confirmation of the cleanness of the cleaned FAs, differential pressure (¨p) measurements
were carried out by using the cleaning tank within the cleaning circuit as illustrated in Fig. 10.
Borated water was used as flow medium and the ¨p-measurements were done under the same
operating conditions as for the cleaning.

Differential pressure was measured across the entire FA inside the cleaning tank for all 7 FAs
one after the other. Therefore, for each FA a separate signal connection line was installed to
the pressure transmitter that was located outside the pool on the reactor podium.

The ¨p-skid consists of a pressure transmitter, multiple pressure port connections to the FAs
including isolation valves and a purge water system for venting and flushing purposes. The
data acquisition and on-line evaluation were performed with the AMDA equipment.

Differential pressure measurements were carried out in order to determine the relative
hydraulic resistance of the cleaned FAs. The procedure for this is graphically illustrated in
Fig. 11.
¨p data were measured for the blocked FAs to be cleaned before starting the cleaning
procedure. To determine the hydraulic level for the cleaned FAs, ¨p measurements were
again done after having finished the corresponding cleaning cycles.

The individual ¨p ratios of blocked FAs and a new FA, represent the degree of blocking for
the FAs to be cleaned, whereas the ¨p ratio of a blocked and a cleaned FA represents the
degree of cleaning. In other words: The magnitude of increased 'p caused by blocking of the
FAs is the required magnitude for decreasing the 'p by cleaning the FAs.

FIG. 10. 'p measuring arrangement.

67
The degree of cleaning was determined for 170 FAs in context with the chemical cleaning
process by utilizing the FANP 'p measuring setup and the degree of blocking was determined
1998 by utilization of the Paflo device. The Paflo device is a setup for exclusive measurement
of the differential pressure of one FA without a cleaning function. NPP Paks has used the
Paflo device for selection of FAs to be cleaned and for establishment of hydraulic limits.

3.4.3. Establishment the cleanness criterion

The information indicated in the upper part of Fig. 11 and the limit value for hydraulic
acceptance were considered for establishment the cleanness criterion. The cleanness criterion
was determined during the test cleaning of 7 FAs with the result, that the cleanness of 7 FAs
was proven by FANP and confirmed by NPP Paks.

The cleanness criterion represents a correlation of measured ¨p ratios obtained from two
different measuring setups, the FANP and the Paflo measuring arrangement. Therefore the
requirement for Framatome was to ensure the same measuring accuracy than the Paflo device
even under the special boundary conditions. The realization of this is shown in Fig. 10.

The cleanness criterion is presented graphically in Fig. 12 indicating the comparison of the
PAFLO and FANP results regarding the relative hydraulic resistances of the cleaned FAs. The
green marked points indicate the FANP results in relation to the defined cleanness criterion.
The cleaning effect is approved when the FANP values of cleaned FAs are located above the
cleanness criterion. Or in other words: The ratio (¨pblocked/¨pcleaned)FANP has to be greater than
(¨pblocked/¨pnew)1998,Paflo/klimit to achieve the hydraulic acceptance criterion. The table beside
the graph contains the ratios indicated in the diagram.

Method: Carrying out differential pressure measurements in order to

determine the relative hydraulic resistance of FAs

Before Degree of blocking Paflo, 1998 Limits

'pblocked FA
cleaning
'pblocked FA
'pnew FA
After
'pcleaned FA Degree of cleaning FANP, 2000
cleaning
'pblocked FA
'pcleaned FA

Establishment of the cleanness criterion during test cleaning of 7 FAs

Proved by The cleanness criterion

FANP, 2000 represents a correlation of
Degree of cleaning
measured 'p ratios obtained
of 7 cleaned FAs
Confirmed by from the FANP and Paflo
Paflo, 2000 measuring arrangements

Fig. 11. Establishment the cleanness criterion to confirm the fuel assembly cleanness.

68
2.0
Measured 'p-ratios
('pblocked/'pnew)1998 Paflo/klimit
1.8
Pos. No. 'pblocked/ ('pblocked/
Cleanness Criterion 'pcleaned 'pnew)/klimit
1.6 ('pblocked/'pcleaned) FANP > FA FA FANP Paflo
('pblocked/'pnew)1998 Paflo/klimit
2 50665 1.240 1.161
1.4 3 53292 1.295 1.266

4 50220 1.343 1.254

1.2
5 53343 1.356 1.268

6 49555 1.313 1.253

1.0
7 48613 1.291 1.252
('pblocked/'pnew)1998 Paflo
0.8
1.2 1.3 1.4 1.5 1.6 1.7 1.8

Clean Not clean

Fig. 12. Assessment of the fuel assembly cleanness.

End of chemical
cleaning process

[g Fe] chemical cleaning

[mbar]

All corrosion products

are safely removed by appliance
'pblocked FA
of the HP/CORD UV process

End of decrease on 'p

Hydraulic resistances of cleaned FAs
are identical to new FAs

'p cleaned FA

Iron release 'p FA

FIG. 13. Indication of the end of the chemical cleaning process.

69
4. SUMMARY

4.1. Chemical cleaning and cleaning effectiveness

Fig. 13 contains two typical curves indicating the end of the chemical cleaning process.
Firstly, the iron removal curve shows all corrosion products are safely removed by application
of the HP/CORD“ UV process. Secondly, the drop of differential pressure indicates the end
of decrease of 'p and also that the hydraulic resistances of cleaned FAs are identical to new
FAs. It is illustrated that after a certain cleaning time the cleaning process has been finished,
the important parameter don’t change anymore. Fig. 14 shows again the assessment of the FA
cleanness from the hydraulic point of view.

The cleaning effect is proved. The relative

hydraulic resistances of 170 cleaned FAs are All 170 FAs are clean.
located above the cleanness criterion.

1.6

1.4
Charge 2 Charge 3
Charge 4 Charge 7
Charge 8 Charge 9
1.2 Charge 10 Charge 11
Charge 12 Charge 13
Charge 14 Charge 15
Charge 16 Charge 17

1.0 Charge 18
Charge 20
Charge 19
Charge 21
Charge 22 Charge 23
Charge 24 Charge 25
Charge 26 Charge 29

0.8
1.0 1.2 1.4 1.6
Clean Not clean

FIG. 14.

4.2. Facts of particular interest

Further facts are of particular interest and may be summarized as follows:

Integrity

Continuous Kr 85 monitoring demonstrated that that the CORD UV process did not impact
the tightness of FAs. No damages occurred on the 170 FAs.

Waste

All corrosion products (17000 g Fe-oxide) and the activity dissolved during the cleaning
process were transferred onto iron exchange resins.

70
Only 914 liters of resins were needed and has to be stored as radioactive waste in the spent
resin tank.

No flushing water was necessary.

Time frame

170 FAs were cleaned within approx. 10 weeks.

Approval

Based on the results obtained from the chemical cleaning and the qualification measures all
cleaned 170 FAs are accepted and approved by the MSZ and the Hungarian Authority for
reuse in the reactor.

Up to now 68 FAs have been successfully reused in the core since March 2001.

The results of the chemical cleaning of 170 FAs at NPP Paks showed that by appliance of
the overall concept consisting of the HP/CORD“ UV cleaning process and the FANP
measuring arrangements for FA qualification, FAs have been safely cleaned and the
cleaning success is safely proved.

ACKNOWLEDGEMENTS

The successful accomplishment of this project is based on the outstanding teamwork between
Framatome ANP GmbH and the Nuclear Power Plant Paks. Therefore the authors wish to
express thanks to all involved staff members of NPP Paks for their excellent cooperation
during the entire project. Especially the authors like to thank Mr. Tilky and Mr. Dr. Schunk
for their helpful and fruitful discussions, Mr. Dr. Miko for his contribution with regard to the
qualification of FAs from the hydraulic point of view and Mr. Skach for his permanent
support.

71
KNPP PRACTICES IN ASSURANCE OF RELIABLE
FUEL OPERATION FOR WWER-440 REACTORS

V. TZOTCHEVA
Kozloduy NPP plc,
Department of Chemistry and Radiochemistry,
Kozloduy, Bulgaria

Abstract

This paper makes an overview of fuel integrity control performed at KNPP. Some of the possible
causes for fuel failure are discussed. The practices in improving of fuel performance and assurance of
reliable fuel operation are represented.

1. INTRODUCTION

Fuel cladding represents the second and the most important barrier against releases of the
fission products beyond the nuclear plant. That is why an effective and reliable cladding
integrity control is essential for the safe operation of the plant. Fuel integrity control at
Kozloduy NPP is realized in two ways:

(a) During reactor operation – through radiochemical control of the primary circuit
coolant;
(b) During reactor reloading (outages) – individual assembly control.

KNPP operates 6 WWER type reactors. Units 1-4 are WWER-440/B230 with a “double unit”
configuration. They were put into operation in 1974, 1975, 1981 and 1982 respectively; units
5 and 6 are WWER – 1000 type and have been operating since 1987 and 1991.

This report reviews the results of the fuel cladding integrity control performed on WWER 440
fuel during KNPP outages and summarizes our practices in improvement of fuel reliability.

2. METHOD OF FUEL CLADDING INTEGRITY CONTROL

The individual assembly control is performed by “wet canister” method. The method consists
in isolating every tested assembly in a can situated in the spent fuel pool, its washing,
followed by forced “pumping-out” of fission products from leaking rods by changing in
system pressure. At the end of this procedure a water sample from the system is taken.
Isotopes 131I, 134Cs and 137Cs are used as indicators for leak. They are measured directly in the
water sample. Two criteria are used:

(a) Fixed criterion for 131I activity in the water sample: 3.7E6 Bq/dm3
(b) Statistical criterion: Aaverage + 3ı (ı - standard deviation)

Fuel assembly (FA) with 131I specific activity in the water sample exceeding the fixed
criterion is considered a failed FA. It is not allowed to put failed fuel assemblies into reactor
core for further use. Assemblies with specific activity of the indicators in the water sample
greater than statistical criterion are considered to be leaking. These FAs might be used again
only upon additional assessment and for each specific case a decision has to be made,
depending on the radiation situation.

72
A new more effective sipping test system within a frame of the IAEA Technical assistance
Project BUL/4/006 is intended for implementation from the next reloading.

3. SUMMARY OF THE RESULTS

A total 78 fuel cycles (FC) passed on units 1÷4, but the number of fuel tests are a little lower.
The scope of the control is shown on Ɍable 1.

The number of the tested fuel assemblies (TFAs) depending on working time into reactor core
(in fuel cycles) is summarized in the Fig. 1.

Table 1

Unit 1 Unit 2 Unit 3 Unit 4 Total

Number of FC 22 23 17 16 78
Number of FA’s cycles 7318 7667 5537 5584 26106
Number of Tested FAs 3141 3590 2133 2196 11060

4000
3500 tested FAs
3000 one year operated
2500
two years operated
2000
1500 three years operated
1000 four years operated
500
0
Unit 1 Unit 2 Unit 3 Unit 4

FIG.1. Distribution of TFAs depending on years of operating.

As it can be seen most of the FAs are tested after their first and second fuel cycle – this is an
important fact, which has to be taken into account when drawing conclusions about their
reliability.
The scope of control is shown on the next figure (Fig. 2) and the working time of the reactor
units is divided into two periods:

(a) From the start up of every reactor unit to the year 1993/4;
(b) Since 1993/4 to now.

Further this work will focus on the last 26 FC for all reactors. One of the reasons for this is
obvious – for the last 26 fuel cycles the scope of control grew more than twice. There are also
some technical considerations – changes in the fuel design, improvements in fuel testing
procedure ensuring better quality of the results etc.

73
 7000
tested FAs
6000
one year operated
5000
4000
two years operated
3000
2000 three years operated
1000
0 four years operated
1975-1993 (52 FC) 1993-2002 (26 FC)

FIG.2. Scope of the control for the different number of cycles.

As it was mentioned above the number of tested FAs differs depending on the working time
in the reactor core. If we draw the distribution of leaking fuel assemblies (LFAs) depending
on working time without taking into account this fact, the conclusions might be wrong. For
example, we may conclude that the most of the leaking FAs are one and two years old.

Having in mind the above mentioned regarding the number of the TFAs we made a
normalization of a LFAs to the number of TFAs from the same type and the results are shown
on the FIG. 3.

6
one year operated
5
4 two years operated
3
three years operated
2
1 four years operated
0
Unit 1 Unit 2 Unit 3 Unit 4 Total

FIG.3. Distribution of LFAs/TFAs depending on years of operating

Several conclusions could be made from this statistics:

(a) A specific distribution is observed for every reactor unit: the distribution for Units 3
and 4 shows that the biggest share of leaking assemblies could be attributed to three
years operated FAs. For Unit 1 the two years operated FAs have shown the greatest
number of LFAs, while for Unit 2 there is nearly uniform distribution of LFAs
between 1, 2 and 3 years operated FAs.
(b) The three years operated FAs have the greatest leaking rate in the total distribution.
The leaking rate of four years operated FAs is the smallest for all distributions except
for Unit 4.

4. ANALYSIS OF SOME POSSIBLE LEAK CAUSES

Generally the causes for fuel failures could be divided in two groups:

(a) external – depending on fuel manufacturing process

(b) internal – reasons related to the fuel operation conditions

74
Although the external causes cannot be neglected they are not of interest for this presentation.
Here has to be mentioned only that there is a big potential in the wide implementation of
quality assurance program covering the entire fuel manufacturing process.

We investigated some of the possibilities within the internal reasons using only “passive”
instrumentations – i.e. without hot cell or direct examinations. (These examinations were
planned to be done at Dimitrovgrad but this issue is still pending due to various reasons).

4.1. Problems with reactor internals

After the first implementation of statistical observation we found separate cases of repeated
fuel failures in one core position. After a careful examination the root cause was identified
(deformed flow restrictor) and eliminated.

Now a new data base for collection, analysis and archiving of every kind of data covering all
fuel “life cycle” is under implementation.

4.2. Implementation of dummy assemblies

The dummy assemblies have been implemented on Unit 1 (1988, 13FC), Unit 2 (1989, 14 FC)
and Unit 3 (1987, 7 FC). The average power was increased from 3.94 to 4.39 MW and the
average linear power of fuel rod was increased from 129 to 143 W/cm (of course all this new
values are within the allowed limits). Although at the beginning of the investigation this fact
seemed to be a very likely reason for failures, now it is rejected as a stand alone reason. May
be it has an impact in combination with other factors. This fact is supported by the same
failure rate for Unit 4 (the dummy assemblies are not installed in this core) as for Unit 1 and
Unit 3.

4.3. Position influence

LFA position analysis makes an attempt to find out some of the possible reasons causing the
assembly failure. Particularly, the influence of the following assembly positions was
investigated, considered to be “risky (dangerous)”positions:

(a) around the control assemblies (group VI region)

(b) In the periphery of the reactor core (with or without dummy assemblies (DA). This
position is interesting with respect to origin of additional hydraulic loading.

Using a methodology proposed by the WNIINM [1] with introducing of minor modifications
[2] we made (in 1999) an attempt to distinguish non typical distribution for every reactor unit
in order to find out if there is a relation between the location of the FA in the reactor core and
the probability for leaking. The analysis of the influence of the “risk position” over a leak rate
shows:
• only for Unit 2 was observed a different from normal distribution with a more LFAs in
the periphery region;
• There is no influence of the working control assembly group on the frequency of
leaks.

The first of these conclusions is in a good agreement with the assumption for vibration related
damages.

75
5. MEASURES TO IMPROVE FUEL RELIABILITY

In order to improve the fuel reliability the following measures are undertaken:

• Elaborating of actions for quality assurance covering “life cycle” of the fuel. A new
detailed quality assurance program was elaborated. This program includes actions for
quality control during manufacturing, transportation and fuel storage. The complete
history of each assembly is followed from its first year insertion into the core to its last
year of performance in the core.
• A Bulgarian-Russian task force was established with participation of representatives
of AO “TVEL”, OAO “MSZ”, VNIINM, OKB “GP” and RNC “KI”. This group is
working on finding of the root causes and elaborating measures to reduce leaking of
the nuclear fuel.
• Hot cell examination are planned to provide further information on both location and
mechanism of leaking;
• Operation of new fuel with enhanced vibration resistance. The first results are
encouraging: for totally 234 working cycles of vibration resistant FAs in the reactor
core of Unit 2 one FA is leaking.
• Enhanced measures during reloading for preventing from loose parts in primary
circuit;
• Purification of the coolant with special filtering device during maintenance.
• A system for neutron noise vibrodiagnostic of the reactor internals is in the process of
development and collecting initial data for its implementation. During each refueling
outages additional inspections of the reactor internals are carried out aimed at reducing
their vibration.
• At the end of each fuel cycle, after carrying out sipping tests a detailed calculation
model of the cycle operation mode is performed and additionally the neutron-physical
characteristics of the LFAs are analyzed.

6. CONCLUSION

• The growth of the scope of control for KNPP fuel during the last years is considerable.
• Implementation of dummy assemblies is rejected as a stand alone reason for fuel
failures.
• It is not observed an influence of the working control assembly group on the increased
frequency of leaks.
• Vibration related fretting damages are very likely reason for the bigger failure rate for
reactor Unit 2
• Implementation of assemblies with enhanced vibration resistance gives encouraging
results.

REFERENCES

[1] WNIINM, Report “Analysis of the Position of Leaking Fuel Assemblies for Kozloduy
WWER440 Units 1 to 4”, 1998, Moscow, Russian Federation.
[2] TZOTCHEVA V., et al., Kozloduy WWER440 Fuel Performance Review, Third
International Seminar on WWER Fuel Performance, Modeling and Experimental
Support, 1999, Bulgaria, BAS-INRNE (2000) 57.

76
UKRAINIAN WWER-TYPE NPP UNITS. RESULTS OF
CLADDING TIGHTNESS INSPECTION

N.YU. SHUMKOVA, O.V. BYKOV, L.P. BELOUSOVA

NAEC "Energoatom",
Kiev, Ukraine

Abstract

In the report the generalized results of cladding leak tightness inspection are reviewed for the
Ukrainian WWER-type NPP units. All observed defect types are described. Influence of reused
untight fuel assemblies on coolant activity level is being discussed briefly. Current state of fuel
performance and joint work with Russian fuel supplier JSC «TVEL» on cladding tightness inspection
instruction are considered. Some results of postreactor fuel examination are included.

1. INTRODUCTION

NAEC "Energoatom" is Ukrainian nuclear utility. There are 4 nuclear power plants (NPP) in
Ukraine. Today 13 units are in operation. 11 of them have WWER-1000 reactor another 2 are
WWER-440. 2 WWER-1000 units are under construction. The eldest unit has more than
20 years operation experience. During units operation large fuel depressurization statistics
(results of cladding leak tightness inspection (CLTI) during refuelling time) was accumulated.
Part of that statistics is presented on Fig.1. For the commercial reasons relative values were
used.

1
Zaporizhzhe NPP (untight total)
0.9
Rovno NPP (untight total)
0.8
Khmelnitskaya NPP (untight total)
0.7
Relative number

South-Ukrainian NPP (untight total)

0.6
Total amount of untight fuel rods
0.5
0.4
0.3
0.2
0.1
0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year

FIG. 1. Results of WWER-1000 CLTI for the period 1990-2002, values are relative.

Year 2002 data are not indicative due to the fact that only part of NPP units have finished its
outage. It is necessary to keep in mind that showed data describe NPP with different number
of units. That fact makes input of Rovno NPP unit 3 dominant in the whole statistics.
Zaporizhzhe NPP units are characterized by the smallest values. Khmelnitsk and South-
Ukrainian NPP have middle values of depressurization per one unit. From Fig.1 it is visible,

77
that the growth of depressurization parameters of fuel was observed last years. The same
quantity of depressurized fuel is predicted from the analysis of coolant activity for 2002 as it
was in 2001.

From the reasons of fuel depressurization analysis it is interesting to take a look at the
dependence of untight fuel assemblies (FA) quantity from the year of operation (see Fig.2).

48.6%
50%
Amount of untight fuel rods

43.8%
1990-2002 2001-2002
40%

30% 26.5%
25.0%

18.8%
20% 16.7%
12.5%
8.2%
10%

0%
1 2 3 4
Year of operation

FIG. 2. Dependence of untight fuel assemblies (FA) quantity from the year of operation.

