Concrete containment management using the Finite Element
technique combined with in-situ Non-Destructive Testing of
conformity with respect to design and construction quality
CONMOD
Co-ordinator
Force Technology
Park Allé 345
2605 Brøndby
Denmark
Tel.: +45 43 267000
Fax: + 45 43 267011
List of partners
1. Force Technology, Denmark
2. Scanscot Technology, Sweden
3. EDF, France
4. Barsebäck NPP, Sweden
CONTRACT No:
EC Contribution:
Total Project Value:
Starting Date:
Duration:
FIKS-CT 2001-00204
EUR 454 000
EUR 1 351 298
Jan 2002
39 months
1
Contents
1. List of abbreviations
3
2. Executive summary
4
3. Objectives and methodology
5
4. Demonstrator 1 – Barsebäck 1 containment
10
5. Demonstrator 2 – MAEVA mock-up
19
6. Conclusions and recommendations
22
7. References
26
8. Tables
28
9. Figures
30
2
1. List of abbreviations
NDE
Non Destructive Examination
FEA
Finite Element Analysis
LCW
Lower Cylindrical Wall
UCW
Upper Cylindrical Wall
ODPT
Over Design Pressure Test
B1
Barsebäck 1 NPP containment
BWR
Boiler Water Reactor
PWR
Pressurised Water Reactor
MAEVA
MAquette Enceinte en Vapeur et en Air, Steam and Air tests
confinement mock-up
AC
As-built Compliance (of structure)
CQ
Condition and Quality (original of structure)
A
Ageing (of structure)
ASR
Alkali Silica Reaction
MASW
Multi-channel Analysis of Surface Waves
SASW
Spectral Analysis of Surface Waves
UPE
Ultra-sonic Pulse Echo
IE
Impact Echo
R
Radar
HECR
High Energy Computed Radiography
P-wave
Compression wave
S-wave
Shear wave
R-wave
Rayleigh wave
The term ”seismic” techniques refers here to all forms of NDE method based on the
transmission of mechanical waves, i.e. all frequencies and modes.
3
2. Executive summary
CONMOD is an exercise project that involves combining advanced finite element analysis
and the latest non-destructive examination techniques with the aim of establishing the
condition of reinforced concrete structures, as well as their present and future behaviour under
various loading conditions including the effects of ageing. The subject of this work has been
the nuclear power containment of Barsebäck 1 and the MAEVA mock-up.
It can be said that CONMOD is a response to recognition of the fact that there is a real need
for a new approach to what is commonly termed ageing management of concrete structures. It
is also an attempt to demonstrate the need to combine new technologies harmoniously and to
explore their mutual benefits effectively.
CONMOD is unique in the sense that the investigative techniques have been applied to an
actual containment structure. In this sense the work is highly qualified to respond to the
question as to whether this kind of investigative process is really necessary and possible, i.e.
for this type of structure and using these techniques. This doubt stems partly from a lack of
understanding of the issues that are relevant, as well as a scepticism towards the reliability of
NDE techniques. These are dealt with in CONMOD.
It has been shown that it is necessary to investigate a structure such as a nuclear containment
using a broad approach and ”penetrative” techniques. It has been shown that traditional visual
inspections and leak-rate tests are insufficient.
The relevant issues are of course structure-specific, but all fall within the following
categories:



as-built compliance
quality and condition
ageing
These factors are all relevant to a greater or lesser extent to complicated structures like
nuclear containments, and their interdependance increases with age. This is demonstrated in
this work with reference to the Barsebäck Unit 1 containment. The condition of the structure at
any time is seen to be dependant on the geometry, detail, quality and materials. These are never
as assumed or strictly according to design and it is therefore not possible to predict condition
and ageing by theory alone.
It is not the aim of CONMOD to specifically find defects in structures but rather to make a
realistic and constructive diagnosis. A certain degree of reticence from utilities in adopting new
approaches to diagnostic investigation may be due to a misconception on this point. The
purpose of CONMOD is to make an accurate and scientific diagnosis of the concrete structure
and in this way detect relevant issues that can then be effectively and timely remedied. At the
same time all non-relevant issues are dealt with and dismissed. The alternative approach is to
do nothing and hope that no problems arise. This latter approach cannot be defended and has
on several occassions lead to serious and expensive consequences. A measure of the success of
CONMOD is the fact that many of the initial fears have been allayed, and the intimate cooperation between utilities, investigative engineers and analysts has been realised – an example
to the industry world wide.