It is necessary to take into account difference in quantity of used FAs in reactor core
depending from the year of operation. If you take into account this difference, the quantity of
untight 4-th year operation FAs per one FA of 4-th year will grow 3 times. It is necessary to
note, that depressurization parameters for Ukrainian WWER-1000 units are a little bit higher
than for the same Russian ones.

The growth of the first year operation FA depressurization, as it figured on histogram (Fig. 2),
causes anxiety of the utility. Unloaded FAs with a small burnup cause great expenses for the
utility, so the main accent in the cooperation with the fuel supplier JSC «TVEL» is made on
that problem.

For the reasons of formation of fuel loading some untight FA were used again. Secondary
CLTI of that fuel were performed after operation. Results of such practical experience
confirm the value of untight FA criterion (Alim1) that is given in fuel supplier CLTI
instruction. In particular:
— Almost all statistically untight FAs (activity level (Ai) is between statistical criterion
<A>+3ı and Alim1) had the same level of activity after the secondary usage.
— Level of activity of almost all untight FAs (activity level (Ai) greater than Alim1) was
greater after the secondary usage. It seems to be due to the defect growth.

78
So, it was recommended for all Ukrainian NPPs to avoid secondary usage of untight FA.
Secondary usage of untight FA leads to the defect evolution and to the radiological
consequences due to the large pollution of surrounding FA fuel cladding by the fuel
composition and fission products.

Depressurization statistics for the Ukrainian WWER-440 reactors are a little bit less than that
in Russia and other countries.

It is necessary to note, that the picture of fuel failures based only on depressurization statistics
is not complete. It is also necessary to take into account fuel failures due to the mechanical
reasons. That fuel can be tight yet, but changes of its mechanical integrity or features will lead
to its depressurization during further operation.

2. THE MAIN MECHANISMS OF FUEL FAILURE

Water chemistry regimes in WWER primary circuit coolant are much softer than the ones in
PWR. It allows us to believe that cladding corrosion in primary circuit coolant is insufficient.
In fact, according to the results of onsite visual inspection and postreactor fuel examination in
RIAR (Dimitrovgrad, Russia) cladding corrosion was never observed. Moreover, deposition
of corrosion products on the cladding and FA elements was also never observed. That allows
us to intend that thermo-hydraulic characteristics of FA are not changing during the operation.
It was frequently mentioned in papers VNIINM (Bochvar institute) that oxide layer thickness
in the cladding is less than 15 microns. Besides it was repeatedly marked, that the alloy E110,
used by the fuel supplier JSC «TVEL» for fuel cladding manufacturing, has a significant
reserve of corrosion resistance. That allows us to use E110 alloy for increasing fuel burnup.

Fuel rod overheating connected with excess of energy release over design limit was also never
observed for the Ukrainian NPP in case of maintenance of design requirements during
operation.

As a rule the following defect types are observed during the reactor operation:

2.1. Fretting-corrosion

It has to be mentioned that FA was designed very successfully. First of all, it is a spacer grid
construction that softens interaction in pair spacer grid - fuel rod and helps us to avoid
fretting-corrosion. But in some last cases of postreactor fuel examination in RIAR fretting
was assumed as one of the fuel failure reasons. It is difficult to prove the fretting nature of
defect without postreactor fuel examination.

2.2. Failure of FA mechanical integrity

As a rule failure of FA mechanical integrity or FA geometry change happens due to two

mechanisms:
— malfunction during the fuel transportation and reload;
— demonstration of defects that laid in design or manufacturing during FA operation.

First kind of failure is easy to detect by continuous control of fuel transport operations. It is
required from the utility to inform fuel supplier about cases of mechanical integrity failure
immediately. Then joint commission of utility, fuel supplier JSC «TVEL», scientific
organizations and designer will investigate causes of fuel failure. As a result commission

79
gives the recommendations how to avoid fuel failure next time. Those can be
recommendations how to change operation process or how to change FA or equipment
design. Then utility has to implement all the recommendations of fuel supplier, scientific
organizations and reactor designer. From the viewpoint of geometrical and mechanical
stability covered FAs is preferable, e.g. the one used in WWER-440. But usage of covered
FAs in WWER-1000 is impossible from many reasons.

2.3. Damage of fuel cladding by the extrinsic subjects (debris) during the reactor
operation

One of the main features of WWER FA is a direct-flow spacer grid. Until recent times it was
considered that little amount of debris in coolant can not damage cladding. Till 2000 there
was not found any WWER-1000 fuel rod with debris damage. But deterioration of the
primary circuit and main equipment leads to the growth of maintenance activities on the
opened surfaces of primary circuit. Obviously it rises quantity of debris that can damage fuel.
Unfortunately technology of debris extraction from primary circuit is not developed.
Operational experience indicates strong necessity of developing such technology. May be
designing the new FA with debris-catching grids will be partial solution of the problem.

2.4. Fuel depressurization with undefined reasons

As it was mentioned above, results of fabrication plants inspection carried out by NAEC
“Energoatom”, state that fuel quality is reliable. Depressurization statistics has sufficiently
risen for the last few years. In addition the reason of depressurization was not defined both by
on-site joint commission of utility and fuel supplier and by the results of postreactor fuel
examination in RIAR. Unfortunately quantity of examined FAs in RIAR is very low and it
can not be raised from many reasons. Depressurization mechanism seems to be as follows:
— Primary defect formation. In many cases primary defect is in the bottom part of the fuel
rod. Primary defect type identification is strongly difficult to make due to the secondary
effects (oxidation and hydrogenizing at the defect zone);
— Opening of primary defect due to cladding cracking up to the direct coolant-pellet
interaction;
— Formation of secondary defects due to cladding hydrogenizing. As a rule it happens
above the primary defect area. Consequences are up to the top plug separation;
— Fuel pellet structure is changing first of all in the area nearby defect. That changes result
in intensifying fission products leakage into the coolant. In case the defect opening at
the beginning of fuel cycle it would be possible to observe less activity of CLTI probe
then the one for the “gas leaking” defect.

3. ONCE MORE ABOUT CLTI RESULTS

Each type of reactor has Cladding Leak Tightness Inspection (CLTI) instructions that were
recommended by nuclear fuel supplier JSC "TVEL". As you know CLTI method is based on
measurements of reference isotopes activity. Results of CLTI measurements of coolant
activity in operating reactor define quantity of fuel assemblies that have to be tested during
refuelling outage in Cask for Cladding Leak Test. So quantity of FAs tested during refuelling
outage can differ from 0 up to whole number of FAs loaded in reactor core. Due to the large
operational experience the separate lacks of the working CLTI instruction were revealed. For
example, the spike-effect during operation specified presence of depressurized FA in the fuel
loading of not tight fuel. But sometimes depressurized FA was not found during refuelling
among tested FA.

80
Now fuel supplier JSC «TVEL» revises its CLTI instructions to comply them with
operational experience. NAEC “Energoatom” is waiting for the development of the new
criteria such as:
— Clarifying quantity of FA tested during refuelling. For example, ratio of Cs-134/137
activities during spike-effect can be used for definition of operational time of
depressurized FA;
— Classifying depressurized FA according to the defect type based on the CLTI results.
For example, value of solid fission products activity can be used.

Now NAEC “Energoatom” is considering possibility of equipping Ukrainian NPP units with
on-line CLTI testing system in the refuelling machine mast (so called sipping-control system).
It will allow us to carry out 100% CLTI tests during every refuelling. We hope that it will
help us to exclude reload of depressurized FA.

Current WWER-1000/440 CLTI instructions divide FA according to the activity of CLTI

probe (Ai) on 4 categories:
— Leak-tight FA (Ai < <A>+3×ı);
— Statistically untight FA (<A>+3×ı < Ai < Alim1);
— Depressurized FA (Alim1 < Ai < Alim2) and
— Failed FA (Alim2 < Ai).

Standards allow only tight FA and FA with defect of "Gas Leak" type for the shipping to the
storage and recycling facilities. FA with “direct contact” defect is prohibited for shipping. We
have to note that definition of “direct contact” defect is too fuzzy.

During the outage of unit #3 at Rovno NPP in 2000/2001 two FA were visually detected as
untight with depressurized peripheral fuel rods. Direct contact pellet-coolant was visually
detected. But both FA had level of activity lower then that one for the failed FA (Alim2) and
were classified according to the CLTI instruction as a depressurized one. From the viewpoint
of operation it is the indicative case as far as the utility has to decide how to use such FA
lacking reliable information. Probability of visual detection of big defect is too small (it is
proportional to the ratio of visible surface of peripheral fuel rods to total surface of all fuel
rods in FA).

It is evident that development of new criteria of large defect detection by CLTI is imminent.
For that purposes widened analysis of CLTI probes was provided. Big amount of solid fission
products activity (barium-140, niobium-95, rubidium-103, cerium-141, 144, etc.) was
detected in the all failed FAs and in some of depressurized FAs. The information was sent to
the fuel supplier JSC TVEL and to all Ukrainian NPPs. From that moment all depressurized
FAs were tested using widened CLTI probe. Some more depressurized FAs with high activity
of solid fission products were detected at Zaporizhzhe NPP. Moreover, activity of that probe
was tested twice: before and after mechanical filtration. Activity of probe has decreased by
order of magnitude after filtration. That means that from the point of view of the utility probe
has activity mainly in a form of fuel grains. Size of grains is about 10-50 micron.

That work was partially carried out together with the fuel supplier JSC «TVEL» and
«Kurchatov Institute». Now a great deal of work on using solid fission products to detect big
defect by CLTI is planned together with JSC TVEL. Additional postreactor fuel examination
in RIAR will be carried out with assistance of the fuel supplier JSC «TVEL» within the
framework of this work. Expected completion date - end of 2002.

81
It is not possible to carry out full-scale researches directly in Ukraine, because of the
following reasons:
— Ukrainian NPPs have no FA inspection stand;
— Both reactor and storage pool are placed in the same containment, thus on-site research
time is limited by outage even if we had a FA inspection stand;
— There is no chance to carry out postreactor examination directly in Ukraine due to the
absence of big hot cells and necessary equipment.

4. EXCLUSION OF FUEL FAILURES. MITIGATION OF FUEL DEPRESSURIZATION

CONSEQUENCES

We can exclude fuel failures or mitigate its consequences in two ways:

— By optimising fuel design;
— By optimising reactor operation and maintenance.

Fuel design optimization is not the subject of this report so only reactor operation
optimization is considered.

Arising fuel failure due to debris damage requires first of all debris extraction from FA and
primary circuit. As during the last 3 years debris have been visually detected in the fuel at the
unit #3 Rovno NPP and there is no technology for its extraction, it was decided to wash FA in
CLTI cask additionally. Visual control after washing did not show previously observed
debris. Unfortunately one cannot guarantee the absence of debris in FA after washing because
of impossibility to observe inner volume of FA. In this connection debris extraction still
remains undecided.

On the basis of the obtained information weak points of existing procedures and technologies
were analysed. Main attention was focused on repair works. Some additional equipment was
developed. Requirements for the preventing ingress of contamination of primary circuit were
strengthened. That changes were extended to all Ukrainian NPP.

Now Ukrainian NPP units are not equipped with the remote continual control systems of
coolant activity. Work is underway on equipping NPPs. Current CLTI instructions stipulate
shortage of interval between activity measurements. As a rule, measurements’ interval is
shorter then it is recommended in instructions. So it allows operating personnel to react on
coolant activity changes more adequately.

In case of coolant activity growth following activities were tested (but they are still not in
reactor operation documentation):

4.1. Reduction of main equipment loading cycles

Loading cycles are the controlled fuel operation parameters and their amount is limited for the
reasons of fuel integrity. So if we have depressurized fuel in the reactor core we have to
minimize amount of loading cycles to avoid further evolution of defect. For example, for the
multiunit NPP it means restriction on changing unit power. Other units will make all changes
of total NPP power output.

82
4.2. Strengthening of the demands to the transients

The aim of such activity is to relieve the stress of cladding due to pellet-cladding interaction.
For that purpose power changing rate reduces up to the lowest reasonable level (less then
existing requirements). Moreover, we try to avoid power distribution changes. For example,
movement range of cluster working group is seriously shortened. Changing of boron
concentration in the coolant provides power maintenance. All main parameters of reactor
(pressure, inlet temperature etc.) are maintained at the stable level.

4.3. Strengthening of the demands to the water chemistry

WWER water chemistry regimes do not provide special measures for binding fission products
(for example, iodine). During the operation no influence of water chemistry on coolant
activity behaviour was observed. But we try to keep stable chemical conditions that prevent
corrosion.

4.4. Unit power reduction

In some cases small power reduction (2-4%) on the unit with high coolant activity results in
sufficient activity drop (2-5 times). Coolant activity changes slowly. Stabilization time is
about a week. Such step in operation is justified in case of coolant activity close to operational
limit.

5. CONCLUSION

For the reliable fuel operation permanent contact on real fuel behaviour in reactor core with
fuel supplier is needed. Feedback of operating experience can ensure prevention of weak
points in design and their elimination. It is better to implement scientific and technical support
with assistance of the fuel supplier. JSC «TVEL» as a fuel supplier has a big deal of
operational information from all consumers and one can analyse and compare it.

In the nearest future it is desirable to equip Ukrainian NPP units with sipping-control systems.
It is evident that units under construction should be equipped with sipping-control system
before starting their operation. Other units can be equipped step-by-step.

It is desirable to pay more attention to the deterioration of the primary circuit and main
equipment. At the same time development of debris extracting technology is urgent. It is also
desirable to modify FA design using grids catching debris.

Existing and developed fuel failure detection criteria should be modified to accurately
separate defects by type. Operational experience should be analysed to select the most
effective way of fuel failure consequences mitigation.

REFERENCES

[1] KOLOBASHKHIN V.M., Radiation characteristics of irradiated nuclear fuel,

Energoatomizdat, Moscow (1983).
[2] 62444-03DIK Instruction. Cladding Leak Test Inspection of WWER-440 reactors (type
V-230 and V-213) at the operation and after cold shutdown, 0KB "Gidropress"(1979).
[3] 0401.00.00.000 DNG Instruction. Cladding Leak Test Inspection of WWER-1000
reactors at the operation and after cold shutdown, 0KB "Gidropress" (1998).

83
[4] Results of Cladding Leak Test Inspection, Rovno NPP, Zaporizhzhe NPP, South-
Ukrainian NPP, Khmelnitskaya NPP (1980-2002).
[5] Reports of participants of Russian-Ukrainian workshop, Rovno NPP, (22-25.01.01).
[6] Reports of participants of Russian-Ukrainian workshop, Collontaevo, (24-26.04.02).
[7] LUZANOVA L.M., MIGLO V.N., SLAVYAGIN P.D., Standardization of maximum
permissible value of WWER fuel cladding failure and fission products activity in
coolant, Journ. Atomic energy, V.74, iss.6, 491-497, June 1993.
[8] RIAR annual reports, Dimitrovgrad, 1996-1999, RIAR, Russia (2000).

84
MITIGATION OF FAILURES BY DESIGN AND MANUFACTURING
(Session 2)
MECHANICAL DESIGN OF THE TRIPLEWAVE DEBRIS FILTER

A. SODERLUND, R. EKLUND, S. HELMERSSON

Westinghouse Atom AB,
Vasteras, Sweden

Abstract

This paper presents the design of the TripleWave debris filter. The function of the debris filter is to
reduce the amount of debris that can enter the active fuel region and thereby reduce the risk for
damage of the fuel cladding by fretting. The design is aimed at catching long and thin debris as this
has proved to constitute the largest fretting risk. The trapping efficiency tests demonstrate that the
TripleWave debris filter reduces the risk for harmful debris to enter the fuel assembly significantly
compared with the current filter design. Fuel assemblies equipped with the TripleWave filter are
thermal hydraulic compatible with assemblies equipped with current filter design even with large
amounts of debris trapped in the filter. A full-scale endurance test showed no signs of wear between
the components in the lower part of the test assembly. The design is robust and redundant, it is
subjected to very small stresses and a very well known and proven material is used and reactor
operation without mechanical problems can be expected.

1. INTRODUCTION

Fuel reliability is basic to a safe and economical operation of nuclear plants. Westinghouse
Atom has developed a more efficient debris filter, the TripleWave debris filter. The function
of the debris filter is to reduce the amount of debris that can enter the active fuel region and
thereby reduce the risk for damage of the fuel cladding by fretting. The design is aimed at
catching long and thin debris as this has proved to constitute the largest fretting risk.

The TripleWave debris filter is placed in the inlet section of the fuel assembly, below the
bottom tie plates of the subbundles.

The design requirements for the TripleWave debris filter are:

• The debris separating efficiency shall be considerably improved and at least 90% for
harmful debris.
• Debris considered not to be harmful should pass the filter, not to risk clogging.
• The pressure drop shall match the pressure drop for current fuel assemblies.
• The design shall be robust and redundant; i.e. shall not contain any loose parts and
shall survive considerable fretting from trapped debris.
• Manufacturing of the filter shall be cost efficient and meet appropriate quality
requirements.

2. DESCRIPTION

2.1. Mechanical and Functional Design

The SVEA-96 Optima2 fuel assembly consists of:

• Fuel Bundle
• Fuel Channel
• Handle with Spring

87
The fuel bundle consists of four 5x5-1 subbundles. The subbundles are separated by a
cruciform internal structure (water cross) in the channel. The subbundles are inserted into the
fuel channel. They are supported at the bottom end by the bottom support and transition piece
(inlet piece), which is bolted to the channel The fuel assembly is lifted in a handle, connected
to the top end of the channel, and supported in the core module by a double leaf spring. The
transition piece of the fuel channel fits into the core support plate.

The bottom support is machined from stainless steel bar material and is on its inlet side
equipped with four TripleWave filter units. The bottom support is designed with grooves for
fitting the TripleWave filter units below each subbundle. The bottom support, supports the
filter units in four directions and the filter units are secured to the bottom support by lock
welds on four sides of the filter unit. The filter mounted in the fuel assembly inlet is shown in
Figure 1.

FIG. 1. Fuel Assembly Inlet Part with TripleWave Debris Filter.

The TripleWave design is aimed at catching long and slender debris as this is regarded to pose
the largest risk for fretting on the fuel rods. Simultaneously small rounded objects like
blasting grit shall be let through not to risk clogging the filter.

The filter is built from small corrugated or wavy plates, formed from 1 mm stainless steel
sheet metal. The nominal pitch for the plates is 3,8 mm and the minimum nominal width of
the flow path is 1,6 mm. The plates have a wavy shape across the flow in the inlet, a wavy
shape along the flow in the centre and then again a wavy shape across the flow in the outlet,
hence the name TripleWave.

The wavy shape of the inlet and the outlet edge serves several purposes. It functions as
support points where the plates can be welded together. It also forms a grid that provides a
first filter for large objects. In the inlet it reorients medium size objects parallel to the flow
that are subsequently trapped at their head-on entrance into the filter. Finally, the design of
the plates forms a flow path with a smooth and vertical outward flow from the filter outlet.

The plates are assembled into filter units. Each filter unit consists of 14 plates welded together
at 10-12 points to the next plate (5-6 points on each edge). The filter units are mounted in the
bottom support, see Figure 2. The filter unit is shown in Figure 3.
The units are 28,5 mm in height.

88
FIG. 2. Bottom Support with TripleWave Debris Filter.

FIG. 3. TripleWave Debris Filter Unit.

89
2.2. Material

The material in the filter plates is stainless steel type AISI 316L. It has the same composition
as the Westinghouse Atom BWR control rod material that has been used in several years and
also the same requirement on low Cobalt content (<200ppm). Initial deliveries will have a Co
content <400ppm.

2.3. Manufacturing

The first step is to cut and stamp individual plates from 1 mm thick sheet metal. Four slightly
different plates are used in each filter unit.

The formed plates are mounted in a welding fixture. All edge contact points between the
plates at the inlet and outlet edges is electron beam welded.

The finished filter units are heat-treated and then finally attached to the bottom support by
welding at a number of spots, both at the inlet and the outlet sides.

3. PERFORMANCE EVALUATIONS

3.1. Test Facilities

The performance is tested in the Westinghouse Atom test loops FRODE and BURE. Most of
the tests are done at atmospheric pressure in the temperature range 20-80°C in the FRODE
loop. The inlet is correctly modelled but only a short mock-up test assembly is used.

The endurance test to prove that the new filter does not negatively influence the fuel were
done in the BURE loop at operating data (about 7 MPa and 270°C) and one-phase flow. A
full-scale test assembly was used.

3.2. Debris Trapping Efficiency

Based on experience from fuel inspections and support from some theoretical considerations a
set of test debris has been composed for the efficiency tests. The amount of debris passing the
new filter is reduced significantly compared with the current filter design.

The TripleWave filter is almost only passed by the shortest and thinnest wire pieces. If pieces
<10 mm are disregarded from the test and only 15 mm or longer is counted, the filter is 99%
effective.

3.3. Pressure Drop

The pressure drop tests were partly run separately and partly combined with the efficiency
tests. In the separate tests Re numbers up to 90000 were obtained. Extrapolation to normal
operating conditions (Re number 150000) is straightforward.

The pressure loss is about 24% higher than for the current filter design. The total core
pressure drop is increased 1 to 1,5%, which in most cases is acceptable and does not imply
any thermal hydraulic compatibility concerns. Core stability is slightly improved through the
higher single-phase pressure drop.

90
The increase in the pressure drop loss coefficient with debris trapped in the filter is about 15%
for the full set of test debris, most of which should have been stuck in the filter during the
pressure drop registration.

4. ENDURANCE

A full-scale endurance test was performed in April - May 2001 (700 h), on a SVEA 96
Optima2 fuel assembly in the BURE loop. The inspection after the test showed no signs of
wear between the components in the lower part of the test assembly. No negative impact on
filter or filter components has been found, and the conclusion is that the bottom support with
TripleWave debris filter has shown satisfactory co-operation with the rest of the test assembly
and that reactor operation without mechanical problems can be expected.

5. CONCLUSIONS

The new TripleWave debris filter design meets or exceeds all the specified design
requirements.

The trapping efficiency tests demonstrate that the TripleWave debris filter reduces the risk for
harmful debris to enter the fuel assembly significantly compared with the current filter design.

Fuel assemblies equipped with the TripleWave filter are thermal hydraulic compatible with
assemblies equipped with current filter design even with large amounts of debris trapped in
the filter.

A full-scale endurance test showed no signs of wear between the components in the lower
part of the test assembly.

The design is robust and redundant, it is subjected to very small stresses and a very well
known and proven material is used and reactor operation without mechanical problems can be
expected.

91
FUEL FAILURES AT ANGRA 1: CAUSE AND MITIGATION

J.L.C. CHAPOT, R. SUANO, N. do COUTO

ELETRONUCLEAR,
Rio de Janeiro, Brazil

J.A. PERROTTA, J.E.R. DA SILVA, L.A.A. TERREMOTO, M. CASTANHEIRA

IPEN/CNEN-SP,
Sao Paulo, Brazil

Abstract

Angra 1 Nuclear Power Plant, a Westinghouse-designed 657 MWe PWR, was prematurely shut down
in cycle 4 due to high activity in the reactor coolant system. Inspections revealed failures in one-sixth
of the fuel assemblies (FAs). According to the fuel vendors the grid-to-rod fretting failures were
caused by grid spring force losses. In order to prevent the recurrence of the fretting failures, a new
spring design was developed. A new reload batch using the new spring design was loaded in the core
for cycle 5. In cycle 6, eight of these FAs failed, showing friction marks at outside strips of spacer
grids, due to excessive amplitude of FA vibrations. Flow tests in laboratory were performed using a
full-scale fuel assembly, and peaks of resonance due to flow induced vibration were observed.
Analysis and modeling of the fuel assembly mechanical behavior can explain the resonance observed.
All these analyses supported the need for a new fuel assembly design. In cycle 7, the reactor core was
loaded with FAs of the same design inserted in cycle 5. Two leaking assemblies were detected during
the fuel inspections conducted at the end of cycle 7. A new core was purchased from Angra 1 former
fuel supplier (cycle 1), to load cycle 8. The use of a proven fuel design got rid of the fuel failures. The
reload batch for the cycle 9 came also from the same origin as cycle 8. The zero-defect aim was also
attained in cycle 9. For cycle 10 it was decided to use twelve once burned assemblies, which had
composed cycle 7, with a damping system that, according to flow tests, could reduce the FAs
vibration. Cycle 10 is presently under operation with no indication of fuel failures. This paper
discusses Angra 1 fuel performance with emphasis to the fuel failures, the inspections and out-of-pile
tests performed, the mechanical modeling, the corrective actions proposed by the fuel vendors and the
failures mitigation.

1. INTRODUCTION

Angra 1 is a Westinghouse-designed 657 MWe PWR, 130 km from Rio de Janeiro, Brazil.
The plant is operated by ELETRONUCLEAR, a government utility. Angra 1 began
commercial operation in 1985. On 31 May 2002, the plant reached about 90% of the planned
burn-up for cycle 10. The reactor core comprises 121 (16x16 array) fuel assemblies (FAs).
Each FA contains 235 Zircaloy-4 fuel rods (FRs) supported at intervals along their length by
eight Inconel-718 spacer grids. The first core (batches A, B and C) was supplied by
Westinghouse. Siemens and Indústrias Nucleares do Brasil (INB) have provided the reloads
for cycles 2 to 7. Due to the failures occurred in cycles 4, 6 and 7 a new core with 121 original
Angra 1 FAs (standard project) was purchased from Westinghouse to load cycle 8. Later on
INB has implemented a contract of technological transfer with Westinghouse in order to
produce the standard project. New reloads of Angra 1 (started with the 10th cycle) use this FA
fabricated by INB with Westinghouse design. Table I show the core configuration and the
failures occurred during the ten operation cycles of the plant.

92
Table I. Angra 1 Fuel Assembly Failures
FA
Fuel Assemblies
Cycle Period Leaking Inspection Comments
Batch Vendor/ Enrich. Number Indication Test
Design (wt %) of F.A
01/85 A W/W 2.1 41 No No indication of fuel failure.
1 to B W/W 2.6 36 No -
01/86 C W/W 3.1 44 No
10/86 A 1 No One failed fuel (estimation of 1 fuel
2 to B 36 No Sipping can rod). Mechanism and root cause not
10/89 C 44 1 determined.
D INB/KWU 3.3 40 No
01/90 C 41 1 One failed fuel (estimation of 1 fuel
3 to D 40 No Sipping can rod). Mechanism and root cause not
08/91 E INB/KWU 3.4 40 No determined.
05/92 C 1 No In-mast Main mechanism rod-to-grid
4 to D 40 17 sipping; fretting; secondary damage; loose
03/93 E 40 4 Visual; fuel rods; some fuel rods slipped
F INB/KWU 3.4 40 1 Ultrasonic down onto end fitting. One FA of
test. batch F was damaged by handling.
12/94 A 36 No Batch G has a new design with a
5 to B 8 No higher spacer grid spring force. (No
03/96 C 1 No flow test was performed and no
F 36 No design change was done to the
G INB/KWU 3.4 40 No spacer grid mixing vanes).
06/96 G 40 8 In-mast Debris or handling failure in batch
6 to H INB/KWU 3.2 40 1 sipping; H. Loose fuel rods in batch G, rod-to
09/97 J INB/KWU 1.9 41 No Visual; grid fretting wears in batch G.
Sipping can Secondary damage observed. Grid-
to-grid fretting (west and east faces,
higher in FA middle position) for all
FA.
12/97 F 4 1 In-mast Failure mechanism was not
7 to H 36 1 sipping; determined. May be the same as
10/98 J 41 No sipping can; cycle 4 and 6. Batch L grid-to-grid
L INB/KWU 3.3 40 No Visual fretting (west and east faces, higher
in FA middle position). Some FA
from batch L showed handling
damage.
12/98 M W/W 2.1 41 No In-mast No indication of fuel failure.
8 to N W/W 2.6 40 No sipping
03/00 P W/W 3.3 40 No
03/00 M 21 No No indication of fuel failure.
9 to N 40 No
04/01 P 40 No
Q*1 INB/S*1 3.4 4 No
R W/W 3.4 16 No
04/01 M 1 90% of the cycle completed, with no
10 to now N 8 indication of failure.
P 40
Q*1 4
R 16
L*2 12
S*3 INB/W 3.4 40

(*1) – New Siemens F.A design with split mixing vanes in the spacer-grids, manufactured by INB.
(*2) – F.A with dampers inside the guide-thimbles.
(*3) – Westinghouse design, manufactured by INB.

93
2. FUEL FAILURES: CYCLES 1 TO 5

During its first three cycles fuel performance at Angra 1 was very good. The fuel inspections
carried out showed: no failures in cycle 1;one failed rod in cycle 2 (undefined cause); one
failed rod in cycle 3 (undefined cause). In cycle 4, the reactor was loaded with 120 FAs
fabricated by Siemens/INB (batches D, E and F) and one assembly that remained from the
initial core. During this fourth cycle fuel failures led to increasing activity levels in the reactor
coolant system (RCS) and the reactor was prematurely shut down on 5 March 1993, as shown
in Fig. 1 and discussed in [1].

Ultrasonic testing performed during outage found 64 leaking FRs in 17 FAs of batch D, that
had been loaded in cycle 2 (Figure 2).

The visual inspections identified grid-to-rod-fretting as the main failure mechanism of the
leaking assemblies. Seventy-six “loose” and/or fretted FRs were observed in the failed 17 FAs
of batch D (see Fig. 3). The RCS trend plots recorded during cycle 4 showed that the FA
containing the first failed FR had an average burn-up ranging from 19 to 24 MWD/kgU.

1.00E+01 100

90
1.00E+00
80

1.00E-01 70
ACTIVITY (µ Ci/cc)

60

POWER (%)
1.00E-02

50

1.00E-03
40

1.00E-04 30

20
DEI-131
1.00E-05
Tec. Spec. Limit 10
POWER (%)
1.00E-06 0
19/6/92

15/7/92

28/7/92

10/8/92

23/8/92

18/9/92

1/10/92

14/10/92

27/10/92

9/11/92

22/11/92

5/12/92

18/12/92

31/12/92

13/1/93

26/1/93

21/2/93
6/6/92

2/7/92

5/9/92

8/2/93

6/3/93

DATE

FIG. 1. Angra1 cycle 4: dose equivalent I-131.

94
 West

F10 F04 F11

F34 F26 F18 D02 F19 F27 F35

F39 E30 E24 D22 E14 D23 E21 E31 F40

F33 E29 E39 D32 E08 D18 E05 D29 E40 E32 F36

F25 E27 D35 D13 D06 E02 D07 D15 D34 E26 F26

F09 F17 D21 E11 D05 D37 E41 D38 D08 E10 D24 F20 F12

North
South F03 D01 E20 D16 E01 E13 C29 E17 E03 D20 E16 D03 F01

F08 F18 D28 E06 D12 D40 E19 D39 D09 E07 D25 F13 F05

F24 E22 D30 D19 D11 E04 D10 D17 D31 E23 F21

F32 E36 E38 D33 E09 D14 E12 D36 E37 E33 F28

F38 E35 E25 D27 E18 D26 E28 E34 F37

F31 F23 F15 D04 F14 F22 F30

F07 F02 F08

East

FIG. 2. Angra 1 cycle 4: Failures location.

In February 1994, Siemens issued a report, Ref. [2], evaluating the root-cause of the failures.
According to Siemens they were caused by grid spring force losses occurred when the fuel
rods were inserted in the skeleton, possibly in combination with loads sustained during
transport to the site. In order to prevent the recurrence of the fretting failures, Siemens
developed a new spring design whose main characteristics were a new shape and a higher
initial force.

The cycle 5 core loading came from three sources:

• A new reload batch G (40 FAs), manufactured by Siemens/INB using the new spring
design;
• Thirty-six FAs from batch F used in cycle 4 (average discharge burn-up of 5 MWD/kgU).
The batch F assemblies have the same spring as batch D, so the burn-up of batch F was
conservatively limited to 18.5 MWD/kgU in cycle 5. Assemblies from batch E could not be
used because they have an average discharge burn-up of 15 MWD/kgU, and they would
soon reach the limit of 18.5 MWD/kgU. This would risk fretting failures similar to those
found in batch D;

• Forty-fiv Westinghouse FAs that had been stored in the spent fuel pool since they were
unloaded after the three first cycles.

No failure was detected during cycle 5 operation.

95
 Fuel
Deflector Vane Fretting Marks Rod
Grid side-strip

A1

Lower end plug of fuel rod A1 fretted away Slipped Fuel Rods

Fretting Fretting

FIG. 3. Angra 1 cycle 4: Fretting on batch D FAs.

96
3. FUEL FAILURES IN CYCLE 6

In cycle 6 Angra 1 core was loaded with 121 Siemens/INB FAs discriminated as follows:
• Forty G FAs used in cycle 5;
• Forty new FAs enriched to 3.2% (batch H);
• Forty-one new FAs enriched to 1.9% (batch J). This additional reload batch had been
purchased to be used exclusively in cycle 6 because of the fuel failures in cycle 4.

Cycle 6 started on 8 August 1996 with a low RCS activity level. On 28 August, with the
reactor at 93% of nominal power, the dose equivalent iodine-131 (DEI-131) was 1.3x10-3
µCi/g, and the sum of noble gases activities was 7.7x10-3 µCi/g. However, on 30 August the
RCS activities increased very significantly. On 4 September, DEI-131 reached 0.1 µCi/g and
the sum of gases 1.6 µCi/g. Following Angra 1 Fuel Failures Action Plan the chemical and
volume control system letdown rate had to be increased from 220 to 440 lpm and the primary
system periodically degassed. DEI-131 activity decreased continuously up to mid-January
1997 (see Fig. 4).

1.00E+01 100

90

1.00E+00
80

70
1.00E-01
ACTIVITY (µCi/cc)

60

POWER (%)
1.00E-02 50

40

1.00E-03
30

20
1.00E-04 DEI-131
Tec. Spec. Limit 10
POWER (%)
1.00E-05 0
7/9/96

6/1/97

8/2/97

2/3/97

4/4/97

7/5/97

9/6/97

1/7/97

3/8/97

5/9/97
16/8/96
27/8/96

18/9/96
29/9/96

1/11/96

4/12/96

17/1/97
28/1/97

19/2/97

13/3/97
24/3/97

15/4/97
26/4/97

18/5/97
29/5/97

20/6/97

12/7/97
23/7/97

14/8/97
25/8/97
10/10/96
21/10/96

12/11/96
23/11/96

15/12/96
26/12/96

DATE

FIG. 4. Angra 1 cycle 6: dose equivalent I-131.

From mid-January, DEI-131 started to increase smoothly up to about the end of July.
However, from August to the end of cycle (EOC) the rate of activity increase was significantly
higher than in the previous 6 months. On 05 September 1997, the last DEI-131 measured
before the scheduled plant shut down was about 0.05 µCi/g. This value is twice as high that
one detected one month before, but approximately the half of the maximum DEI-131 value
measured during cycle 6 (0.1 µCi/g, on 4 September 1996). Therefore, it should be
emphasized that dose-equivalent I-131 remained significantly below the Technical
Specifications limit (1µCi/g) during cycle 6.

97
 West

H10 H04 H11

H34 H26 H18 G04 H19 H27 H35

H39 G10 G34 J35 J09 J36 G35 G11 H40

H33 G09 G39 G26 J27 G18 J28 G27 G40 G12 H36

H25 G33 G25 J40 J19 J05 J20 J41 G28 G36 H28

H09 H17 J34 J26 J18 J12 G19 J13 J21 J29 J37 H20 H12

South
H03 G03 J08 G16 J04 G17 J01 G13 J02 G20 J06 G01 H01 North

H08 H16 J33 J25 J17 J11 G15 J10 J14 J22 J30 H13 H05

H24 G32 G24 J39 J16 J03 J15 J38 G21 G29 H21

H32 G08 G38 G23 J24 G14 J23 G22 G37 G05 H29

H38 G07 G31 J32 J07 J31 G30 G06 H37

H31 H23 H15 G02 H14 H22 H30

H07 H02 H06

East

FIG. 5. Angra 1 cycle 6: Failures location.

A number of different inspections were carried out: in-mast sipping (IMS), wet sipping and
visual inspection, Ref. [3,4,5,6,7,8]. The tests identified the following:
• Eight failed FAs in batch G [G12, G16, G19, G21, G26, G28, G29 and G40] and one in
batch H [H07] (see Fig. 5);
• Fretting wear on spacer grid side-strips of several FAs.

The main observations on grid side-strip fretting wear were:

• Fretting limited to west and east faces; and hammered out areas on south and north faces;
• Fretting most severe on grids 4 and 5, less on grids 3 and 6.
• Fretting marks observed in batches G, H and J (see Fig. 6).

The key observations on the fuel failures are shown below:

Batch G
• At least 5 “loose” rods (rod movement): 2 rods on east face of G28 and one corner rod each
in G16, G19, and G21.
• Rod fretting wear under contact spring: G21 south left corner; G19 east corner (both
“loose” rods).
• Hydride blister: G28 east; G21 south.
• No apparent primary failure locations.
• All eight FAs appeared to maintain good structural integrity; no degradation except the
“loose” rods.
Batch H:
• Severe damage on FA H-07 at FR 14, at face north below grid 2. Possible debris mark
below grid 8 (bottom) on same rod. A metallic debris found at face north, above grid 8,
between rods 6 and 7 (see Fig. 7)

98
 Fuel Assembly G04, grid 4, face west.

Fuel Assembly H19, grid 6, face west. Fuel Assembly J01, grid 3, face east.

FIG. 6. Angra 1 cycle 6: wear marks on spacer grids side strips.

Face north, fuel rod 14, between grids 2and 3. Face north, between rods 6 and 7, grid 8:
debris.

FIG. 7. Angra 1, cycle 6, fuel assembly H07.

99
3.1 Root cause evaluation

3.1.1. Fuel assembly H07

H07 is most likely to have failed by debris or by damage during core loading. The FR 14
probably degraded by secondary hydriding below grid 2. This large failure could explain the
activity behavior in cycle 6. Our interpretation is that the high I-131 activity at the beginning
of cycle has been caused by FR 14 failure early in cycle. It should be still emphasized that the
high RCS activity due to H07 failure masked batch G failures.

3.1.2. Batch G

The root-cause analysis presented below is based on the information available at the end of
1997,Ref. [9]. The visual inspections could neither show the primary failure locations of batch
G nor determine conclusively the failure mechanism. However, the failures and the fuel
damages detected in cycle 4 can help us interpreting the event observed in cycle 6. As stated
in item 2, grid-to-rod fretting has been identified as the failure mechanism of batch D in
Angra 1 cycle 4. Comparisons between the predominant characteristics registered by the
inspections carried out during the outages of Angra 1 cycles 4 and 6 are shown in Table II.
The analysis of this table shows that there are significant differences between the visual
observations in cycles 4 and 6. However, the wear marks on some batch G fuel rods, together
with “loose” rods on leaking FAs, are strong evidences that grid-to-rod fretting was the main
failure mechanism of batch G in cycle 6.

Table II. Comparison between Angra 1 cycles 4 and 6 fuel failures

CYCLE 4 CYCLE 6
(BATCH D) (BATCH G)
FRETTING MARKS ON MANY 6
THE FUEL RODS
SLIPPED FUEL RODS
DOWN ONTO THE FUEL
ASSEMBLY LOWER END MANY NONE
FITTING
“LOOSE” RODS IN THE MANY ( grids #1 to #8) 5 (mid-grids/corner rods)
GRID CELLS
GRID SIDE-STRIP MANY (mid-grids/east-
RARE
FRETTING WEAR west direction)
AVERAGE DISCHARGE 26 22.8
BURN-UP (MWD/kgU)
SECONDARY DAMAGE YES YES
(HYDRIDE BLISTERS)

The number of leaking FRs in cycle 4 was much higher than in the sixth cycle; no sliding of
FRs through the grid cells was observed in cycle 6. Therefore, apparently the rod/spring grid
contact of batch G was better than that one of batch D. Nevertheless, as can be seen from
Table II, the average discharge burn-up of batch D in cycle 4 was 3.2 MWD/kgU higher than
the average discharge burn-up of batch G in cycle 6. A simple calculation shows that, in order

100
to reach the same discharge burn-up as batch D, fuel assemblies G would have to remain more
103 EFPD in the core. Probably, the damages caused by fretting during an additional period of
about 3.5 effective full power months would not be negligible. However, as we have no means
to predict the future fuel deterioration, we were constrained to evaluate the failure root-cause
in function of the evidences arising from the inspections. These evidences pointed out to a
better rod/spring grid contact of batch G, when compared to batch D. Nevertheless, in spite of
the higher grid spring force of batch G, there has been a recurrence of fretting in cycle 6.
Undoubtedly, the friction observed on spacer grids side-strips of several FAs is the key for
understanding this phenomenon.