4
This summary outlines the methodlogies used in the MAEVA and Barsebäck projects. These
methodologies are partly demonstrated with a description of the several stage investigative
process and a proposed method of seismic monitoring as a tool for predicting ageing. The work
has been described more as a feasibility study than a solution to the problem of manageing
ageing of concrete structures. It cannot, however, do more than describe a general
methodology since a fully-detailed analysis would only be relevant to the structure in question
and would not be of universal use. Many avenues of future development have been identified
and recognising the urgency of needs in this area some short-term workable ideas are
presented.
The main conclusion of CONMOD is that a new approach combining non destructive
examination with finite element analysis methods is both workable and necessary in order to be
able to accurately determine and predict the condition of nuclear power containment structures.
In order that this technology can be applied in practice in the near future and in the light of the
fact that all concrete structures are unique, requires that site-specific procedures be developed,
implemented and validated. This requires initiative primarily from utilities. The subject of
”Ageing” of concrete structures is often defined as being a priority although it has not been
defined what the issues really are and far less has it been shown what relevance they actually
have. A generalisation of ageing issues does not benefit individual needs since these are often
specific. CONMOD has demonstrated that other factors such as defects and unknown detailing
introduced during construction can be at least and often much more critical than ”classic
ageing” mechanisms. This applies to critical defects introduced at the construction stage as
well as conditions that can initiate and accelerate ageing processes. It has also been seen that
the condition of older containment structures can be generally good and better than might be
expected. Understanding the structure requires however that a full and detailed investigation be
made and followed up, as demonstrated in this work.
3. Objectives and methodology
3.1 Objectives
The objective of CONMOD has been to find a practical means to determine the condition of
a containment structure, as well as how this can be expected to change with time. The
methodology described and to some extent demonstrated here could be applied to other
structures. Information about the condition and ageing processes would be used to predict the
behaviour and functioning of the structure under loading and how this might change with time.
It was to be shown how this would be possible by combining Non Destructive Examination
(NDE) and Finite Element Analysis (FEA) in a series of steps.
The original objectives were to locate critical parts of the Barsebäck 1 structure, as well as
critical parameters, with the help of FEA and to use this information in planning how and
where NDE would be carried-out. In other words it would be possible to concentrate
investigations on the most critical parts of the structure and obtain the most critical information
on the basis of FEA analysis. At the same time the FEA models of structural behaviour would
be up-dated with information obtained by NDE made on the structure. For example, parametar
studies could be made to study the effects of non-compliance with design such as deviations in
reinforcing geometry. Similarily the beneficial effects of long-term strength gain and drying
could be evaluated as could a hypothtical effect such as lack of bond between liner and
concrete caused by corrosion to the liner. It was not clear at the outset how NDE and FEA
could be combined in more sophisticated analyses, e.g. of ageing trends and the consequences
5
thereof. The MAEVA project was focussed on the problems of predicting structural behaviour
and the initiation and growth of cracks, with consequent risk of increased leakage. Crack
growth, for example, has been modelled and predicted with a FEA technique and compared
with information obtained by acoustic modelling.
There are many unknown factors about a large concrete structure, especially one that is first
examined 30 years after construction, like the B1 Unit in this work. For it to be possible to
predict the future condition of a structure it is necessary to construct a Profile of that structure.,
in the same way as a Doctor would diagnose a patient. All factors relating to geometry,
detailing, condition and ageing together describe the Profile of the structure. The objective of
CONMOD has also been to demonstrate the importance of all these factors.
The structure Profile is described by:
The as-built Compliance with design
(AC)
The Condition and Quality of the concrete (CQ)
The Ageing processes (A)
A means by which NDE and FEA can be used together has been sought as an alternative to
traditional inspection and analysis methods. This is in recognition of the fact that for
complicated structures and processes then these disciplines are on their own inadequate –
sometimes referred to as ”islands of knowledge”. The objective has been to combine
sophisticated theoretical structural analyses with real, practical data collected from the
structure. In this way a purely theoretical discipline is vastly reinforced with real, experiencebased knowledge backed up with scientific NDE data. At the same time the inspection process
using NDE can be guided with the help of FEA to enable more efficient and rational data
collection. In addition the quality and correctness of NDE data interpretation is improved with
the help of FEA modelling. Understanding the ageing process using these methods and
following the proposed methodology requires that predicted and measured NDE responses are
the same. These responses reflect a condition at a given time, as well as a direct or indirect
change in condition resulting from some ageing process.
The process of establishing the condition of a structure is an iterative one, which in practice
revolves around NDE and FEA in a series of both independant and combined steps. NDE
responses are affected by the condition of the structure and they will change with time as the
concrete changes in the ageing process. The condition and changes in condition due to ageing
processes that concrete may undergoe can be described by non-destructive, seismic techniques.
These techniques are the ”windows” to the present and future condition.
In this study some deviations have been found, which can be directly related to as-built
compliance and quality. These observations are in some cases of no direct relevance to the
structure and its behaviour. They will however have significant effect on the seismic response
of the structure and must therefore be described accurately.
6
Basic objective of CONMOD
The basic objective of CONMOD is to construct a profile of a concrete containment that
describes the condition and ageing. The theoretical profile of the structure is formed into the
actual profile in a series of investigative steps. These steps include a diagnosis of the ageing
processes, which enable future condition to be predicted (Figure 1).
3.2 Methodlogy
The CONMOD methodology is a diagnostic tool used to investigate the conformity and
condition of concrete containments. This can be incorporated in an ageing management plan. It
makes use of NDE and FEA combined if possible with other forms of testing such as material
sampling, site monitoring, lift-off tests and pressure tests. In CONMOD B1 then NDE, FEA
and material sampling have been employed.
3.2.1 FE analysis in general
When carrying out finite element analyses (FEA) of the reactor containment, a library of FE
models has to be established, which can be used during initial and up-dated analyses of the
structure. These can also be used to assist in planning NDE and other inspection methods. The
FE models range from global and local models for evaluation of the leak-tightness and loadbearing capacity of the structure, to “digital mock-ups” of important areas that simulate seismic
responses that can be used to predict and interpret the NDE response on site.
This library of FE models must be site-specific, and must cover the issues of relevance. If a
complete study is at hand, a fully 360-degree global model of the containment is
recommended, complemented with local models at critical areas (singularities) such as the
major penetrations. In some situations, so-called wedge models, or even axi-symmetrical
models are sufficient. At Barsebäck Unit 1 a combination of an axi-symmetrical model, a
horizontal plane strain model, and a detailed 3D-model of a pipe entry (Figure ) was used. This
was used in order to study the global behaviour of the structure in an undisturbed region, as
well as the response of a typical wall section with pre-stressing buttresses, and a detailed study
of the pipe entry. A digital mock-up of a typical wall section at Barsebäck Unit 1 has also been
established (Figure 24).
Within the scope of CONMOD and the ISP 48 project, several important reactor containment
design types have been studied;
-
-
BWR (Barsebäck Unit 1) and PWR (MAEVA, ISP 48)
Steel liner on the inner surface (ISP 48), embedded liner (Barsebäck Unit 1), without
liner (MAEVA)
Bonded pre-stressing tendons (Barsebäck Unit 1) and unbonded tendons (ISP 48)
3.2.2 Initial FE analysis
In the case of a structure as large as a containment, it must be possible to rationalise the
testing and sampling and include parts of the structure, which are both representative of the
structure as a whole, as well as possible weak areas. This is the first step in establishing a
profile of the structure, i.e. normal and special (critical) parts of the structure are identified.
7
This can be done by combining an inventory of the structure with an intial structural analysis
based on nominal data. The inventory will typically include design detailing, construction
methods, in-service history and results from leak-rate tests etc. The initial structural analysis
will be used to find critical sections of the structure as well as critical parameters (Figure 2.)
The FEA analysis is also used to establish the nominal safety levels with respect to leaktightness and load-bearing capacity. The effect of ageing trends such as creep and shrinkage
and relaxation of pre-stressing steel should be considered with the help of theoretical models
and/or site monitoring.
3.2.2 Preliminary NDE investigation
The next step in the process is an intial site investigation using NDE and traditional
techniques. The objective is two-fold:
1) to compare theoretical NDE responses with those measured and establish a general
picture of the condition and compliance
2) to evaluate the NDE methods w.r.t. capability and accuracy and if necessary apply
alternative techniques
Having completed this then the main site investigation can be planned. The techniques have
been chosen and the objectives of the investigation have been decided upon, including where to
test and what to measure. The first objective is to establish details such as the internal and
external geometry and details of the structure, such as quality variations and trends. Site tests
are compared with tests on mock-ups and responses are compared with predicted, theoretical
responses, e.g. using FEA modelling techniques (Figure 3.)
3.2.3 Main site investigation
The main site investigation is first made when the methods, their application and their
predicted, typical responses are known for the structure in question. This cannot be established
without initial structural analysis, a preliminary site investigation and NDE-modelling.
Inspections are carried-out at a representative number of positions in parts of the structure that
are representative of the structure as a whole, including parts that have been found to deviate
from the normal. The predicted NDE responses are established on the basis of actual structure
geometry and condition, which will normally require some core-sampling at the outset. The
objective of the main inspection is to sub-divide and classify the main parts of the structure in
terms of normal and special responses. These normal and special NDE-responses may be
observed globally and may be due to the effect of different construction techniques, or local
environment and / or damage resulting from these factors. The expected result is a series of
typical or normal responses to one or several NDE-techniques that describe various normal
conditions as well as special conditions. At this stage the NDE techniques are analysed with
the help of FEM and other means in order to interpret the responses in a way that will describe
the geometry and condition for the concrete at the time of test (Figure 4.)