Rare in cycle 4, the wear marks were the dominant characteristic of cycle 6 event, indicating
that occurred a unidirectional contact (east-west) between neighbour FAs during operation.
Abnormal vibrations of the FAs probably caused the grid-to-grid contact. The unexpected
vibrations were most likely induced by the reactor coolant flow. In other words, flow-induced
vibration (FIV) would have been again the root-cause of the fretting failures in Angra 1. The
new spring design introduced in batch G, having as main characteristics a new shape and a
higher initial force, did not correct the problem, because FIV is fundamentally related to the
mixing vane design (shape/orientation), which was not changed from batch D to batch G. The
reinforced springs led to a better rod/spring grid contact, resulting in less “loose” rods in cycle
6. On the other hand, the higher spring forces changed the vibrational characteristics
(amplitudes and frequencies of the different modes) of the FAs. Due to mechanisms not
clearly understood, under the action of the coolant flow the vibrations of the reinforced FAs
were intensified, leading to the strong grid-to-grid interactions showed by the visual
inspections in cycle 6. Although these hypotheses were reasonable - the fuel vendor agreed in
general with them, Ref. [5] - they had to be confirmed by experimental tests (as it will be
shown in item 6).

4. FUEL LOADING CYCLE 7

The cycle 6 fuel failures had some important effects:
• The Regulatory Authority has imposed tougher licensing requirements.
• During cycle 7, the RCS radiochemistry analyses were performed at least once a day.
• Batch G was discarded.
• Cycle 7 startup delayed approximately 2 months.
• Cycle 7 was very short (223 EFPD).
• Batches H (twice burned) was discarded at the end of cycle 7.
• An entire new core with 121 original Angra 1 fuel assemblies has been purchased from
Westinghouse to load cycle 8.

Cycle 7 core loading came from four sources:

• A new reload batch L (40 FAs), manufactured by Siemens/INB using the same design as
batches G, H and J.
• Forty-one assemblies from batch J used in cycle 6.
• Thirty-six assemblies from batch H used in cycle 6.
• Four F assemblies that had been used in cycles 4 and 5.

Batch G was scheduled to return to the core in cycle 7. However, due to the systematic
failures detected, ELETRONUCLEAR decided to discard the 40 G FAs. They were replaced
by J batch that had been originally purchased to be used exclusively in cycle 6. Besides, H07
and its three symmetric FAs were replaced by four F assemblies, which had been used in

101
cycles 4 and 5. As the residual reactivity of batch J was very low, cycle 7 length was the
shortest one of Angra 1 operation history: 223 EFPD.

The Brazilian Regulatory Authority (CNEN) main requirements to license cycle 7 operation
were related to the estimation of the fuel failures propagation in cycle 7 as a function of burn-
up, and the evaluation of the mechanical integrity of the 81 FAs that would return to the core
in cycle 7. For the first item, ELETRONUCLEAR estimated the limiting burn-up as a
function of cycles 4 to 6 experience. For the second item, Siemens, INB and the brazilian
research institute IPEN/CNEN-SP analyzed the influence of the friction at the outer surface of
the spacer grids on the operational behavior of the FAs, and the influence of impacts due to
vibrations on the integrity of guide thimbles. The results arising from experimental tests, Ref.
[10], and theoretical calculations pointed out to the conclusion that the FAs would remain
functional during cycle 7. Therefore, safety related control rod insertion malfunctions and loss
of fuel assembly cooling geometry were not expected to occur. Based on these analyses,
CNEN allowed ELETRONUCLEAR to operate cycle 7 initially during only 80 EFPD, and
could grant an extension depending upon the RCS activity level detected at the end of this
licensed period.

Cycle 7 started on 04 December 1997. Next to the end of the licensed operation period (80
EFPD), theoretical calculations based on the RCS activities led to a number of about 2 leaking
fuel rods in the core. However, due to the relatively low activity levels detected, which were
well below the Technical Specifications limit (see Fig. 8), CNEN authorized the continuation
of the cycle operation. Cycle 7 ended on 17 October 1998. Inspections performed during
outage showed 2 failed assemblies: H36 and F29 (see Fig. 9). It should be stressed that the
single failure of batch H was on H36, the lowest burnt FA of this region (18.1 MWD/kgU).

1.00E+01 100

90

1.00E+00 80

70
ACTIVITY (µCi/cc)

1.00E-01 60
POWER (%)

50

1.00E-02 40

30

1.00E-03 20
DEI-131
Tec. Spec. Limit
10
POWER (%)

1.00E-04 0
4/12/97
15/12/97
26/12/97
6/1/98
17/1/98
28/1/98
8/2/98
19/2/98
2/3/98
13/3/98
24/3/98
4/4/98
15/4/98
26/4/98
7/5/98
18/5/98
29/5/98
9/6/98
20/6/98
1/7/98
12/7/98
23/7/98
3/8/98
14/8/98
25/8/98
5/9/98
16/9/98
27/9/98
8/10/98
19/10/98

DATE

FIG. 8. Angra1 cycle 7: dose equivalent I-131.

102
 West
L17 J21 L33

J17 L24 L07 L26 L23 L39 J14

L06 L14 H32 J25 J05 J22 H29 L09 L27

J20 L12 J38 H26 J31 H10 J32 H27 J39 L08 J19

L40 H35 H25 H37 H20 J09 H17 H38 H28 H34 L01

L31 L36 J28 J30 H15 F31 H04 F33 H14 J33 J27 L04 L18

J10 L15 J04 H08 J08 H03 J01 H01 J06 H12 J02 L29 J12
South North

L34 L38 J23 J37 H18 F29 H02 F35 H19 J34 J24 L05 L19

L22 H30 H24 H40 H13 J07 H16 H39 H21 H31 L21

J15 L02 J41 H23 J36 H06 J35 H22 J40 L28 J16

L13 L10 H33 J26 J03 J29 H36 L03 L11

J18 L30 L20 L25 L35 L16 J21

L37 J13 L32

East

FIG. 9. Angra 1 cycle 7: Failures location

5. CYCLE 8 FUEL PERFORMANCE: THE FAILURES MITIGATION

Perhaps the most important consequence of the failures in cycle 6 was that an entire new core
with 121 original Angra 1 fuel assemblies was purchased from Westinghouse to load cycle 8.
This corrective action, i.e., the use of a proven fuel design, was expected to eliminate the
fretting failures experienced by Angra 1. Additionally, as the FAs supplied by the plant
designer have built-in anti-debris filters it was also expected that the recurrence of failures by
debris would be minimized.
Three batches composed the cycle 8 core:
• M (41 FAs enriched to 2.1%)
• N (40 FAs enriched to 2.6%)
• P (40 FAs enriched to 3.245%)

Cycle 8 length was 394 EFPD. It started on 11 December 1998 and ended on 19 April 2000.
The fuel performance was very good. No failures were observed.

6. FLOW EXPERIMENTS OF ANGRA 1 FUEL ASSEMBLIES

Along 1998, Siemens/INB performed laboratory flow vibration tests in order to assess the
susceptibility of G fuel assemblies to flow-induced vibration, Refs [11,12,13,14]. The tests
were done in a low-pressure loop (< 110oC, < 6,5 m/s, <10 bar). At select locations, the flow
velocity, and the vibration of the rods and that of the fuel assembly were measured. Inductive
displacement measurement devices were applied to determine the distance between the spacer
grids and the inner test channel wall. Several windows, in the test rig wall, allowed
measurement of the flow velocity.

103
The main observations were:
• The vibration was only significant perpendicular to the cross flow direction, determined by
the unidirectional mixing vane pattern. It was assumed that this self-induced excitation
was directly related to cross flow. There was a sharp resonance, for the fuel assembly
vibration perpendicular to the mixing vane direction, in the range of 25-27 Hz.
• The amplitude of vibration was dependent on the coolant flow velocity. For flow
velocities from 5.2 to 5.6 m/s the amplitude increased and for flow velocity from 5.7 to
6.1 m/s the amplitude decreased. The maximum FA vibration amplitude was on the range
of 100 microns.
• The resonance seemed only to occur for sufficient strong grid spring forces, related to the
beginning of life (BOL) condition. The resonance was not observed for the simulation of
end of life (EOL) condition, with very low force at the grid spring.
• For BOL condition the fuel assembly, including the cage and fuel rods, was vibrating in a
highly synchronized manner. The neighbour spacer grids were vibrating with a phase
difference of 180o, and the motion of the rods in each span was fully determined by the
corresponding spacers.
• The resonance was considerably reduced by the insertion of a damper inside the fuel
assembly (rods inside the guide tubes).

Based on the test observations, Siemens/INB recommended the reinsertion of the partially
burned FAs (Batch L) in the core, but using dampers inside them.

Based also on the test information, it was developed a new mixing vane design, using split
pattern. Tests carried out on a modified fuel assembly (PS4) using the new spacer grid design
showed that the resonance was not any more a problem and no restriction was found to the
usage of this new design related to the fuel rod fretting behavior.

7. CYCLE 9 FUEL PERFORMANCE

Since the end of cycle 7 there were more than 150 FAs prematurely stored in Angra 1 spent
fuel pit, due to propensity for fretting failures. As the remaining reactivity of batches E, F, G,
H (twice burned) and L (once burned) is highly significant, ELETRONUCLEAR analyzed the
viability of starting a fuel assembly reconstitution program. Nevertheless, there was an
obstacle for accomplishing that goal: the lack of a qualified Siemens/INB skeleton. However,
the flow test results presented in item 6 indicated that, under laboratory condition, the new FA
designed by Siemens/INB (PS4) had no susceptibility to flow-induced vibration. Eletronuclear
decided then to launch a qualification program for the PS4 design. Four lead test assemblies
(batch Q - 3.4% enriched) were then inserted in the Angra 1 core in cycle 9. The other FAs
came from the following sources:
•A new reload batch R (16 FAs - 3.4% enriched), manufactured by Westinghouse (the same
design as batches M, N and P).
•Twenty-one assemblies from batch M used in cycle 8.
•Forty assemblies from batch N used in cycle 8.
•Forty assemblies from batch P used in cycle 8.

Cycle 9 started on 11 July 2000 and ended on 07 April 2001. The goal of zero fuel failures
was achieved again.

104
8. CYCLE 10: REINSERTION OF PARTIALLY BURNED FUEL ASSEMBLIES WITH
DAMPING DEVICES

At the end of cycle 7, the burn-up of batch L (Siemens/INB design) ranged from 5 to
11 MWD/kgU. As a result of the flow experiments discussed in item 6, the fuel vendor
recommended the reinsertion of partially burned FAs of batch L, with damping device, into
Angra 1 core. Due to the high remaining reactivity of those elements, ELETRONUCLEAR
decided then to use 12 L FAs to compose cycle 10 core.

Before the use of the batch L fuel assemblies all them (40 FAs) were visually inspected, Ref.
[15]. The main observations of the inspection were:
• All FAs showed wear marks at the side strip of spacer grids, fretting limited to west and
east faces; and hammered out areas on south and north faces (see Fig. 10);

FA L21 – face east– grid 3 – side strip wear FA L21 – corner south/east– grid 3 – side,
hammered out (south), strip wear (east)

FA L21 – face south – grid 4 – side strip FA L27 – face west– grid 4 – side strip wear
hammered out (baffle position)
FIG. 10. Angra 1, batch L, wear marks at spacer grids side strips.

• FA L04 had one of the grids damaged by excessive wear (see Fig. 11);
• 25% of FAs showed also wear at the grid spring of the side strip, which can cause loss of
the FR fixing force and can lead to fretting (see Fig. 12);
• FAs L08 and L01 showed damage in spacer grids due to improper handling during core
loading or unloading (see Fig. 13).

105
 FIG. 11. Angra 1 batch L - FA L04, face south, grid 4, damaged strip by wear.

FA L27 – face east – grid 5 – spring wear FA L18 – face east – grid 4 – spring wear

FIG. 12. Angra 1 batch L spacer grid strip and spring wear.

FA L08 – face north – grid 2 FA L08 – face north – grid 2

FIG. 13. Angra 1 batch L, grid damaged by handling.

106
The pattern of wear verified in batch L, which has a low average burn-up, matches well with
the flow induced vibration for beginning of life conditions simulated in flow test at laboratory
as shown in item 6. Nevertheless the amplitude of vibration observed were much bigger (more
than 10 times) in reactor than in laboratory rig.

Based on the visual inspection results, 12 batch L fuel assemblies were chosen for being used
with dampers. Besides these 12 FAs, Angra 1 cycle 10 core was loaded with:
• a new reload batch S (40 FAs - 3.4% enriched), manufactured by INB using
Westinghouse technology;
• one assembly from batch M used in cycle 8;
• eight assemblies from batch N used in cycle 9;
• forty assemblies from batch P used in cycle 9;
• four lead test assemblies (batch Q) used in cycle 9;
• sixteen assemblies from batch R used in cycle 9.

Cycle 10 started on the 6 June 2001 and will finish at 20 of July 2002. No failures had been
detected up to the middle of June 2002. Visual inspections of cycle 10 FAs are scheduled for
being done during the next outage, when the mechanical behavior for flow induced vibration
will be verified.

9. FAILURE ANALYSIS - DISCUSSION

The structure of a PWR fuel assembly is formed by a support structure (skeleton) composed
by end nozzles, guide tubes and spacer grids. The guide tubes are rigidly fixed to the end
nozzles and the spacer grids. The spacer grids support the fuel rods through fixing devices that
allows the differential movement between the fuel rod and the spacer grids (or the fuel rod and
the skeleton). The fuel assembly mechanical stiffness depends mainly on the stiffness of the
fuel rod-spacer grid fixing device. This fixing device stiffness (in each of the three directions
of displacement and the three directions of rotation) depends on the spacer grid spring and
dimple stiffness, and on the geometry characteristics of the pair spring-dimple in each spacer
grid cell. It is important to realize that the forces acting on a fuel rod “flow” from the rod to
the spacer grid, then to the guide tubes and then to the end nozzles which receives the reaction
of the core support plate structure. The forces acting on the fuel rod are the hydraulic ones and
those from differential displacements due to thermal gradients and to irradiation growth. The
fuel assembly displacements and rotations result from all forces acting in all fuel rods. An
important force acting on the fuel rod is that due to the coolant flow. The force acts mainly
perpendicular to the axis of the fuel rod and it is due to the parallel flow of the water and due
to cross-flow originated from mixing vanes in the spacer spring or from thermal gradients and
flow gradients along the fuel assembly. It is well known among fuel designers that the fuel rod
is mainly excited in its lower mode of vibration by the coolant flow, see [16]. This is an
important point in the behavior of FAs in the reactor and to identify fuel rod failure by
fretting.

Perrotta, Ref. [17], developed, using a matrix method, the computer code ELCOM for PWR
fuel assembly static structural analysis. The method description helps the understanding of the
fuel assembly structural behavior. The main assumption of the method is that the spacer grid
behaves like a rigid body, so each of its cells node displacement or rotation can be related to
the displacement and rotation of the grid center of gravity. The method takes into account all

107
fuel rods, guide tubes, spacer grids and end nozzles and constructs an equivalent beam to the
fuel assembly where the beam nodes are at the center of gravity of each spacer grid and end
nozzles. Each fuel rod or guide tube is considered as a local system and the equivalent beam
as the global system in the matrix method. The method considers up to three displacement and
three rotation directions (three-dimensional problem). The method considers linear behavior
for small displacements.

The fuel rod local system can be described by:

{Rvi}+{RMi}= [Kvi]{rvi} (1)

{Rvi} – vector of the equivalent nodal external forces acting on the ith fuel rod (this could be
from thermal gradients, mechanical forces, irradiation growth, and hydraulic forces);
{RMi} – vector of the reaction forces from the fixing device in the spacer grid (due to the
spring-dimple system in the grid cell);
[Kvi] – stiffness matrix of the ith fuel rod;
{rvi} – vector of the nodal displacement of the ith fuel rod.

The equation (1) can be set for all fuel rods taking the subscript i out, and the size of the
vectors will be 6m, being m the number of fuel rods.

The guide tube local system can be described by:

{Rtj}+{Rsj}= [Ktj]{rtj}
(2)

{Rtj}– vector of the equivalent nodal external forces acting on the jth guide tube;
{Rsj} – vector of the reaction forces from the fixing device in the spacer grid (considered to
be a rigid joint);
[Ktj] – stiffness matrix of the jth guide tube;
{rtj} – vector of the nodal displacement of the jth guide tube.

The equation (2) can be set for all guide tubes taking the subscript j out, and the size of the
vectors will be 6n, being n the number of guide tubes.

The displacement vector of the local system of the fuel rod - grid joint (spring or dimple) is
given by:

{rMi} = {rgi} – {rvi} (3)

{rgi} –displacement vector of the ith node at the spacer-grid cell (external node);

The reacting force in the fuel rod-spacer grid-fixing device is given by:
{ RMi} = [KM]{rMi} = [KM]({rgi} – {rvi}) (before fuel rod sliding) (4)

{ RMi} = {Ro} (after fuel rod sliding) (4a)

[KM] is the stiffness matrix (assumed linear) of the joint device (spring-dimple).
{Ro} is the vector of the limiting forces for fuel rod sliding in the spacer grid.

108
The equation (4) can be set for all fuel rods – spacer grid system taking the subscript i out.
The size of the vectors will be 6m, being m the number of fuel rods.

Assuming rigid body movement for the spacer grid, the following relations are obtained:

{rg} = [AM]{rG} (5)

{rt} = [At]{rG} (6)

{rG} –displacement vector for the grid center of gravity;

[AM] – transformation matrix for the fuel rods;
[At] – transformation matrix for the guide tubes.

The equilibrium equation for the spacer grids and end nozzles (global system) is given by:

{RG} = [AM]T{RM} + [At]T{Rs} + [Af]T{Rf} (7)

{RG} – vector of the external forces acting in the grids and nozzles;
[Af]T.{Rf} – this term represents the vector of reacting forces of the fuel assembly fixing
spring at the top nozzle, and this term can be written as:

[Af]T{Rf} = [Af]T(-[Kf]{rf0} – [Kf][Af]{rG}) (8)

[Af] – transformation matrix for the fuel assembly fixing spring;

[Kf] – stiffness matrix (assumed linear) of the fuel assembly fixing spring;
{rf0} – initial displacement vector of the fuel assembly fixing spring;

The following system of equations can be written:

{Rv}+[KM]([AM]{rG}– {rv}) - [Kv]{rv} = {0} (9)

{RG}-[AM]T[KM]([AM]{rG}–{rv})+[At]T{Rt}-
[At]T[Kt][At]{rG}+[Af]T[Kf]{rf0}+[Af]T[Kf][Af]{rG}) = {0} (10)

Taking the value of {rv} in (9) and replacing it in (10) gives the equilibrium equation for the
equivalent beam to the fuel assembly:

{RG}+[At]T{Rt}+[Af]T[Kf]{rf0}+[AM]T[KM]([KM]+[Kv])-1{Rv} =
([AM]T[KM][AM]-[AM]T[KM]([KM]+[Kv])-1[KM][AM] +[At]T[Kt][At]+ [Af]T[Kf].[Af]).{rG} (11)

Equation (11) can be written as:

{FG} = [KG]{rG} (12)

Where:

{FG} = {RG} + [At]T{Rt}+[Af]T[Kf]{rf0} + [AM]T[KM]([KM]+[Kv])-1{Rv} (13)

[KG] = [AM]T[KM][AM]-[AM]T[KM]([KM]+[Kv])-1[KM][AM]+[At]T.[Kt][At]+[Af]T[Kf][Af] (14)

109
It can be seen from equation (14) that the stiffness of the equivalent beam of the fuel assembly
takes into account the stiffness of all components: fuel rods, guide tubes, fuel rod-spacer grid
fixing device, fuel assembly fixing spring. As the value of the stiffness of the fuel rod fixing
device increases, the stiffness of the fuel assembly increases. The opposite is true, that means,
as the stiffness (or the fixing force) of the fuel rod in the spacer grid decreases, the stiffness of
the fuel assembly decreases. As the fixing force of the fuel rod decreases along burn-up
(mainly by creep), the stiffness of the fuel rod (local system) and the stiffness of fuel assembly
(global system) decrease along burn-up either.

It can be seen from equation (13) that the forces acting in the equivalent beam of the fuel
assembly come from forces acting directly on the spacer grids, the compression of the fuel
assembly fixing spring, the forces acting on the guides tubes and the forces acting on the fuel
rods. These last ones are transmitted to the fuel assembly by the fuel rod – spacer grid fixing
device. As the fixing force of the fuel rod decreases along burn-up (mainly by creep), the
forces transmitted from the fuel rod (local system) to the fuel assembly not necessarily
decreases, but the differential displacement between the fuel rod and the grid (fuel rod sliding
in the grid) certainly increases, which may increase fretting in the fuel rod along irradiation.

It is interesting to compare the values of stiffness for the fuel rod – spacer grid fixing device
and the fuel rod stiffness (terms of [KM] and [Kv]). Table III presents these values. From this
table is seen that the shear stiffness of the fixing device is much bigger than the fuel rod value.
This means that for the term ([KM] + [Kv]) [KM] is the most important for beginning of live
and also for end of live. For rotation there is a higher importance of the term [KM] for
beginning of live, but [Kv] gets important for end of live.

Table III. Comparison between stiffness values

Fuel Rod
Spacer Grid Device Stiffness Stiffness Value Between 1st and Other segments
Value for Beginning of Live (*) 2nd spacer grid (522 mm)
(620.5mm)
Shear ~60 N/mm 12 EJ/l3 0.69 1.16
(Spring) N/mm N/mm
Shear ~ 600 N/mm 12 EJ/l3 0.69 1.16
(Dimple) N/mm N/mm
Rotation ~ 400000 4EJ/l 88345 105016
Nmm/rad Nmm/rad Nmm/rad
(*) E= 7,8x104 N/mm2; J= 175,7 mm4

From the static equilibrium system shown before, one can estimate the natural frequencies of
the local system (fuel rod) and the fuel assembly equivalent beam. The complete time
dependent equation may be written:
.. .
[mvi]{rvi}+[c]{rvi}+([KM]+[Kv]){rvi}={Rvi(t)} (15)

[M]{rG}+[C]{rG}+[KG]{rG}={FG(t)} (16)

[mvi], [M], [c], and [C] are the mass and damper matrix at the local system (fuel rod) and fuel
assembly equivalent beam. {Rv(t)}, and{FG(t)} are the forces, same definition as Eq.(13), but

110
time dependent. The damping behavior for the fuel rods depends on the structural damping,
the pellet cladding interaction (burn-up dependent), friction between the fuel rod and the grid
spring and the viscosity damping (temperature dependant). The damping behavior of the fuel
assembly depends on the damping of all fuel rods plus the interaction of the guide tubes with
the grids and the rods inserted (control rods or burnable poison rods).

As in the case of the system stiffness, the fuel rod and the fuel assembly natural frequencies
will increase or decrease as a function of the stiffness (or fixing force) of the fuel rod fixing
device. As the fixing force of the fuel rod decreases along burn-up (mainly by creep), the
natural frequencies of the fuel rod (local system) and of the fuel assembly decrease. Fig. 14
shows the modes of vibration of the fuel assembly and fuel rod in the plane along the axial
direction (dry condition, without added mass). Table IV presents the variation of the natural
frequencies with the stiffness values of the spacer grid spring device.

F.A, KX=60 ; Krot=100000

0,6

Mode 1 f=2,55 Hz
Mode 2 f=5,40 Hz
Amplitude

Mode 3 f=8,53 Hz
0 Mode 4 f=11,98 Hz
Mode 5 f=15,44 Hz
Mode 6 f=18,17 Hz

-0,6

Fuel Assembly Axis (8 Spacer Grids)

Fuel Rod, KX=60 ; Krot=100000

0,7
Amplitude

Mode 1 f=28,41 Hz
0 Mode 2 f=34,92 Hz
Mode 3 f=36,69 Hz

-0,7
Fuel Rod Axis (8 Spacer Grids)

FIG. 14. Fuel Assembly and Fuel Rod Natural Frequencies and Modes of Vibration (BOL).

The main cause of Angra 1 fuel failure in cycles 4,6 and 7 was related to rod to grid fretting
due to fuel rod and fuel assembly vibration. Cycle 4 (batch D) fuel failures showed a pattern
related to fuel rod vibration. Fuel assembly burn-up were higher than 20 MWD/kgU, and were
related to the third cycle of the fuel assembly inside the core. Ultrasonic tests performed
showed that the majority of failed fuel rods were located in the peripheral rows of the failed

111
fuel assemblies and there were a higher concentration of failed rods at the north and south
sides, which gave an indication of preferential direction of flow induced forces acting on the
fuel rods (perpendicular to the mixing vane direction). The grid position where fretting and
loosed rods were observed in visual inspection can be seen in Fig. 15. The axial position of
fretting in the fuel rod matches very well with the highest amplitude of the first natural mode
of vibration of the fuel rod. The grids 3 and 4 (from bottom to top of the F.A) would be the
ones where spring relaxation along burn-up would be the highest. Although for some
assemblies (D14 is a good example, see Fig. 16) fretting was observed also in the first grid
(F.A bottom). These fuel assemblies should have had problems during assembling (procedure
or equipment) and the first grid spring forces should be very small. Again, in these cases, there
is a good match of the highest amplitude of vibration for the first natural mode of the fuel rod
with the fretting position. The root cause of the cycle 4 (batch D) failure is related to the cross-
flow generated by the mixing vane pattern that leads to a force acting on the fuel rod much
higher than that assumed by design.

Table IV– Comparative values of frequencies for Angra 1 fuel assembly

Fixing Device Fixing Device FA Flexure FA First Natural FR First Natural
Shear Stiffness Rotation Stiffness Equivalent Frequency Frequency
(N/mm) (Nmm/rad) Stiffness (Hz) (Hz)
(N/mm)
60 107 5621/22.12=254,1 3.72 38.9
60 105 5621/47=119.6 2.55 28.4
60 0 5621/112=50.2 1.66 21.7
30 107 5621/22,2=253.2 3.71 35.4
30 105 5621/47=119.6 2.55 27.4
30 0 5621/112=50.2 1.66 21.2
5 107 5621/23,6=238.2 3.57 19.0
5 105 5621/47,8=117.6 2.52 18.3
5 0 5621/112=50.2 1.65 16.4
1 107 5621/29.19=192.6 3.12 8.9
1 105 5621/50.87=110.5 2.41 8.8
1 0 5621/112=50.2 1.65 8.6

After cycle 4 failure, the fuel designer issued a report, Ref. [2], evaluating the root-cause of
the failures. According to it the failures were caused by grid spring force losses occurred when
the fuel rods were inserted in the skeleton, possibly in combination with the loads sustained
during transport to the site. The designer developed then a new grid spring design with a new
shape and higher initial fuel rod fixing force (batch G fuel assembly). Neither flow
experiments in hydraulic loops nor any change in the grid mixing vane were done, this means,
the exciting force in fuel rods coming from the water flow through the fuel assembly was not
evaluated or changed. Looking to the equations (9) to (16) one can conclude that the designer
increased the stiffness of the grid spring [KM] and the fuel rod sliding force {Ro}, and
consequently increased the fuel assembly equivalent beam stiffness [KG]. However there was
no change in the forces acting on fuel rods {Rv} and, consequently, in the forces acting on the
fuel assembly equivalent beam {FG}. Then the problem of excessive fuel rod vibration or fuel
rod fretting might not be solved.

112
 45

40 Fretting

Loose Rods
35

Number of Visual Observations

30

25

20

15

10

0
1 2 3 4 5 6 7 8
Grid Position

FIG. 15. Angra 1 – cycle 4: fretting and loose FRs observed during visual inspection in all
failed FAs.

20

18
Fretting
16
Loose Rods
Number of Visual Observations

14

12

10

0
1 2 3 4 5 6 7 8
Grid Position

FIG. 16. Angra 1 – cycle 4: fretting and loose FRs observed during visual inspection in FA
D14.

Cycle 6 (batch G) showed a pattern related to fuel assembly vibration. Grid wear (east-west
side), due to fretting between adjacent fuel assemblies, was located at fuel assembly position
where the highest amplitude of higher (7th/8th) modes of fuel assembly vibration would
appear. Visual inspection of batch L (same design as batch G) used in cycle 7, but just once
burned, showed the same pattern of grid wears as batch G. Fig. 17 presents the numbers of
observation of wears in the fuel assemblies grids, and shows similarity to the amplitude of
vibration of higher modes of the fuel assembly equivalent beam. As discussed in item 6, the
designer performed flow experiments in a hydraulic loop with the batch G fuel assembly

113
design. It was verified that the fuel assembly vibrates perpendicular to the spacer grid mixing
vane direction in a sharp resonance at the range of 25-27 Hz for beginning of life mechanical
condition. This resonance was not observed for end of life mechanical condition. It was
concluded that the spacer grid mixing vane pattern design was responsible for the excitation
mechanism. Looking to equation (15), one can assume that the fuel rod excitation force
{Rvi(t)} is a consequence of the coolant flow and cross-flow generated by the mixing vane. As
has been observed that the first mode is predominant for the fuel rod vibration under flow
condition, it can be also assumed that the frequency of this excitation force is in the range of
the first natural frequency of the fuel rod. So, looking to equation (16), one can assume that
the fuel assembly equivalent beam will also receive an excitation in the range of the first
natural frequency of the fuel rods (taking into account the phase angle among them). The
resonance observed in the laboratory flow experiment, similar that one observed inside the
reactor, is explained by this assumption presented before. 25 Hz represents the first natural
frequency of the fuel rod (excitation) for BOL condition, and resonance occurs at the 25 Hz
frequency of the higher mode of vibration of the fuel assembly equivalent beam. For EOL
condition there is a decrease of the natural frequency (smaller stiffness) and an increase in the
damping factor for the fuel rod and also to the fuel assembly equivalent beam, which may
decrease in a sharp way the amplitude of vibration for a resonance condition.

40
Hamered out Face South

35 Hamered out Face North

Wear Face East
Wear Face West
Number of Visual Observations

30

25

20

15

10

0
1 2 3 4 5 6 7 8
Grid Position

FIG. 17. Angra 1, batch L, number of visual observations - wear marks at spacer grids side
strips.

Laboratory flow tests with batch G fuel assembly design showed that the use of dampers
would lower the amplitude of vibration observed at the resonance. The use of dampers
(zircaloy rods inside guide tubes) was decided to be used to the batch L, low burn-up fuel
assembly. Looking to equation (16) one can verify that the damping factor [C] will increase
due to change in the damping factor of the guide tubes. This damper device does not change
the damping factor of the fuel rods. Assuming the previous experience of cycle 4, 6, and 7, rod
fretting may occur after 19 MWD/kgU. This is assumed because the forces due to the mixing
vane are still acting on the fuel rod and are higher than that predicted by design, Ref. [18].
Twelve fuel assemblies of batch L are being used (with dampers) in cycle 10 with a maximum
planned discharge burn-up of 25 MWD/kgU.

114
INB received from the designer an alternative fuel assembly design that uses split mixing
vanes at the spacer grids. Four fuel assemblies (batch Q) were manufactured by INB with this
new design and used (in an experimental basis) since cycle 9. This design with split mixing
vane would certainly mitigate the fuel assembly vibration and the fuel rod fretting due to the
smaller forces acting on the fuel rods compared to that of directional mixing vane. Visual
inspections will be carried out on this fuel, after each reactor cycle, in order to verify its
performance.

ACKNOWLEDGEMENTS

The authors would like to thank the IAEA for supporting the presentation of this work at this
TCM in Bratislava, Slovakia, and particularly to thank Dr Vladimir Onoufriev of IAEA
Division of Nuclear Fuel Cycle and Waste Technology.

REFERENCES

[1] CHAPOT, J.L.C. and FREIRE, J.S., Tracing Fuel Failures at Angra 1, Nuclear
Engineering International, p32-34, September 1994.
[2] Report KWU BT52/BT51/94/E079: 21 February 1994.
[3] CHENG, B., (EPRI/USA), Angra 1 Cycle 6 Fuel Performance Evaluation Report,
October 31, 1997.
[4] Remote Visual Inspection of Fuel Assemblies, TECNATOM, November 1997.
[5] Fuel Failures: ANGRA 1/6th cycle, SIEMENS, Report NBTM/11.97/St.
[6] PERROTTA, J.A., SILVA, J.E.R., Relatório de Acompanhamento dos Testes de
Elementos Combustíveis no Descarregamento do Ciclo 6 de Angra 1, Convênio
IPEN/CNEN-SP-ELETRONUCLEAR , 30 Setembro, 1997 (in Portuguese).
[7] PERROTTA, J.A., TERREMOTO, L.A.A., Relatório de Análise dos Testes de Sipping
dos Elementos Combustíveis no Descarregamento do Ciclo 6 de Angra 1, Convênio
IPEN/CNEN-SP-ELETRONUCLEAR , 15 Outubro, 1997 (in Portuguese).
[8] PERROTTA, J.A., TERREMOTO, L.A.A., Relatório de Análise dos Espectros gama de
Amostras Líquidas do SRR para Identificação de Np-239 no Ciclo 6 de Angra 1,
Convênio IPEN/CNEN-SP-ELETRONUCLEAR , 19 Outubro, 1997 (in Portuguese).
[9] CHAPOT, J.L.C., FREIRE, J.S., SOUSA, U.C.S., Fuel Failures in the Sixth Cycle of
Angra 1 (1998 OECD/NEA/IAEA Joint Meeting to Exchange Information on Recent
Nuclear Power Plants Events and Anual Meeting of IRS National Coordinators, Paris,
May/1998).
[10] PERROTTA, J.A. Relatório de Análise do Ensaio de Compressão de Grade Espaçadora
do E.C de Angra 1 para Verificação de Resistência e Função de Segurança com
Existência de Trinca na Tira Externa, Convênio IPEN/CNEN-SP-ELETRONUCLEAR ,
22 Novembro, 1997 (in Portuguese).
[11] KOEBKE, K., STABEL ,J. Flow Experiments with Siemens Fuel Assemblies for Angra
1- Part 1: Description of the flow experimental device – KWU Report A1C-1305570-0,
17.12.98.
[12] KOEBKE, K., STABEL, J.,NEUMANN, H. Flow Experiments with Siemens Fuel
Assemblies for Angra 1- Part 2: Description of the flow experiments – KWU Report
A1C-1305570-1, 29.1.99.
[13] KOEBKE, K., STABEL, J. Flow Experiments with Siemens Fuel Assemblies for Angra
1- Part 3: Conclusion – KWU Report A1C-1305570-2, 1.2.99.

115
[14] KOEBKE, K., STABEL, J., STEPHENS, L.G. Simulation of PWR Fuel Vibration
Response due to Different In-reactor Flow Conditions (ANS-International Topical
Meeting on Light Water Reactor Fuel Performance) Park City, Utah, USA, April 10-13,
2000.
[15] PERROTTA, J.A., SILVA, J.E.R, TERREMOTO, L.A.A. Inspeção Visual dos
Elementos Combustíveis da Série L Estocados na PCU de Angra 1, Convênio
IPEN/CNEN-SP-ELETRONUCLEAR , 24 Março, 2000 (in Portuguese).
[16] PADOUSSIS, M.P. A Review of Flow-Induced Vibration in Reactors and Reactors
Components, Nuclear Engineering Design 74(1982) 31-60.
[17] PERROTTA, J.A., MACEDO, .V.S. Structural Analysis of PWR Fuel Assemblies, 13th
SMIRT, Porto Alegre, Brazil, August 13-18, 1995.
[18] PERROTTA, J.A. Utilização de Elementos Combustíveis Irradiados de Fabricação INB
em Recargas de Angra 1, Convênio IPEN/CNEN-SP-ELETRONUCLEAR , 16
Novembro, 1999 (in Portuguese).

116
OPERATION EXPERIENCE OF WWER-440 FUEL ASSEMBLIES AND MEASURES
TO INCREASE FUEL RELIABILITY

V. CHIRKOV*, V. NOVIKOV**, A. SHARIKOV**

* Joint Stock Company MSZ, Electrostal, Moscow Region, Russian Federation

** Joint Stock Company TVEL, Moscow, Russian Federation

Abstract

The paper presents technical data for the fuel cycles used in 14 WWER-440 reactors of B-213 type
situated outside CIS-territory on the basis of the 2001 operational results. The paper reflects the
dynamics of average and maximum fuel burnup as well as information on the annual rate of the
leaking fuel rods for the above reactor group identified during the 1997-2001 discharge period. As an
example of work performed by RIAR in 2001 the paper brings forth the PIE-results of a leaking
WWER-440 fuel assemblies (FAs). It is reported that the reason behind the leaking and failed fuel
rods of the FA was interaction with a foreign object being in the coolant flow. The paper describes the
measures taken by the NPPs together with the Supplier (JSC TVEL) and Manufacturer (JSC MSZ) to
enhance the fuel operational safety.

1. INTRODUCTION

At present JSC MSZ is a major manufacturer of the nuclear fuel. The fuel is supplied and
operated at 58 units of 11 European and Asian countries, half of these NPPs having WWER-
440 reactors (28 units). JSC MSZ and JSC TVEL have a data acquisition system being used
by all NPPs to collect operational information about FAs discharged from the reactors. This
information allows to follow the dynamics of the main reliability indicators for the WWER-
440 fuel rods and assemblies as well as to take necessary actions to improve FA manufacture
for the purpose of enhancing fuel operational safety.

The analysis of the WWER-440 fuel operation is carried out on the basis of the official data
received from the NPPs.

2. ANALYSIS OF WWER-440 FA’S OPERATION

The results of the nuclear fuel operation were evaluated both from the point of view of
effectiveness of the fuel cycles being used and from the point of view of operational reliability
that should ensure a reactor safety during its service life. According to the results of the 2001
reloads the fuel cycles were characterized which are used at present in 14 reactors WWER-
440 (B-213 type) of the second generation located outside CIS countries. Fig. 1 demonstrates
that 8 reactors of those 14 ones are in the state of transfer from 3-years cycle to a 4-years
cycle with maximum enrichment 3,82%.

Fig. 1 demonstrates the dynamics of the average and maximum burnup of the fuel in the 14
reactors WWER-440 during the period of unloads 1997-2001.

117
 45
43 44.63 44.764
43.97
42.61
BURNUP, MW*day/kgU

41 41.79
39
37
35 37.40 37.75
36.45
33 34.83
34.41
31
29
27
25
1997 1998 1999 2000 2001

YEAR OF UNLOADING

ȿavg ȿɦɚɯ

FIG. 1. Dynamics of average and maximum burnup of WWER-440 (type V-213) in 1997-2001
(14 units).

The highest values of average and maximum burnup in 2001 for the above group of the
reactors were achieved at the NPP “Dukovany” (40,41(unit 4) and 44,76 MWd/kg U (unit 2),
respectively), the NPP being in a transfer state to a 4-year cycle. The highest values of the of
average and maximum burnup in 2001 among all the WWER-440 reactors were achieved at
the Kola-3 NPP (46,7 and 51,04 MWd/kgU respectively), the unit operating on the 5-year
fuel cycle with maximum enrichment 4,4%.

One of the most important indicators characterizing the performance reliability is the fuel
leakage rate.

Fig. 2 shows the annual leakage rate of the fuel rods over the 1997-2001 discharge period for
the 14 reactors of the second generation, located outside the CIS countries, which use the fuel
manufactured by JSC MSZ. It should be noted that in 2001 not a single leaking fuel assembly
was unloaded. The fuel rod leakage rate for the last 5 years was 1,5*10-6.

The average leakage rate of the fuel rods in the 14 above WWER-440 reactors (B-213 type)
within the unloading period of 1997-2001 is shown in Fig. 3. It demonstrates that within the
unloading period 1997-2001 the average annual rate of leaking fuel rods in reactors 1-2 of the
Loviisa NPP, in reactor 4 of the Bohunice NPP and in reactor 4 of the Dukovany NPP made
up 4,5˜10-6 and there were no failed fuel assemblies. As for the rest 10 reactors presented in
Fig. 3 not a single unloaded fuel assembly happened to be leaking or failed.

3. POST-IRRADIATION EXAMINATION

Figures 4 and 5 give the results of the post-irradiation examinations of one leaking WWER-
440 fuel assembly as an example of work carried out by RRC RIAR in 2001.