3.2.4 Special investigations
Special investigations are conducted when responses from the structure are found to deviate
from the normal established responses. Deviating responses may occur globally or at a given
position on the structure with the passing of time.
8
At this stage both NDE and FEA have been used as separate tools for analysis of the
structure. It has been necessary to apply FEA techniques to interpret the NDE responses
obtained. Also, as a final stage, FEA has been used to predict the changes in NDE responses
that may occur with time. The condition of the structure has been established, as have the asbuilt conformity and the ageing processes. In other words a Profile of the structure has been
established.
The Profile of the structure obtained in this way is used to up-date the structural model. The
function of the structure (leak-tightness and load-bearing capacity) is then re-evaluated. This
must include time-dependant (ageing) parameters that can directly or indirectly (as a
consequence of) affect the structure. These changes will affect mechanical wave transmission
and they can be modelled using FE-analysis. The key to applying NDE in the ageing
management process is in being able to recognise the changes that might occur in the structure
with time. These changes may be slow and progressive and may or may not in themselves have
any serious effect on the functioning of the structure. On the other hand, they may lead to a
gradual change in the performance and function of the structure and may result in a sudden and
critical change, for example, the development of cracks or loss of liner bond with concrete.
NDE techniques are used as a set of windows to the condition of the structure and changes in
the condition that may occur with time.
3.2.5 Use of NDE and FEM together in the ageing management process
There are several ways in which NDE and FEA are combined:
Identification of critical parts of the structure for NDE evaluation including critical
parameters
Up-dated structural analyses using input from NDE examinations
Prediction of NDE responses for a known condition at a given time using FEM
modelling techniques
Prediction of NDE responses using FEM modelling techniques based on known condition
and how this will change due to ageing processes
Points 3) and 4) apply to both normal condition of the structure and to the case of a particular
”ageing” process.
4 Demonstrator 1 – Barsebäck 1
In this section, a typical pipe entry including surrounding wall are considered. The pipe
entry passes through the lower cylindrical wall at the level of the equipment hatch, i.e. just
above the slab separating the condensation pool and compression space from the primary
compartment. The choice was made to inspect a pipe entry, as there was reason to suspect that
voids may be found in the concrete surrounding it. This particular pipe entry was chosen since
it was easy to access both from the outside and inside of the containment. A list of inspections
made at the pipe entry and surrounding wall (in order of execution) and observations from
9
these is given in Table 1. The FEM analysis required that the size of any void be determined in
order that the integrity of the liner under internal pressure be established.
4.1 Summary of observations with respect to condition at time of inspection
4.1.1 Pipe entry: It was suspected that there could be voids in the injection grout above the pipe
entry and this was confirmed. However, there was no apparent corrosion to the liner plate,
which was the major concern. The wall thickness was found to be considerably greater than
design. The moisture level inside the void was (rH) 82%, which was as expected (Figure 5.)
4.1.2 Wall surrounding pipe entry: The wall condition and thickness was found to be typical of
the lower cylindrical wall in general. The thickness of the outer pre-stressed wall was 880 mm,
which is 140 mm greater than design. The outer 250 mm of the pre-stressed wall has some
significant anomalies, due to the slip-form casting process. The reinforcing and cable ducts
have been set at a depth from the surface, which is approximately proportional to the greater
wall thickness. The concrete wave velocities were found to be similar to the LCW in general,
i.e. high velocities at depths > 250 mm, consistent with a strong and uniform concrete. The
relative humidity of the pre-stressed wall was similar to the LCW in general, i.e. significant
increase from the relatively dry outer surface (40-50% rH) to quite humid conditions (> 90 %
rH) near the liner plate. The seismic tests indicated that the liner and concrete were bonded
together, since sound transmission through the entire wall thickness was possible.
In Table 1 a list is given of the deviations from the assumed (design) detailing and condition.
These are classified according to whether or not they are related to an ageing process (A), to
the original as-built compliance (AC) or the original condition and quality (CQ). Each is also
considered with respect to possible influence on ageing processes in the past and future.
Each significant deviation is therefore related either to the original quality and/or the as-built
compliance. The fact that the wall thickness is significantly greater than design will have had
an influence on moisture movement (drying) and subsequent long-term strength gain. No
damage mechanism or effect was observed related to ageing. The significance of the nearsurface anomalies in the concrete is less since the wall thickness is greater than design by an
amount which is approximately equivalent to the damaged layer thickness. The reinforcing and
cable ducts are not within the damaged layer (Figure 6.)