118
 NPP"Loviisa"-2

NPP"Loviisa"-1

NPP"Dukovany"-4

NPP"Bohunice"-4

NPP"Paks"-4

NPP"Paks"-3

3.00E-04 NPP"Paks"-2

NPP"Paks"-1
LEAKING RATE

NPP"Dukovany"-3
2.00E-04 NPP"Dukovany"-2

NPP"Dukovany"-1
R=L/N
1.00E-04 NPP"Bohunice"-3
R - annual share of leakages
NPP"Mochovce"-2
L - quantity of leaking fuel rods
NPP"Mochovce"-1
N - q-ty of fuel rods in the core
0.00E+00 * - considered on condition that: 1 leaking fuel rod
1997

1998

1999

2000

2001

for 1 leaking fuel assembly

FIG. 2. Annual rate of WWER-440 (type V-213) fuel rod leakages during the period of 1997-2001 (14
units). Average value during 5 years-1.5E-06. Average value during 2001-0.

7.0E-06 leaking failed

type V-230 type V-230
type V-179 type V-179
6.0E-06 type V-270 type V-270
type V-213 type V-213

5.0E-06
LEAKING RATE

4.0E-06

3.0E-06

2.0E-06

1.0E-06

0.0E+00
NPP NPP NPP NPP NPP NPP NPP
"Loviisa"-1 "Bohunice"- "Bohunice"- "Dukovany"- "Moch "Paks"-1 "Paks"-3
4 3 2 ovce"-1

FIG. 3. Average rate of WWER-440 leaking and failed fuel rods (1997-2001 period of fuel discharge).

119
Fig. 4 shows that the FA contained 5 leaking fuel rods as well as a piece of a metal wire, stuck
in the lower lattice. Comparing the appearance and geometry parameters of debris-damages of
the fuel rods as well as considering the orientation of these damages in the bundle (see Fig. 5),
there are good reasons to assume that the leakage of the above fuel rods could be caused by
interaction with a similar wire. The analysis of the wire chemical composition showed that the
wire was made of carbon or low-alloy steel, and the deposits found in the place of leakage of
one of the five fuel rods of the bundle are the corrosion products of that material. It can be
assumed that the wire fragment causing leakage of the four fuel rods has corroded by the time
of the FA discharge from the reactor (650 EFPD) and was not discovered during PIE.

FIG. 4. Leaking fuel rod locations in the FA.

Thus, the interaction with the foreign objects of the coolant flow caused leakage and failure of
the fuel rods.

More detailed information on failure mechanism is given in paper presented by Mr.

A.Smirnov at this meeting.

4. MEASURES AIMED AT ENHANCEMENT OF FUEL SAFETY

4.1. Implemented measures

Some modifications have been recently made in the FA design (FA of the 2-d generation) to
improve the economic indices of the fuel operation which are expressed in higher burnup,
increase of effective operation and reduction of possible fuel failures.

4.1.1. Working fuel assembly:

• increased fuel load due to the increase of the active fuel length;
• dismantling design that enables to dismount the working fuel assembly, withdraw leaking
fuel rods from the bundle and replace them with the dummies;
• engineering solutions (optimization of spacer grid positioning along the bundle height,
backlash-free securing of the fuel rods in the lower lattice, a system of supporting ribs)
aimed at the improvement of the working fuel assembly stiffness and vibrostability.

120
 FIG. 5. Appearance and orientation of fuel rod debris damages in the FA cross-section.

4.1.2. Fuel follower of control rod:

• An engineering solution has been implemented that eliminates power density ramping in
the working fuel assemblies surrounding the fuel follower of control rod, thus reducing
possible failure of the fuel rods of the working fuel assembly during higher burnup fuel
cycles.

4.2. Measures under development

To prevent ingress of the debris-particles into the fuel assembly an anti-debris device is
supposed to be used. The device might enable to reduce a potential interaction of the debris-
particles with the fuel rods and FA bundle components and will decrease possible FA leakage
respectively. Such activities are currently carried out by JSC MSZ in association with OKB
Gidropress.

5. CONCLUSIONS

The NPPs in association with the Supplier (JSC TVEL) and Manufacturer (JSC MSZ) are
steadily improving the effectiveness of the fuel cycles being used and fuel operational
reliability.

The actions to be taken for the enhancement of the fuel safety could be divided into two
groups. One group should combine the activities aimed at the FA improvement. JSC TVEL
together with the Manufacturer are currently carrying out such work, an implementation of
vibro-stable fuel assemblies and those with an anti-debris lattice included.

121
The other group of activities should be carried out by NPPs in the framework of the quality
assurance of the fuel performance. It will require a system of measures and techniques to be
developed by the NPPs for cleaning the primary coolant from the corrosion products and
foreign objects. The latter group of activities is becoming of essential importance because of
the work carried out for the purpose of extension of the service life of the WWER-440 first
generation reactors.

Table 1. Brief description of fuel cycle

Unit/Units Brief cycle description
Bohunice 3 Mixed 3-years fuel cycle, a part of FAs is left for the 4th year; max.
and 4 enrichment-3.6%
Dukovany 1-4 Transitional from a mixed 3-years fuel cycle to a 4-years cycle; max.
enrichment-3.82%
Loviisa 1 and Transitional from a 3-years fuel cycle (max. enrichment-3.6%)to a 4-years
2 cycle with max. enrichment-3.82%
Mochovce 1 Transient cycle from an unsteady 3-years fuel cycle with max. enrichment-
3.6% to a 4-years cycle with max. enrichment-3.82%
Mochovce 2 Unsteady 3-yeares fuel cycle with max. enrichment-3.6%
Paks 1, 2 and Mixed 3-years fuel cycle, a part of FAs is left for the 4th year; max.
4 enrichment-3.6%
Paks 3 Transient cycle from an unsteady 3-years fuel cycle with max. enrichment-
3.6% to a 4-years cycle with max. enrichment-3.82%

122
FUEL FAILURE MITIGATION AT THE RINGHALS PLANT

T. ANDERSSON
Ringhals AB,
Väröbacka, Sweden

Abstract

Based on the various types of fuel failures that have occurred at Ringhals, the paper describes the
efforts made at Ringhals and by the fuel suppliers to reduce the risk of fuel failures. The defense
against secondary degradation at Ringhals 1 includes 10x10 liner fuel with debris filters, a
conservative management of PCI margins as well as measures to prevent debris from entering the
coolant system. For the PWRs the risk for fuel rod fretting remains a concern. The design margins
against fuel rod fretting need to be improved on new fuel types.

1. INTRODUCTION

During the past decade 13 new fuel types have been introduced at the Ringhals plants (the
number depends somewhat on how a new fuel type is defined). All of these fuel types have
been or will be delivered in reload quantities. In addition, within a given fuel type, there are
usually a number of varieties: different material compositions, different dashpot designs etc.
With the exception for Ringhals 2, the Ringhals cores are therefore composed of a mixture of
different fuel types and varieties of a given fuel type.

The reason for this diversity is both technical and commercial. To maintain competition, new
DEMO/LFA fuel types are introduced. Fuel contracts are usually signed for 4-year periods,
which is less than the period required to fully replace the fuel types of a given core.

Assembly bow has required a number of design changes within a given fuel type - reinforced
dashpot, relaxed hold-down forces and thicker guide thimbles. Further modifications have
been introduced in order to limit assembly growth and to improve corrosion properties.
Fragema, Siemens, Westinghouse and ABB-Atom have supplied the fuel.

The introduction of new fuel types requires an assessment of both performance and reliability.
New fuel types generally have better thermal performance than their predecessors, which
gives an economic incentive to introduce new fuel types. New fuel types are also expected to
be more reliable than their predecessors. On the other hand there’s the risk that the new fuel
types have unknown reliability problems that will show up at high burnups.

One difficulty in the reliability assessment is to account for the variability in the fuel’s
operating environment: surrounding fuel types, core locations as well as power dependent
flow conditions. The manufacturers flow testing hasn’t been able to account for all these
variables.

Fuel types that operate without problems in Ringhals have been subject to flow induced
fretting failures in other plants. The fact that it has been possible to verify failure mechanisms
in test loops after the delivery of the fuel to Ringhals or to verify that the Ringhals reactor in
question was not subject to a particular fretting mechanism shows that loop testing can be
made more efficient.

123
The management of the PCI margins gained by the introduction of 10x10 liner fuel requires a
trade-off between the demand for improved fuel economy and operating flexibility and the
requirement to maintain sufficient margins to PCI.

The decision to introduce a new fuel type is also influenced by the marginal costs. If
improved thermal or mechanical properties come at little or no extra costs it is likely that that
such features will be introduced irrespective of if they are needed or not from a licensing
point of view - resulting in better operating margins if the boundary conditions remain the
same. If the cost is too high, much needed hardware or methodology improvements might not
be made.

Considering all aspects of fuel reliability the total costs associated with failures and poor
performance can be quite substantial. There are costs for fuel repair, fuel inspection programs,
safety analysis, restrictions on loading and handling of the fuel, increased core tilts, increased
uncertainties in the peaking factors, increased background activity levels, reduced power
peaking factors, prolonged outages, PIE, restrictions in core design, burnup restrictions etc.

2. THE RINGHALS PLANTS

Ringhals is Sweden’s largest power plant. Last year the production from the four units
equaled 18% of the Swedish electricity consumption. The thermal ratings of the Ringhals
plants are the following:

Ringhals 1: Asea-Atom BWR, external pumps, 2500 MWth

Ringhals 2: Westinghouse PWR, 15x15 lattice, 2652 MWth
Ringhals 3: Westinghouse PWR, 17x17 lattice, 2775 MWth
Ringhals 3: Westinghouse PWR, 17x17 lattice, 2775 MWth

The plants are operated on 12-month cycles. Refueling outages are in the period May to
September. Extensive periods of coast down operation are used in order to improve the fuel
economy and to adjust production to the reduced demand in the summer period.

The backend costs in Sweden are at present very low. There's therefore little incentive to go to
very high burnups. Presently the most economical discharge burnup is in the 45–50
MWD/KgU range.

3. FUEL TYPES AND FUEL SUPPLIERS

Siemens replaced ABB-Atom as fuel supplier to Ringhals 1 in 1999. The present replacement
fuel to Ringhals 1 is ATRIUM 10B. For the period 2003–2006, Framatome ANP will deliver
replacement fuel to Ringhals 1 (ATRIUM 10B). The core composition of Ringhals 1 is shown
in fig. 3.

Fuel supplier to Ringhals 2 during the period 1995–2002 is Westinghouse. Delivered fuel type
is Performance + with IFMs. Cladding and grid material is ZIRLO. Framatome ANP is
contracted as the fuel supplier for the period 2003–2006. Fuel type to be delivered is AFA 3G
with M5 cladding and grid material.

Fragema has delivered fuel to Ringhals 3 and 4 during the period. In 1999, AFA 3G replaced
AFA 2G as replacement fuel. For the period 2003–2006 Framatome ANP will supply HTP as
replacement fuel.

124
4. NUMBER OF FUEL FAILURES

The total number of fuel failures (leakers) at Ringhals between 1989–2001 were 38. Failure
mechanisms have been PCI (Ringhals 1), debris fretting, fuel rod fretting wear and
mechanisms related to manufacturing. For 9 of the failures, the failure causes are still
unknown. The division of the failures among the units is shown in table. 1. Most failures have
appeared at Ringhals 3.

Prior to 1989, a large number of fuel failures on Ringhals 2 were caused by baffle jetting. In
1989, the barrel coolant flow in Ringhals 2 was converted from “down-flow” to “up-flow” to
prevent baffle-jetting failures.

Starting in 1993, debris filters have been introduced in the Ringhals plants. All of the fuel
assemblies in the PWR cores are equipped with debris filters. The fraction of fuel assemblies
with debris filter on Ringhals 1 is about 30%.

Table 1: Fuel failures at Ringhals units between 1989 and 2001

Failure mechanism Ringhals 1 Ringhals 2 Ringhals 3 Ringhals 4

Debris 3 1 3 5
PCI 6
Rod fretting 3
Corrosion 1
Primary hydriding 5
Manufacturing 1 1
Unknown 1 7 1
Total 9 2 20 7

5. STRATEGIES TO IMPROVE FUEL RELIABILITY

Thermo-mechanical spare margins are included in the design and licensing of the fuel. The
fuel is presently licensed to a burnup of 60 MWD/kgU despite the fact that the average batch
discharge burnup is in the range 45–50 MWD/kgU.

The introduction of IFMs in PWR gives additional DNB-margins that have not been licensed
as higher peaking factors. Parts of the design margins have been used to account for assembly
bow and thimble plug removal.

The design margins also simplify burnup extensions for lead assemblies within a given batch.
Examples of margin recoveries are the introduction of IFMs on the PWRs and liner 10x10
fuel on the BWR.

An extensive reliability assessment done for all offered fuel types is made prior to contract.
The assessment includes fuel economy, thermal performance, mechanical design and
operating experiences. Important parameters are debris filter efficiencies, assembly bow
properties and fretting resistance.

New fuel types of commercial or technical interest are usually introduced as LUAs. The
objective with the LUA program is generally a complete licensing of the fuel type covering
possible future reloads.

125
A long term and standardized fuel assembly inspection program is used to verify performance
and for early detection of problems. The inspection program includes oxide thickness
measurements (5-10), assembly growth measurements (during unloading), visual inspection
and assembly/channel bow measurements. CRUD sampling are made every two years as a
part of the water chemistry follow-up program.

Ringhals along with the other Swedish plants participates in a long-term PIE program with
Studsvik. The main purpose of the program is to examine the burnup properties of various
types of cladding materials as well as primary failure mechanisms. A few fuel rods are sent
each year to the Studsvik facility for inspection. A typical PIE includes measurements of
oxide thickness, hydriding, creep and growth.

6. PERFORMANCE OF MODERN FUEL TYPES

To what extent have the fuel suppliers been able to make the fuel products more robust and
reliable given the lessons learned from older designs? To what extent have the operators been
able to introduce reliable fuel types and prevent debris from entering the coolant systems? Do
the most recent designs perform better then the previous designs?

Our expectations for modern fuel products is an average failure rate of one leaking fuel
assembly per 10 reactor cycles, corresponding to a failure rate of 3x10–6.

In general terms there has been an improvement of the leakage situation with the deliveries of
more advanced fuel types. This can be seen in fig.1. With the exception for Ringhals 2, the
cores are still a mixture of older and advanced fuel types. It is therefore too early to tell if the
goal of less than one failure per 10 reactor cycles can be met.

IFMs are being introduced on all PWRs for better DNB performance. During the transition
period there's a pressure difference between fuel with and without IFMs. Rod fretting wear
caused by this pressure difference has not been observed.

Fuel rod fretting wear hasn’t caused any failures in recent years but remains a latent threat.
Variations in the flow environment of the fuel as well as variations in the grids mechanical
properties related to burnup and manufacturing might create conditions that leads to fretting
wear. The design margins against fretting needs to be improved on new fuel types.

No debris failures have been found on fuel with debris filters so far. Introduction of more
efficient filters will further reduce the risks. All of the assemblies in the PWR cores have
debris filters.

7. FUEL FAILURES AND REMEDIES

7.1. Ringhals 1

Ringhals has adopted a defense in depth strategy (Section 8) to prevent secondary

degradation at Ringhals 1. A range of precautionary measures has been taken in order to
prevent debris from entering the coolant system during repair works and refueling. The
introduction of efficient debris filters on the fuel prevents debris induced fretting. The
cyclones installed in the feedwater pipes further separate debris from the coolant.

Activity monitoring with the FLEA code enables an early detection of failures. If a leaker is
detected, operating restrictions are imposed in order to delay the secondary degradation. If

126
the activity level reaches a threshold level, the plant is shutdown for replacement of the
leaking assembly.

If the plant is shutdown for removal of a leaking assembly, full core sipping is performed in
order to prevent reinsertion of undetected leaking assemblies.

The defense in depth strategy includes PIE at Studsvik for better understanding of the
mechanisms behind the degradation. In the technical assessment of new fuel types, the
cladding material’s secondary degradation properties are evaluated. Finally Vattenfall
participates in various R&D activities aimed at finding cladding materials with better
protection against secondary degradation. The water chemistry specifications ensure that the
cladding corrosion is not adversely affected by the water chemistry. The defense in depth
strategy is illustrated in table 2 and figure 1.

Table 2: Defense in depth matrix: For each main core component, two defenses are
introduced. The full defense includes 18 areas
Level of General design Failure mechanisms First defense Second defense
defense criteria
1 Core design a) Corrosion Chemistry restrictions Fuel inspection program c)
2 Assembly design b) Debris fretting Debris prevention Debris filters
3 Grid design Fuel rod fretting Flow loop testing d) Design margins
4 Fuel rod design PCI Power ramp restrictions Liner/Conditioning
5 Fuel pellet design Impurities/cracks Quality control Manufacturing procedures
6 Pellet matrix composition Secondary degradation Activity monitoring Operating strategies
Fission products

a) Includes burnup limitations and thermal margins.

b) Material composition, debris filters efficiencies, hold-down forces.
c) Includes leak testing and repair.
d) Includes operating experience.

Fuel failures at Ringhals 1989 - 2001

6
Number of leaking assemblies

R4: ?
2 R3: Manufacturing R4: Manufacturing

R1: Debris

0
89 90 91 92 93 94 95 96 97 98 99 00 01 02
Year

FIG. 1. Fuel failure development at Ringhals.

127
The 6 PCI failures on Ringhals 1, fig. 2, were the results of unconservative PCI thresholds in
control cells. More conservative PCI limits were therefore introduced in 1995. The
introduction of 10x10 liner fuel in 1997 has further improved the PCI margins. About 60% of
the fuel assemblies of the core have liner fuel.
Less than half of the fuel assemblies in Ringhals 1 have debris filter. Presently only 1/6th of
the core is replaced at each refueling. The small reload batches means that there’s a
considerable time lag before improvements are fully in effect.

Ringhals 1 - leaking fuel

4
Number of leaking fuel assemblies

PCI

Debris

2
Secondary degradation

0
89 90 91 92 93 94 95 96 97 98 99 00 01
Year

FIG. 2. Fuel failures at Ringhals 1.

Ringhals 1: Core fuel composition

100

90

SVEA 64
80 8x8

70

60
Fraction (%)

50
SVEA 96S/L
Liner
40
SVEA 96S
ATRIUM 10B
30 SVEA 100 10x10 + liner + debris filter

20

10

0
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Year

FIG. 3. Composition of the Ringhals 1 core.

128
The introduction of debris filter was delayed when rig testing showed that the filter in
question was subject to fretting wear at critical flow conditions.

The PCI failures in 1994 and the associated secondary degradation lead to a large increase in
the background activity levels. The present policy is to shut the plant down and replace the
leaking fuel if secondary degradation develops.

7.2. Ringhals 2

Only two failures have occurred in the period 1989–2001–both failures on fuel types without
debris filters. The Performance+ 15x15 fuel has not been subject to the kind of rod fretting
failures that have affected similar 17x17 Westinghouse fuel.

The risk of fretting wear caused by pressure drop between assemblies with and without IFMs
was a reliability concern during the introduction of IFMs at Ringhals 2. However, no fretting
failures have occurred on Ringhals 2. All assemblies of the core now have IFMs.

Assembly bow has emerged as a problem on Ringhals 2 during the last cycles. Bow
measurements made in the early 90-ties showed no indications of bow. Measurements made
during the most recent refuelings show the same bow magnitudes similar to those found at
Ringhals 3 and 4. The countermeasure has been to introduce top nozzles with a lower hold-
down force. The fuel supplier and Vattenfall have extensively analyzed the impact of the bow
on the safety analysis.

Ringhals 2 - leaking fuel assemblies

2
Number of leaking assemblies

Not repaired Debris

1

Old assembly Assembly without debris filter

0
89 90 91 92 93 94 95 96 97 98 99 00 01
Year

FIG. 4. Fuel failures at Ringhals 2.

7.3. Ringhals 3 and 4

More than half of all fuel failures in Ringhals during the period 1989–2001 has occurred in
Ringhals 3, or 20 out of 38 in total. The result is puzzling considering that the same fuel types
were delivered to Ringhals 4, manufactured in the same facilities. The core operating
conditions are also very similar.

129
Ringhals 3 and 4 have the same mix of fuel types as in Cattenom 3 (AFA 2G and AFA 3G).
Fuel inspection of individual rods in one AFA 2G assembly shows no indications of fretting
wear (12 ft cores as opposed to 14 ft core).

In addition to debris, also primary hydriding related to manufacturing has caused failures at
Ringhals 3. Grids with insufficient spring forces caused the fretting damages at Ringhals 3.
The problem was related to two reloads. A failed end plug weld caused one of the failures.