4.2 Next stages in the CONMOD methodology
4.2.1 Pipe entry
Initial structural analyses have been made of the containment wall and the pipe entry using
the available design data. A thorough investigation has been made of the structure using NDE
and material sampling. The data obtained has been used to up-date the structural analyses, i.e.
the effect of the observed void above the pipe entry.
The pipe entry problem is a specific one and considering that 2 out of 2 pipe entries
inspected have shown similar voids, and based on previous experience, then it is reasonable to
assume that this is a systematic fault which applies to most if not all pipe entries. The
conditions may vary between pipe entries, if the size of the void varies and if there is
separation of the injection grout materials. In the worst case these factors could lead to liner
corrosion. The problem should be dealt with by 100 % inspection of the pipe entries and repair
10
of the concrete where voids are found to occur. B1 has however been out of service for some
years and no repairs are necessary. See 4.6 below.
4.2.2 Wall surrounding pipe entry
4.2.2.1 Stage 1: preliminary NDE tests and modelling
The surrounding wall is typical of the lower cylindrical wall of the containment. A profile of
the wall has been constructed using various forms of NDE combined with material sampling
(coring). The condition of the concrete wall may change with time due to various factors*:
Drying shrinkage and creep with micro-cracking as a result of differential movement in the
wall x-section
Expansion of the concrete due to ASR
Loss of bond between the liner and concrete
Corrosion to the liner
Cracking or delamination induced by outside force
Other unknown
Further long-term strength gain of the concrete
* These are hypothetical and have not been found to apply in this case at this time.
These represent possible ageing mechanisms, which may be either slow and progressive,
accelerating or diminishing. They may also be in the form of a consequence of some slow
process, e.g. the development of cracks. They would affect either the concrete or steel
(reinforcing, liner plate and ducts) but regardless of which, they would affect the seismic
response of the wall.
The Profile of the containment wall is described by the condition, geometry and ageing. The
condition may change with time and this can be detected and described by seismic profiling.
This has been the basis of the CONMOD demonstrator.
4.2.2.2 Normal seismic responses.
In this section, the sequence of site tests and comparison with expected and measured
responses is demonstrated. Further investigations are motivated by the discrepencies between
predicted (theoretical) and measured responses.
Initial NDE-tests : comparison of predicted and measured responses
The measured (apparent) P-wave velocity is based on a full wall thickness of 1000 mm, the
grade of concrete specified and the age of the structure (as well as experience of the concrete at
this site). There are clearly significant differences between predicted and measured values of
the observed four normal seismic responses.
Normal seismic responses
The condition of the concrete is described by the normal seismic responses. The normal
response of the structure depends on the geometry and condition of the structure:
N  (G, C)
11
G is a geometry factor
C is a condition factor
Reference is made to the normal seismic response, NS. The condition of the concrete can be
described by both stable and time-dependant factors, e.g. the condition after casting and the
changes that the concrete undergoes with time due to drying.
NS  (G, (CSt + Ct))
CSt is a stable condition factor
Ct is a time-dependant condition factor
See Figure 7.
In Figure 8, the predicted and measured seismic responses (SASW – Rayleigh Wave
dispersion) are shown for the lower cylindrical wall. These are based on the geometry and
quality of the concrete, as well as measured shear wave velocities. At this stage there was
clearly some unknown factor (s) causing the discrepency between predicted and measured
values. There was also reason to believe that these observations (of seismic and radar
responses) may be linked to each other. The seismic response for the lower and upper
cylindrical walls are shown in figure 9.
Initial modelling of seismic response
There is a clear difference in the quality of data and measured apparent Rayleigh wave
velocity when comparing the Lower and Upper cylindrical walls. The suspected cause of this
difference was the fact that the LCW was slip-formed while the UCW was not. Slip-forming
could be expected to have caused some near-surface damage to the LCW. The effects on a
dispersion curve of a low-velocity surface layer were investigated by modelling. In addition,
FEM-modelling has shown that the ratio of surface wave wavelength to wall thickness is such
that the seismic response will be affected. In this case, assuming a wall thickness of 1000 mm,
the wave velocities would dip downwards at wavelengths around 500 mm (Figure 10 and 11.)
Note that the low velocity layer is assumed in this first model to be only 100 mm thick,
while later investigations showed this in fact to be 200 – 250 mm.
Summarising initial NDE results and FEM-modelling of predicted responses:
The measured apparent P- and R- wave velocities were found to be significantly lower than
predicted.
The depth to the reinforcing and cable ducts was found to be significantly greater than design.
The wall thickness measured by radiography appeared to be approximately 15 % greater than
design.