Ringhals 3 - leaking fuel assemblies

Corrosion
3
Primary hydriding

Debris
Fuel rod fretting
Number of leaking fuel assemblies

Not repaired

1
Manufacturing

0
89 90 91 92 93 94 95 96 97 98 99 00 01
Repaired, cause unknown Year

FIG. 5. Fuel Failures at Ringhals 3.

Ringhals 4 - leaking fuel assemblies

3
Number of leaking fuel assemblies

Debris

Repaired, cause unknown

Debris Manufacturing

0
89 90 91 92 93 94 95 96 97 98 99 00 01
Year

FIG. 6. Fuel failures at Ringhals 4.

130
8. DEFENCE IN DEPTH

The core region might be partitioned in logical units in the following way. The core is
composed of fuel assemblies that in turn comprises a skeleton/grid structure with suspended
fuel rods. The fuel rods encapsulate fuel pellets. The pellet’s matrix finally encapsulates the
fission products. Qualitatively, therefore, six different shells surround the fission products. To
prevent the fission products from escaping, the integrity of all six shells must be maintained.
This is illustrated in figure 7. The integrity requirements are specified as general design
criteria.

Fuel failure mechanisms tend to appear at the interfaces between the shells. Main failure
mechanisms are corrosion, debris fretting, fuel rod fretting, PCI and primary hydriding. The
probabilistic nature of these mechanisms makes it difficult to establish simple design criteria.

Table 2 describes the defense in depth strategy to prevent failures. Fuel failures are prevented
by the general design criteria applicable to all six layers of defense. In addition, two defensive
measures are introduced for each failure mechanism at the interface between the main core
components - in all 18 areas of fuel failure defense.

Main components
Interfaces
(Areas)
(Lines)
Core
Corrosion
Fuel assembly
Debris
Grid/Skeleton
Rod fretting
Fuel rod
PCI
Fuel pellet
Impurities/cracks
Matrix
Activity release
Fission products

FIG. 7. The defense in depth strategy requires that attention is paid to the interfaces between the
main core components where failures tend to appear. The fuel cladding is exposed to all main
components.

9. HOT CELL EXAMINATIONS AT STUDSVIK

Fuel rods are regularly sent to the hot cells at Studsvik for root cause examination of failed
rods and for more general examination within the long term R&D program.

Hot cell examinations at Studsvik verified the PCI failures at Ringhals 1 as well as the
primary hydriding failures at Ringhals 3. Hot cell examinations have also been used to
investigate the corrosion properties of advanced alloys such as M5 and ZIRLO as well as
creep and growth rates.

The secondary degradation failure mechanisms have been investigated thoroughly at

Studsvik.

Valuable results have also been gained by ramp testing of various fuel rods at Studsvik.

131
10. ASSEMBLY BOW

The most demanding problem related to fuel reliability has been the assembly bow problem in
the PWRs. All PWRs are affected by assembly bow.

Fuel types with thicker guide thimbles, reinforced dashpot and reduced hold down spring
forces have been introduced in order to reduce the bow. In addition, materials with a lower
growth rate and better creep properties have been introduced. As a result of the mechanical
modifications, we see little impact of the bow on control rod friction and drop time of the
bow.

On Ringhals 3 and 4 we have seen a gradual reduction of the bow. Last year, however, we
saw an increased bow on Ringhals 3. It remains to bee seen how effective the
countermeasures are in the long term.

Extensive studies performed by our fuel supplier and by Vattenfall have shown that the
assembly bow causes additional uncertainties in the peaking factors of around 10%. These
additional uncertainties have been accounted for in the transient analysis. Some of the
previously mentioned spare design margins have been utilized to account for assembly bow.

To verify the safety limits, the assembly bow is measured on 10-20 assemblies during each
refueling. The maximum bow is verified by the measurements to be less than the largest bow
assumed in the safety studies.

In addition to the mechanical forces giving rise to bow, there are measurements as well as
calculations showing that the coolant flow through the core gives rise to bow forces. Various
forms of bow management or reshuffling strategies are discussed to counteract the bow.

11. PRODUCT LIFE CYCLES

New fuel types generally have better thermal mechanical properties than their predecessors.
Improved thermal performance translates to better safety margins if properly managed. The
dilemma is to avoid "children's diseases" i.e. undetected reliability problems. The dilemma is
illustrated in figure 8.

12. COOLANT ACTIVITY ANALYSIS AND LEAK TESTING

On-line leak testing of the whole core is generally performed irrespective of the activity levels
prior to shutdown. On-line sipping is reliable, efficient and does not impact critical path
during outage. Wet sipping is used to verify fuel integrity after fuel repair. Full core leak
testing eliminates the risk of reloading leaking fuel.

Activity analysis is a sensitive instrument for detection of small failures, the outset of
secondary degradation and for estimation of the amount of tramp uranium in the system.

Secondary degradation in the BWR in most cases requires shutdown and replacement of the
leaking fuel.

132
 Number of assemblies Product life cycle
delivered

Proven design?
Low performance/High reliability

Introduction

Development High performance/Uncertain reliability

Time
FIG. 8. Product life cycle.

13. CONCLUSIONS

‰ Prevention of secondary degradation requires a defense-in-depth strategy.

‰ Flow testing of fuel prototypes can be made more efficient.

‰ The design margins to fuel rod fretting needs to be improved on new fuel types.

‰ Fuel failures related to manufacturing indicate a need to further improve the quality
control during manufacturing.

‰ The mechanisms behind assembly bow must be better understood. Bowing adds to the
peaking factor and fuel duty uncertainties and hence might impact fuel reliability.

‰ A long-term fuel inspection program combined with PIE makes it possible to detect
deviations from expected performance. PIE is essential for root cause investigation.

‰ A conservative management of thermal and mechanical margins is required in order to

improve both reliability and fuel economy.

‰ Efficient activity analysis is necessary in order to detect small leakers and the outset of
secondary degradation.

‰ Full core leak testing eliminates the risk of reloading leaking fuel

133
EXPERIMENTAL STUDIES OF FUEL FAILURE AND
DEGRADATION MECHANISMS
(Session 3)
PWR FUEL FAILURE ANALYSIS DUE TO HYDRIDING BASED ON PIE DATA

YONG-SOO KIM
Department of Nuclear Engineering, Hanyang University,
Sungdong-Ku, Seoul, Republic of Korea

Abstract

Recently failures of nuclear fuel rods in Korean nuclear power plants were reported and their failure
causes have been investigated by using PIE techniques. Destructive and physico-chemical
examinations reveal that the clad hydriding phenomena had caused the rod failures primarily and
secondarily in each case. In this study, the basic mechanisms of the primary and the secondary
hydriding failures are reviewed, PIE data such as cladding inner and outer surface oxide thickness and
the restructuring of the fuel pellets are analyzed, and they are compared with the predicted behaviors
by a fuel performance code. In addition, post-defected fuel behaviors are reviewed and qualitatively
analyzed. The results strongly support that the hydriding processes, primary and secondary, played
critical roles in the respective fuel rods failures and the secondary hydriding failure can take place
even in the fuel rod with low linear heat generation rate.

1. INTRODUCTION

One of the frequent fuel failures in light water reactors is hydrided or hydriding-related
failure [1,2]. When a reactor fuel element is defective during operation, cladding no longer
provides a barrier between the fission-induced burning fuel element and the coolant. The
presence of this defective fuel may present economic penalties to the power utility.

Sometimes severe secondary damage can follow the primary failure. Existing leak path can
let the coolant enter the element, the coolant flashes into steam, and then complicated
processes such as steam oxidation of UO2, oxidation and hydriding of the clad inner surface,
restructuring of UO2, are developed, ending up with catastrophic failure [3]. Recent
investigations have provided a better understanding of the failure processes, physical and
chemical [4,5].

Fuel rod failure were recently reported in the two different Korean nuclear power plants. Two
rods were failed during cycle 1 start-up operation in the prototype Korean standard nuclear
power plant and one rod was failed during cycle 7 reload core operation in the other plant.
Through the post irradiation examination it was revealed that the causes of the rod failures
are ascribed to the primary and secondary hydriding. These findings are supported by the
mechanistic review of the hydriding failures, fuel behaviors predicted by a fuel performance
code, and qualitative post-defected fuel behaviour analysis.

2. POST-IRRADIATION EXAMINATION

Failed fuel rods were transported from the storage pool in the plant sites to the PIE facility in
KAERI. The examinations, non-destructive, destructive, and chemical, were carried out in the
hot cells. Whole rods were thoroughly inspected using telescope, gamma-scanned, and the
diametral change of the cladding was measured in the non-destructive test. Then the oxide
thicknesses of clad inner and outer surface and micro-structural change of UO2 pellet were
examined. In addition, axial hydrogen content distribution of the clad was determined by
using LECO analyzer.

137
Through PIE, it is found that D103-K2 rod was failed due to the random internal hydring
while in B208-R8 rod through-wall defect was developed by the debris caught in the bottom
grid and the defect gradually induces the breach of the cladding due to the secondary massive
hydriding. These two rods were failed during cycle 1 start-up operation in the prototype
Korean standard nuclear power plant. With the design calculation in detail and the gamma
scanning measurements, the burn-ups and the ratios of q’/q’core avg for both rods were
determined: 1875 MWD/MTU and 1.27 for D103-K2 rod, and 2013 MWD/MTU and 1.38 for
B208-R8, respectively.

Figure 1 shows the visual inspection results of the damaged D103-K2 rod due to hydride
blister and hydride-induced crack. Figure 2 renders the cross sectional view of the
microstructural change of UO2 and the breached cladding of the rod. As shown in the middle
micrograph of Figure 2, bright spots in the fuel pellet are observed in an annular band at
around r = 0.6R. Extensive investigation reveals that the annular band of the bright spots are
observed at about the elevation (1200~2000mm) with high linear power and possibly under
the highly oxidizing environment. Thus the spots were closely examined and the micro-
hardness was measured in the radial direction (Figure 3). It is found that extra-ordinary
columnar grain growth took place in the annular band.

Figure 4 shows the visual inspection results of the secondarily damaged B208-R8 rod due to
massive hydride formation. Figure 5 provides the cross sectional view of microstructural
change of the fuel pellet and the breached cladding of B208-R8. In this rod the annular band
of bright spots were also observed at the high linear power elevation even though they are not
as clear as those of D103-K2 rod

The third fuel rod for which PIE was carried out is the J09-L01 rod burned up during cycle 7
reload core operation in another plant. The PIE results are shown in Figure 6. It is obviously
seen that this rod was also damaged secondarily owing to the massive hydriding. However,
this result is very unexpected because q’/q’core avg and dischge burnup of the rod are found to
be only 0.66 and 11806 MWD/MTU, respectively, based on the design calculation in detail
and the gamma scanning measurements. In fact, the rod was located outmost periphery of the
reactor core during the operarion, facing the baffle of the reactor vessel.

FIG. 1. Visual Inspection of D103-K2 Fuel Rod.

138
FIG. 2. Micro-structural Change of UO2.

139
 FIG. 3. Columnar Grain Growth in Steam Oxidation Environment.

FIG. 4. A View of Hydride Failure on B208-R8 Cladding Tube (2,660mm).

3. RESULTS AND DISCUSSION

3.1. Internal Hydriding Failure of D103-K2

Hydrogenous impurities inside a fuel rod will ultimately hydride the Zircaloy cladding,
regardless of their initial chemical state. Massive localized hydriding leads to the hydride
blisters where the volume change is visually evident on the outside of the fuel rod, to the
serious deterioration of the mechanical properties of the clad so that splits can easily develop,
and eventually to the perforation of the clad after local breakthrough [3]. The main source of
the hydrogen in the typical hydriding rod failures is the residual moisture in the UO2 fuel
pellets [1].
In the primary hydriding process, the residual moisture (steam) oxidizes the inner surface of
the clad. Thus the thickness of oxide, over the whole length of the rod, generally follows the
axial temperature profile since the Zircaloy oxidation is an activated process thus the
thickness depends on the temperature. Figure 7a) and 7b) show that in the D103-K2 rod the
inner and outer oxide thickness profiles are generally in good agreement with the temperature
profiles predicted by the best estimate fuel performance code, ESCORE [6]. The code
simulation is based on the nuclear design report for the fuel cycle of the unit [7].

140
FIG. 5. Micro-structural Change of UO2.

141
 FIG. 6. PIE Results of J09-L01 Rod.

142
 FIG. 7. Comparison Between the PIE Results and Code Prediction (D103-K2 Rod).

High temperature inside the fuel pellet, especially during start-up operation, and oxidizing
environment in the moisture-rich environment induce fast restructuring of the pellet micro-
structure, i.e., grain growth and stoichiometry changes. Through PIE these phenomena are
observed at the elevation of high linear heat generation rate of the rod, which is consistent
with the fuel centerline temperature profile predicted by the code (Figure 7c). The extra-
ordinary columnar grain growth is discussed in Section 3.3.

3.2. Secondary Hydriding Failure of B208-R8 and J09-L01 Rod

Occasionally, small primary defect leads to heavy secondary hydride formation [8,9]. During
the initiation stage, coolant enters the fuel rod through the defect and flashes into steam. Once
the internal and external pressures have equalized, steam oxidizes the internal Zircaloy
cladding surface into ZrO2, resulting in the release of hydrogen. As the gases react in the fuel-
cladding gap, the concentration of hydrogen continuously increases while steam is depleted.
Thus, the ratio of H2/H2O increases rapidly in the gap. In the secondary hydriding process,

143
molecular diffusion of the steam in the gap is required, that is, the steam must be supplied
from the primary defect site along the rod axis for the continues oxidation. If the ratio of
H2/H2O exceeds a certain critical value at a certain elevation and the conditions for the
accelerated hydriding is achieved in the region, such as damaged protective oxide and/or
flawed surface, then the massive hydriding can instantaneously take place thus breaches the
mechanically-degraded cladding [4,5].

In general, therefore, it is believed that the inner surface oxide is thickest at the primary defect
site and gradually decreases when it gets far from the primary defect point. However, it is not
true if temperature is varying along the rod since the temperature plays the most significant
role in the Zircaloy oxidation kinetics. In the failed fuel rod B208-R8, fuel performance code
predicts the cladding inner surface temperature profile which has a maximum about mid-point
(Figure 8a). Therefore, instead of gradual decrease from the primary defect site, the thickness
of the inner surface oxide is increasing, following the temperature profile (Figure 8b). The
oxide profile and the results of the restructured pellet examined in the PIE are consistent with
the predicted axial power profile of the rod.

As previously discussed, it is known, in general, that the hydriding failure, primary or

secondary, takes place in the fuel rod with high linear heat generation rate since rapid
oxidation and hydriding reaction of Zircaloy cladding are required for abundant hydrogen
production in the gap between cladding and pellet. However, the PIE results of J09-L01 rod
reveal that this rod was damaged secondarily due to the massive hydriding even though its
linear power was only 66% of the core average. This implies that in the defective fuel the
ambient environment is also very crucial as well as the fuel temperature.

FIG. 8. Comparison Between the PIE Results and Code Prediction (B208-R8 Rod).

144
3.3. Enhanced Oxidation and Extra-ordinary Columnar Grain Growth

In Figures 7 and 8 it is seen that the inner surface oxide thicknesses of the two rods are
generally in good agreement with the temperature of the surface predicted by the ESCORE
code. However, close and careful examination reveals that the oxide in the neighborhood of
the defect is thicker than that in other regions. The cause of the oxidation enhancement is
believed to be due to the hydride precipitation in the Zircaloy matrix during the oxidation
reaction.

Now it is well-known that the zirconium alloy oxidation kinetics is enhanced when the
hydride precipitation takes place in the metal-oxide interface [10,11]. Recently it is
demonstrated that the oxidation can be accelerated under the hydrogenous environment with
high H2/H2O ratio even without the hydride precipitation in the interface, if the oxidation and
hydriding reactions take place simultaneously (Figure 9) [12]. Thus, the unusual oxide
thickness near the defect seems be ascribed to the simultaneous reactions in the gap between
the cladding and the pellet during the internal hydriding process.

FIG. 9. Zircaloy Oxidation Enhancement in Hydriding Environment.

It is understood that, as the reactions goes on in the gap, steam is depleted and hydrogen
builds up rapidly at the high temperature spot because the oxidation kinetics is faster at the
high temperature spot than that at low temperature region. This leads to further oxidation,
more hydrogen production, and further oxidation again, ending up with the enhanced
oxidation and massive hydride formation and the eventual breach of the spot.

The observed remarkable columnar grain growth in the pellet is very unusual because UO2
grain growth kinetics is not fast enough to grow the grain size up to 200 µm under the
temperature predicted by the code. However, when the thermo-physical property changes of
defected fuel and thermal conductivity degradation of the gap are taken into consideration, the
extraordinary growth is understandable.

During the internal or secondary hydriding process, water entering the fuel rod turns into
steam and thus the fuel pellet-to-cladding gap is quickly filled with steam. The thermal
conductivity of steam is less than 10% of the helium filling gas, thus, the temperature

145
difference in the gap drastically increases, leading to the abrupt fuel temperature increase. In
addition, it has been reported that the thermal conductivity of UO2 is degraded in the
hyperstoichiometric transition during the steam oxidation 13,14]. Therefore, the increase of
the fuel pellet temperature may be at least a few hundreds degree, which makes the grain
growth fast.

In the mean time, steam oxidation of UO2 start to take place from the pellet outer surface and
the oxygen begins diffusing towards the center, which induces the oxygen concentration
gradient. A few reports are available that under the oxidative environment, stoichiometry of
the UO2 increases and this induces the rapid grain growth [15,16].

Thus, in the defective fuel, once the gap is filled with steam fuel temperature goes up thus the
centerline temperature of the fuel pellet also increases. This temperature is believed not to be
high enough for the grains of the stoichiometric pellet to grow fast, however, high enough for
the grain growth of the oxidizing fuel pellet in transition to hyperstoichiometry.

Therefore, on summarizing these property changes, we can explain the extra-ordinary

columnar grain growth. If in a certain circumferential region the fuel temperature is high
enough for the grain growth of a hyperstoichiometric UO2 which can be achieved by the
steam oxidation in the gap, the remarkable grain growth can take place in an annular band. If
the temperature of the pellet is too low or the gap does not provide the oxidizing environment,
any noticeable grain growth never occurs even in the local spot.

4. CONCLUSIONS

PIE data analyses reveal that fuel rod D103-K2 was failed by internal hydriding and the fuel
rods, B208-R8 and J09-L01 with non-hydriding primary defects, were additionally damaged
by secondary hydriding. These results are generally in good agreement with fuel performance
code predictions.

Careful observation of the oxide thicknesses of clad inner surface and microstructural change
of the fuel pellet raises some questions on the oxidation enhancement and extra-ordinary large
columnar grain growth in the pellet. The unusual oxide thickness near the defect is ascribed to
the hydride precipitation in the zirconium matrix, which accelerate the oxidation.

The extra-ordinary large columnar grain growth can be explained in terms of the fuel pellet
temperature increase, the local stoichiometry change of the pellet, and the fast grain growth of
the UO2 in transition to hyperstoichiometry.

It is reported that secondary failure can take place in the rod even with low linear power
because the oxidizing environment in a defected rod plays the most significant role in the
thermo-physical properties changes. Finally, PIE data support the proposed mechanism for
the defective fuel behaviors such as hydriding-enhanced corrosion and the secondary
hydriding.

ACKNOWLEDGEMENT

This project was supported by Ministry of Science and Technology (MOST) as a part of the
nuclear R&D programs.