12
Some assumptions have also been made at this stage based on experience and preliminary
FEM-modelling:
 The wall geometry can be expected to cause a drop in the R-wave velocity at wavelengths
greater than half the wall thickness.

A low velocity surface layer will affect the dispersion curve signicantly.

The slip-forming technique can have caused a low-velocity surface layer.
Based on this combined information, it seemed reasonable to assume that the total wall
thickness was significantly greater than design and that the outer pre-stressed wall had a lowvelocity layer. The apparent low seismic wave velocities were therefore probably due to these
factors together with the effects of the wall thickness:wavelength ratio.
4.2.2.2 Stage 2: Material sampling and NDE on cores
Core sampling and a fibre-optic inspection at the pipe entry revealed the following (applicable
to the lower cylindrical wall):
 The wall thickness at this position was found to be 880 mm (pre-stressed outer wall)
compared with a design value of 734 mm.
 The outer 200-250 mm of concrete was damaged by the slip-forming process causing this
to be a low-seismic-velocity layer.
 The properties of the concrete vary significantly with depth due to the long-term strength
gain of the concrete.

The average P-wave velocity of the concrete in the pre-stressed wall was 4500-4600 m/s.
4.2.2.3 Stage 3: Up-dated seismic modelling of containment wall based on new parameters
Choice of seismic testing method
The methods used up to this point for seismic testing have been SASW and IE. The former
technique, SASW, is perhaps more valuable since it provides a characteristic profile of the wall
section from which it is possible to extract information about wave velocities and their
variation with depth, as well as total wall thickness. The method is however sensitive to local
inhomogeneities in the concrete and for this reason a more robust and stable method was
sought.
Multi-channel Analysis of Surface Waves (MASW)
This method involves collecting seismic wave data via a sensor placed on the concrete
surface and multi-shot impulses (hammer strikes) at increasing distance from this. The
13
information is transformed from the offset-time (x-t) domain into the frequency-phase velocity
(f-VPH) domain. The method makes no assumptions about the nature of the seismic event in
relation to the phase velocity, and the construction of the frequency-phase velocity image is
performed through an objective pattern-recognition technique. The MASW method enables a
record of the total wavefield of both surface and body wave (P-wave) events (Figures 12 & 13)
The seismic response of the Lower Cylindrical wall was studied by first making
complementary MASW measurements on site and then FEM-modelling of the response, using
the up-dated information on the thickness of the wall and condition of the concrete. The updated information included the velocity profiles obtained from the cores taken from the wall,
the depth of the observed damaged surface layer and the total wall thickness. The objective has
been to attempt to match predicted and measured seismic responses for the LCW.
A more refined evaluation using MASW in combination with Impact Echo was seen to have
the best potential as a suitable seismic method. A sensitivity analysis of MASW data for the
structure can indicate which parameters (thickness, velocities, steel liner plate etc) that can be
evaluated from MASW data. This analysis can also give an indication as to the extent and
quality required of the measured data (separation and number of points). A first step analysis
involves an analytical solution (1 D model) instead of the more lengthy FE analyses in 2D. The
FE analysis can be made in a second stage of the sensitivity analysis in order to study the bond
conditions between concrete and steel liner, the effect of anomalies and the (disturbing) effect
of pre-stressed cable ducts etc. The results of the sensitivity analysis can show which
parameters are worthwhile trying to evaluate from the MASW data. An automatic evaluation
(inversion) is then tested on real MASW data from positions at which as much reference data
as possible is known.
Figure 14 & 15 show the phase velocity spectrum with amplitudes as a function of frequency
and phase velocity. The upper and lower diagrams in Figure 15 represent the measured and
best-fit synthetic data. The latter is based on a known layer model and the field set-up
geometry. Sensitivity analyses were made of the containment wall with reference to shear
velocity in the outer 100 mm (assumed) layer, the thickness of the inner (high velocity) layer at
a fixed velocity and the thickness of the outer layer with a fixed velocity (lower). Note that
their is no clearly defined layer thickness and that the velocities vary with depth, i.e. the
method used is an approximation. These layers refer to the load-bearing pre-stressed wall. The
predicted and observed phase velocity spectra are compared by a function known as the
mismatch ( M ). This can be a value between 0 and 1 where 0 represents a perfect match. A
sensitivity analysis is made to establish which parameters in the model have a significant effect
on the phase velocity spectrum (Figure 16 & 17.)
Ref. 2. ”Inversion of surface wave dispersion curves is a non unique and non linear inverse
problem, where a direct solution is not possible. Non unique means that different layer models
may match the measured data equally well. Non linear means that small changes in the data
can correspond to large changes in the model and vice versa.
The new procedure used for inversion of surface wave data is based on the complete phase
velocity spectrum rather than discrete dispersion curves. A global optimisation technique, Fast
Simulated Annealing (FSA), is implemented for the inversion algorithm. With this procedure,
data reduction and the extraction of an experimental dispersion curve is avoided.
Consequently, the complete procedure from data collection to evaluation of layer properties
becomes fully automatic and objective.”
14
...The goal of the inversion procedure is to find a layer model whose corresponding phase
velocity spectrum matches the measured spectrum best...
...The basic concept of the simulated annealing method is to perform a random walk in the
multidimensional parameter space with the intention of finding the global minimum of the
objective function ”
4.3 Reference data used in the seismic model - Wall thickness
According to HECR measurements the average wall thickness is 1130 -1144 mm and IE on
average 1070 – 1090 mm (Figure 18.) At the pipe entry wall considered, the actual thickness of
the pre-stressed part was found by drilling to be 880 mm, which would correspond with a full
wall thickness of 1146 mm assuming an inner wall thickness of 260 mm.
Measurement of damaged layer thickness – CONMOD B1
This has been possible thanks to the fact that a large number of cores were removed from the
containment wall. The damage can in some cases be seen without visual aids, but the true
extent of the damage in the outer zone is first seen with the help of a plane section of one of the
cores. The cores were examined with small transducers in the transverse and longitudinal
direction along their length. Both P-wave and S-wave velocities were measured (see inset in
Figure 22). The depth of the damaged layer has been established to about 220 – 250 mm, as
suggested by the inconsistency in the ratio of shear to compression wave velocities in this
zone.
4.4 Summary – Demonstrator 1 Barsebäck containment
The seismic response of the containment wall at Position 2 has been modelled using a Finite
Element technique, in which the in-going parameters have been determined using NDE and
other investigative methods on site. The solution to this model is not unique and it is therefore
necessary to construct the model with the help of complementary techniques, which in this case
includes NDE and material sampling. A match between predicted and actual seismic responses
has been achieved. This reflects the agreement between the assumed (according to initial NDE
investigations and not theoretical) and actual condition and geometry of the containment wall
(Figure 16, 17 & 22.)
The geometry of the wall, i.e. the thickness, the depth to liner and cable ducts etc. can be
determined by combining NDE techniques. Some global variation in these parameters can be
expected. In an ageing management scheme, reference sections should be established in which
these parameters have been established along with a corresponding seismic model. Any
changes in the seismic response with time can then be analysed with respect to changes in the
condition of the concrete alone. The effects of variations in internal geometry, e.g. reinforcing
and cable duct depth and liner:concrete bond, should be evaluated in a sensitivity analysis of
seismic response.
4.5 Initial FE analysis
Inital FE`analyses of the Barsebäck Unit 1 containment showed that the area around the pipe
entry was critical wth regard to stresses in the steel liner. As can be seen in Figure 19, the
15
accidental internal overpressure at liner yielding is approximatly 10% lower in the pipe entry
area (position 2) compared with an undisturbed region (position 1).
4.6 Updated FE analysis
Penetration (pipe entry)
FE analyses have been carried out to study the effect of the following deviations in
immediate vincinity to the penetration:
-
Void in concrete above penetration (found at B1)
Friction between steel liner and the surrounding concrete (contact examined at B1)
Loss of one pre-stressing tendon (fictitious)
Void
Non-destructive examination as well as fiber optic inspections have shown that there was a
void in the concrete above the pipe entry. The FE model used in the initial analysis has been
up-dated with this information, i.e. a void has been introduced into the model (Figure 20). The
analysis shows that the formal leak-tightness capacity is decreased by approximatly 5% if the
void reaches out to the critical zone identifed in the initial analysis. However, the void
identified at this position did not reach this area, and hence no reduction in capacity is at hand.
The void is thus mainly an ageing problem (possible risk for corrosion).
Friction
If the coefficient of friction between the concrete and the steel liner is increased from zero to
0.6 in the penetration area, an increasing interaction between the inner and outer cylinder wall
occurs. This leads to a reduction in the stress in reinforcement, tendons and steel liner
compared with the original analysis at the same internal over-pressure level. Yielding would
occur in the steel liner at a slightly higher pressure compared with the initial analysis.
Tendon
If a critical tendon is removed near the pipe entry, the cracking of the concrete in the
containment wall initiates at a lower internal overpressure than in the initial analysis. This
implies that the stresses in reinforcement, hoop tendons and steel liner start to increase at an
earlier stage. The stress in the steel liner would reach the yield stress limit at an overpressure
that is approximatly 6% lower than in the initial analysis.