146
 REFERENCES

1. F. GARZAROLLI, R. von JAN, and H. STEHLE, The Main Cause of Fuel Element Failure
in Water-Cooled Power Reactors, Atomic Energy Review, 17 (1979) 31.
2. S.M. STOLLER Co., Nuclear Unit Operating Experience, EPRI NP-5544 (1987).
3. D.H. Locke, Review of Experience with Water Reactor Fuels in 1968-1973, Nucl. Eng.
Desg., 33 (1975) 94.
4. S. VAKNIN and D.R. OLANDER, Secondary Hydriding of Defected Zircaloy-Clad Fuel
Rods, EPRI RP-1250-23 (1972)
5. J.C. CLAYTON, Internal Hydriding in Irradiated Defected Zircaloy Fuel Rods, Zirconium
in the Nuclear Industry: Eighth International Symposium, ASTM STP 1023 (1989) 266.
6. Combustion Engr. Inc., ESCORE-the EPRI Steady State Core Reload Evaluator Code,
EPRI NP-5100 (1987).
7. KAERI and ABB/CE, Nuclear Design Report for Yonggwang Unit 4 Cycle 1, Rev. 0
(1995).
8. A. JONSSON, L. HALLSTADIUS, B. GRAPENGIESSER, and G. LYSELL, Failure of A
Barrier Rod in Oskarshamn 3, Fuel for the ‘90’s: Proc. ANS/ENS International Topical
Meeting on LWR Fuel Performance, Avignon, France, Vol. 1 (April 1991) 371.
9. J.H. DAVIES and G.A. POTTS, Post-defect Behavior of Barrier Fuel, Fuel for the ‘90’s:
Proc. ANS/ENS International Topical Meeting on LWR Fuel Performance, Avignon, France,
Vol. 1 (April 1991) 272.
10. A.M. GARDE, Enhancement of Aqueous Corrosion of Zircaloy-4 Due To Hydride
Precipitation at the Metal-Oxide Interface, Zirconium in the Nuclear Industry: Ninth
International Symposium, ASTM STP 1132 (1991) 564.
11. M. BLAT and D. NOEL, Detrimental Role of Hydrogen on the Corrosion Rate of
Zirconium Alloys, , Zirconium in the Nuclear Industry: Eleventh International Symposium,
ASTM STP 1295 (1996) 319.
12. S. KIM and Y. KIM, Oxidation vs. Hydriding of Zircaloy-4, J. Nucl. Mater. (to be
published).
13. B.J. LEWIS, F.C. INGLESIAS, D.S. COX, and E. GHEORGHIU, Nucl. Technol. 92
(1990) 353.
14. H. STEHLE, H. ASSMANN, and F. WUNDERLICH, Uranium Oxide Properties For
LWR Fuel Rods, Nucl. Eng. Desg., 33 (1975) 230.
15. I. AMATO, R.L. COLOMBO, and A.M. PROTTI, J. Amer. Ceram. Soc., 46 (1963) 407.
16 H. ASSMANN, W. DORR, and M. PEEHS, Control of UO2 Microstructure By Oxidative
Sintering, J. Nucl. Mater., 140 (1986) 1.

147
OUTSIDE-IN FAILURE OF BWR SEGMENT RODS DURING
POWER RAMP TESTS

H. HAYASHI, Y. ETOH , Y. TSUKUDA

Nuclear Fuel Department, Nuclear Power Engineering Corporation,
Tokyo, Japan

S. SHIMADA
Global Nuclear Fuel-Japan Co., Ltd,
Ibaraki-gun, Ibaraki-ken, Japan

H. SAKURAI
Nippon Nuclear Fuel Development Co., Ltd,
Ibaraki-gun, Ibaraki-ken, 311-1313 Japan

Abstract

BWR 8x8 fuel assemblies with segmented rods (Step II LUAs) were irradiated up to 5 cycles in
Fukushima Daini Nuclear Power Station No. 2 Unit. The ramp tests of 25 segments were conducted in
Japanese Material Test Reactor. One segment rod irradiated for 3 cycles (43 GWd/t) failed by a single
step ramp test after 9 minutes at terminal ramp power of 614 W/cm with a pinhole due to PCI/SCC.
One segment irradiated for 4 cycles (56 GWd/t) failed by a single step ramp test after 149 minutes at
551 W/cm with an outer side axial crack. Among 9 segments irradiated for 5 cycles (61 GWd/t) two of
them failed by a single step ramp test after 100 and 68 minutes at 421 and 428 W/cm, respectively and
one of them failed at 446 W/cm by a stair ramp test with outer side axial crack. The decrease of the
failure threshold for higher burnup segment rods is clear and the failure mode of higher burnup
segment rods is different from that of low burnup (less than 43 GWd/t) segment rods. This new failure
mode is caused by the combination of the high stress and the radially oriented hydride precipitated
during ramp test. Through detailed PIEs before and after ramp tests, following characteristics on failed
segment rods were observed. Hydrogen contents in the cladding tubes increased with burnup and
exceeded solubility limits during base irradiation of segment rods. Radial hydrides were observed at
the outer rim of the cladding tubes irradiated for 4 and 5 cycles and ramp-tested rods, while few radial
hydrides were observed before ramp tests. Crack started from the outer surface of the cladding tube
and propagated to inside and axial directions and finally failed by ductile fracture. The fracture surface
showed brittle features developed by the combination of the stress due to PCMI and hydride in the
cladding.

1. VERIFICATION TEST ON BWR HIGH BURNUP FUEL

Under the sponsorship of the Ministry of Economy, Trade and Industry (METI), Nuclear
Power Engineering Corporation (NUPEC) has completed the fuel irradiation test program
̌The Verification Test on BWR High Burnup Fuel". The objectives of this program were to
verify fuel integrity and to study fuel behaviors of high burnup 8x8 fuel (BWR Step II Fuel)
at high burnup [1], [2], [3], [8].

Step II fuel has an 8˜8 lattice configuration of 60 fuel rods in which Zr-lined and high
corrosion resistant cladding tubes are used. Its design maximum bundle burnup is 50GWd/t.
The bundle design has a central large diameter water rod to soften the neutron spectrum. It


Present address: Global Nuclear Fuel-Japan Co., Ltd,
2-3-1 Uchikawa Yokosuka Kanagawa-ken, 239-0839 Japan.

148

uses ferrule-type spacers in place of the earlier grid-type ones to improve thermal hydraulic
performance. Furthermore, some design improvements to reduce fission gas release, such as
an increased initial He pressure, increased pellet density and decreased gap width between
pellet and cladding, are incorporated. The design parameters and fuel rod arrangement in the
fuel assembly are shown in Figure 1 [1], [8].

Figure 1. Design Parameters and Rods Arrangement.

Figure 2. Irradiation History of Step II luas.

Eight lead use assemblies (LUAs) equipped with segment rods were loaded in a relatively
high power and symmetric position in the core of Fukushima Daini Nuclear Power Station No.
2 Unit, operated by Tokyo Electric Power Co. They were irradiated up to 5 irradiation cycles
under normal BWR conditions and one assembly was discharged after each irradiation cycle
for detailed PIEs. The reactor was operated at almost full power through the five cycles.
Irradiation history of LUAs is shown in Figure 2. Maximum linear heat rates of LUAs were
kept above 300 W/cm in the first cycle, above 250 W/cm in the second and third cycles and

 149
decreased to 200 W/cm in the fourth cycle and 80 W/cm in the fifth cycle. The integrity of
high burnup 8x8 fuel was confirmed up to the bundle burnup of 48 GWd/t after 5 cycles of
irradiation.

After each of the reactor cycles, LUAs were subjected to basic non-destructive examinations
in the spent fuel pool, including visual examination with an underwater TV camera. Five
LUAs have been examined in the hot laboratory of Nippon Nuclear Fuel Development Co.,
Ltd. (NFD) and detailed PIE data were systematically obtained. The burnups of these five
LUAs were about 13, 24, 35, 44 and 48GWd/t, respectively. At each burnup stage,
non-destructive tests (NDT) were executed on principally 16 rods and destructive tests (DT)
were performed on 2 or 3 rods, including a (U,Gd)O2 rod.

In order to confirm both the integrity of fuel rods during normal/off-normal operational
transients and the margin of ramp performance, power ramp test series were executed using
25 segment rods irradiated for three to five cycles (burnup range: from 43 to 61GWd/t).

The ramp sequences are a stair case power ramp (Ramp sequence A), a single step power
ramp (Ramp sequence B), and a power cycling test (Ramp sequence C). Ramp sequence A is
used to calibrate heat generation by He-3 gas pressure and to determine power to failure.
There are three types of ramp sequence B, that is, Bs, Bt and Br. Ramp sequence Bs is
employed to confirm the integrity of fuel during normal operational transients as is ramp
sequence C also. In this case, ramp terminal power (RTP) corresponds to the designed
maximum power level. Ramp sequence Bs is also employed to investigate the margin of ramp
performance by increasing RTP up to about 150% of the designed maximum power level. The
integrity of fuel rods under abnormal operational transients, such as control rod withdrawal, is
studied by ramp sequence Bt in which RTP and the holding time are selected as about 125%
of the designed maximum power level and 10 minutes, respectively [2]. RTP and the
cumulative holding time of ramp sequence Br are the same as those of ramp sequence C, so it
is possible to obtain the characteristics of fuel behavior during cyclic operation by comparing
with ramp sequence Br. Four typical power ramp sequences (two Bs, Bt and C) are shown in
Figure 3.

Figure 3. Power Ramp Sequence.

150

2. CHARACTERISTICS CHANGES OF HIGH BURNUP FUELS

Cladding water-side corrosion, resulting hydrogen uptake is considered to be a key factor for
PCMI failures at high burnup. Corrosion and hydrogen uptake behaviors are examined in
detail.

Visual appearances and cross-sectional micrographs of the cladding are shown in Photo 1.
Two types of corrosion were observed. Nodular corrosion was clearly observed at the gas
plenum region, although the oxide layer of fuel stack region was uniform and the thickness
was 10 - 20 micrometers in the transverse sections of each fuel rod, as shown in Photo 1.

Burnup dependence of maximum oxide thickness in fuel stack regions of Step II LUAs is
shown in Figure 4, which also includes those of 8x8 fuels and Step I LUAs [3], [4], [8]. In
this figure, oxide thickness obtained by metallography is plotted as a function of specimen
burnup. Maximum oxide thickness of Step II LUAs was less than one half of 8x8 fuels
throughout their irradiation cycles.

Photo 1. Visual Appearance and Cross Sections of Oxides on Fuel Rods.

 151
 Figure 4. Burnup Dependence of Maximum Oxide Thickness.

Figure 5. Hydrogen Content in Cladding v.s. Irradiation Period.

The content of hydrogen in cladding was measured by an inert gas fusion method. Hydrogen
content of BWR fuel cladding is shown in Figure 5 as a function of irradiation period.
Maximum hydrogen content of Step II LUA cladding was less than that of 8x8 fuels up to
three irradiation cycles, but it tended to increase with an increase in irradiation period
thereafter, and after five irradiation cycles, the content was almost the same as those of the
reported.

Hydride morphologies in the cross-sectional metallography of the cladding are shown in

Photo 2. Hydrides in the cladding irradiated for 3 cycles are small and distribute uniformly.
Hydrides increase with burnup after 4 cycles of irradiation and the number densities of
hydride in outer peripheral and Zr-liner are high [8].

152

 Photo 2. Hydride Morphology of Step II Fuel Cladding.

With increasing burnup, pellet-cladding gap in fuel rods tends to close due to pellet swelling
and cladding creep-down, and eventually a bonding layer is formed between the pellet and
cladding under contact conditions at high burnups. The bonding layer was observed for Step
II LUAs after three cycles of irradiation. The thickness of the layer is about 10 – 20 Ǵm and
remained almost constant in spite of the increase in burnup. Considering EPMA measurement
results at the pellet-cladding interface and published information [5], this layer may be
produced by subsequent mutual diffusion of UO2 and ZrO2 [8].

3. FUEL RODS FAILURE DURING POWER RAMP TESTS

A series of ramp tests for 25 segment rods of burnup ranging from 43 to 61 GWd/t was
carried out in JMTR using the Boiling Water Capsule under simulated BWR temperature and
pressure conditions. The ramp test results are shown in Figure 6 with published data [2], [3],
[4], [8].

Figure 6. Power Ramp Test Results for Zr-lined Fuel Rods.

 153
Five segment rods failed during the power ramp tests. The appearances of failed segment rods
are shown in Photo 3. One segment rod irradiated for 3 cycles failed at RTP of 610 W/cm by
PCI/SCC mechanism as shown in Photo 4, while four segment rods irradiated for 4 and
5 cycles failed at about 550 and 420 W/cm, respectively and had axial cracks starting from the
outer surface of cladding tubes.

Photo 3. Visual Appearance of Failed Segment Rods.

Photo 4. SEM Image and Cross-sectional Metallography of Failed Segment (3cy. Irradiated).

154

Following characteristics on failed segment rods were observed through detailed PIEs before
and after ramp tests.

The cross-sectional micrographs at four elevations of the through wall crack of the segment
rod irradiated for 5 cycles are shown in Photo5. The shape of cross section near the center of
the axial crack is perpendicular to the outer surface of the cladding tube and the failure mode
is strongly brittle. At the upper part of the crack, there is a small brittle portion in the outer
rim of the cladding tube and a large ductile area in the middle part [8].

Photo 5. Visual Appearance and Cross Sectional Metallograpy of Failed Segment Rod.

Radial hydrides are observed at the outer rim of the cladding tubes in Photo 6 of which was
taken from the failed segment rod during power ramp test after 5 cycles of irradiation. The
length of radial hydrides is about 70 micrometers. These radially-oriented hydrides were
observed on all rods failed by ramp-tests after 4 and 5 cycle irradiations, and the length of the
hydrides depended on RTP. No radially-oriented hydrides were found on rods before ramp
tests [8].

Photo 7 shows the SEM images of fracture surface. The fracture surface is composed of three
regions. The region near outer surface is strongly brittle, but that of the near inner surface is
ductile while the middle region is macroscopically brittle but microscopically the mixture of
brittle and ductile features. The width of strongly brittle region is about 50 to 80 micrometers
and that of macroscopically brittle crack is about 700 micrometers from outer surface. The
width of strongly brittle region is close to the length of radial hydrides at the outer rim of the
cladding tubes. In Photo 8 two pairs of the two opposing fracture surface are shown. It is
evident that there is a good fit between the two sides of the crack, i.e., a “hill” on one fracture
surface corresponds to a “valley” on the matching fracture surface [6], [8].

 155
 Photo 6. Hydride Distribution in Cladding of Failed Segment Rod.

Photo 7. SEM Images of Fracture Surfaces.

156

 Photo 8. SEM Images of two Matching Fracture Surfaces.

The cross-sectional micrographs of another crack sample from the segment rod irradiated for
5 cycles are shown in Photo 9. There are many hydrides perpendicular to fracture surface.
These hydrides are also seen in Photo 6.

In order to get the root cause of the crack from the outer surface of the cladding tube, many
metallurgical observations were done, from three different directions, i.e. axial, radial and
circumferential directions. Photo 10 shows the view from axial direction at non-penetrated
cracks. There are many hydrides perpendicular to the crack on both sides and some hydrides
are gathering at the crack tips. There is a small hydride just at the crack tip as shown in Photo
11. View from radial direction close to the cladding outer surface at the axial ends of crack
indicates a nest of hydrides at the small crack tip as shown in Photo 12. These photos tell us
the crack propagation is strongly connected with hydrides gathering at the crack tip.

4. FAILURE MECHANISM [5], [6], [7], [8]

By the detail PIE and the analysis we estimated the failure mechanism by axial crack would
be as follows.

 157
 Photo 9. Hydride Metallography at the Crack.

Photo 10. Hydride Distribution at Non Penetrated Cracks.

158

 Photo 11. Hydride Observed at Crack Tip (Radial Cross Section).

Photo 12. Hydride Observed at Crack Tip (Tangential Cross Section.

 159
 Photo 13. Hydride Distribution in Cladding of Base Irradiated Fuel Rod and Segment Rod.

Hydrogen contents in the cladding tubes increased with burnup as shown in Figure 5 and
exceeded solubility limits of cladding material after 4 and 5 cycle of base irradiations. Photo
13 shows the distribution of hydride in the cladding tubes irradiated 4 and 5 cycles. In the
outer region of the cladding tube there are many circumferentially oriented hydrides. Inner
portion of the cladding tube, Zr liner has more hydrides due to lower solubility than
Zircaloy-2. Estimated relative hydrogen content distributions are shown in Figure 7. These
hydrogen contents were evaluated using the densities of hydrides in the micro photos.

Figure 7. Hydride Distribution in Cladding after Base Irradiation.

160

During power ramp tests, some hydrogen diffused from inner to outer region in the cladding
tube due to temperature gradient. The distribution of hydrogen solubility limit in the cladding
tube is so steep as shown in Figure 8 under power ramp conditions. Diffused hydrogen
precipitated radially at the outer region due to the combined effects of relatively lower
solubility and high tensile stress in the cladding caused by PCMI.

Figure 8. Hydrogen Solubility Limits in Cladding.

Radial hydride was cracked by high stress or strain by PCMI.

This is the initiation of the failure due to crack and the schematic diagram is shown in
Figure 9(1).

When initial crack formed during the power ramp test, hydrogen concentrated at the crack tip
due to stress concentration and small hydride was formed as shown in Photo 10 and Photo 12.
These small hydrides at the crack tip cracked and formed “hill” and corresponding “valley”
on matching fracture surface as shown in Photo 8 by the stress due to PCMI. Then fresh crack
tip is formed as shown in Photo 11.

These stress concentration, hydrogen diffusion, hydrogen precipitation and the cracking of
hydride repeated and the crack propagated. This is the propagation mechanism of the crack
and the schematic diagram is shown in Figure 9(2).

When stress exceeded the yield strength of residual cladding thickness, ductile failure
appeared.

 161
 Figure 9. Schematic Diagram of Crack Initiation and Propagation.

5. SUMMARY

In this verification test program, well-characterized LUAs of high burnup 8x8 fuel were
irradiated in a typical commercial BWR in Japan, Fukushima Daini Nuclear Power Station No.
2 Unit, under normal operational conditions. After each irradiation cycle, detailed PIEs at a
hot laboratory and power ramp tests on segment rods, which were installed in LUAs, were
carried out.

As the results of these examinations, some segment rods irradiated for 4 and 5 cycles failed at
about 550 and 420 W/cm, respectively, during power ramp tests and showed axial cracks
initiated at outer surface of cladding tubes.

162

Estimated failure mechanism by axial crack can be summarized as follows:

1) Hydrogen contents in the cladding tubes increased with burnup and exceeded solubility
limits during base irradiation of segment rods.
2) During power ramp tests, some hydrogen diffused from inner to outer region in the
cladding tube due to temperature gradient. Diffused hydrogen precipitated radially at the
outer region due to coupled effects of high hydrogen content and high tensile stress by
PCMI.
3) Radial hydride cracked by high stress and/or strain.
4) New, small hydride precipitated at the crack tip due to stress concentration and cracked
repeatedly.
5) When stress exceeded the yield strength of residual cladding thickness, it failed ductilely.

ACKNOWLEDGEMENTS

This study was carried out mainly under the direction of committees of NUPEC. The authors
wish to express their gratitude to the members of thěCommittee of Verification Tests on
High Burnup Fuel", especially Dr. T. Okubo and Dr. M. Yamawaki for their discussions and
evaluations with respect to this program. The authors also gratefully acknowledge METI for
the sponsorship on this project.

REFERENCES

[1] T. KUROSU, et al.,̌Verification Test Program of High Burnup Fuel", Japan-Russia

Experts Meeting on Power Reactor Fuels, held in Tokyo, Japan, 19-21 September (1994).
[2] H. HAYASHI, et al., ̌Irradiation Characteristics of BWR Step II Lead Use Assemblies",
Proc. 1997 ANS Int. Mtg. on LWR Fuel Performance, Portland, Oregon, (1997), 296.
[3] H. SAKURAI, et al., “Irradiation Characteristics of High Burnup BWR Fuels”, Proc. 2000
ANS Int. Mtg. on LWR Fuel Performance, Park City, Utah, (2001), 151.
[4] H. OHARA, et al.,̌Fuel Behavior During Power Ramp Tests", Proc. 1994 Int. Topical
Mtg. on LWR Fuel Performance, West Palm Beach, Fl, (1994).
[5] D. SCHRIRE., et al., “Secondary Defect Behavior in ABB BWR Fuel”, Proc. 1994 Int.
Topical Mtg. on LWR Fuel Performance, West Palm Beach, Fl, (1994) 338.
[6] G. LYSELL et.al., “Axial Splits in Failed BWR Fuel Rods”, Proc. 2000 ANS Int. Mtg. on
LWR Fuel Performance, Park City, Utah, (2001), 216.
[7] D. SCHRIRE., et al., “Testing Cladding Integrity at High Burnup”, IAEA Tec. Committee
Mtg., Nyköping, Sweden, (1998).
[8] Nuclear Power Engineering Corporation, The Annual Report on Irradiation Tests of High
Burnup Fuels in FY 2001 -- Integrated Evaluation on Irradiation Behaviour of BWR High
Burnup Fuel --, March 2002.

 163