General wall section (not specific to pipe entry)
The following deviations in an general area of the containment wall have been studied:
-
Increased thickness of the containment wall (found at B1)
Friction between steel liner and the surrounding concrete (contact examined at B1)
20% lower pre-stressing force (than long-time design value) in all tendons (ficticious)
Loss of one pre-stressing tendon
Loss of pre-stressing force in one tendon (maintaining steel area)
10% general reduction in rebar steel area
Friction between steel liner and the surrounding concrete
16
Wall thickness
If the increase in wall thickness is assumed to imply a global egg-shaped geometry (see
Figure 21) then the radial deformation of the containment will be changed, especially after
cracking of the concrete has occurred.
Friction
A higher frictional value allows a greater ability to transfer shear forces between the
different structural parts. This means that the outer part of the wall will be helped by the inner
part, mainly at vertical loading. Since the vertical loading is critical when it comes to the
ultimate load bearing capacity, then this interaction will increase the capacity to some extent.
Pre-stressing force
A reduction of the pre-stress force by 20 % has no noticeable effect on the ultimate load
bearing capacity (as is expected). However the reduction results in an initially weaker
structure, which leads to an earlier yielding of the steel liner, as well as the vertical
reinforcement connected to the conical roof.
Horizontal tendon
When a critical horizontal tendon is removed (level + 116.82) the effect is local. The
ultimate load bearing capacity is not affected due to the fact that it is still the vertical tendons
that govern the ultimate capacity of the containment (connection between LCW and conical
roof). The leak-tightness capacity is reduced by approximately 5 % in the area where the
tendon is removed.
Re-bar x-sectional area
A reduction of the reinforcement area by 10 % does not affect the leak tightness, and affects
the ultimate loading capacity only marginally. This is reasonable since this reduction only
gives an insignificant reduction of the section stiffness and an even smaller reduction of the
total tensional capacity of the steel components (reinforcement, pre-stressing tendons and steel
liner).
A higher frictional value allows a greater ability to transfer shear forces between the
different structural parts. This means that the outer part of the wall will be helped by the inner
part, mainly at vertical loading. Since the vertical loading is critical when it comes to the
ultimate load bearing capacity, then this interaction will increase the capacity to some extent.
4.7 Seismic modelling as a diagnostic tool for monitoring ageing of the structure
Having access to a diagnostic tool that can describe the condition and geometry of the
containment wall can be used for:


determination of global trends (normal responses) and deviations from these
determination of changes that may occur to the material and structure with time
The change that concrete or concrete and steel bond might undergo may be a slow process,
such as drying of the concrete or long-term strength gain. It may on the other hand be a
relatively sudden change such as cracking or corrosion to the liner. Whatever the case, the
consequence of an ageing process can be modelled as can the effect this will have on, for
17
example, the seismic response. An example is given in Figure 24. In this case a loss of bond
between liner and concrete occurs at a given time.
Seismic mock up
Seismic non-destructive evaluation holds the key to condition assessment and ageing
monitoring of complicated structures like nuclear containments. The seismic response can be
predicted on the basis of experience and with the help of both physical and digital mock-ups
(using FE-analysis). Digital mock-ups have the benefit that they can be made quickly, accurately
and to low cost. In the planning stage, they would be of great value in establishing which
methods are viable for the structure and problem case considered. They can be used to establish
method set-up on site and predict different types of responses. Parametric studies can also be
made, for example, to determine the smallest defect that can be detected.
Modelling wave propagation with the help of digital mock-ups will enable:
• Increased understanding of wave propagation in complex structures.
• Detailed parametric studies to quantify what NDE can ”see”.
• Pre-study to optimize measurement procedures and field set-up geometry.
• Evaluate material properties such as a match between field data and a known system.
5 Demostrator 2: MAEVA
EDF, with contributions from its partners, constructed a concrete test model (mock-up) at
CIVAUX in France. This represents the normal section of a pressure, water reactor (PWR)
containment, (Figure 25): MAEVA (MAquette Enceinte en Vapeur et en Air, Steam and Air
tests confinement mock-up). The purpose of this was:

To obtain a better understanding of the behaviour of a, reactor containment
subjected to the combined loading of pressure and temperature

To study the evolution of permeability by leak rate measurements and also the
state of cracking of the concrete containment wall

To study the behaviour of a composite liner and its contribution to the leaktightness of the concrete containment wall
Five campaigns of tests have been carried out from December 1996 to December 2002. For
each campaign, an average of 2 sequences (one test with air and another with steam) have been
performed. The CONMOD Project refers to the 7th sequence "Over design pressure test" of
MAEVA of the sixth and last campaign performed in December 2002 before decommissioning
the mock-up.
5.1 Over design pressure test and NDE concrete crack detection
18
During the "Over design pressure test" (ODPT), pressure within the cylinder of MAEVA has
been twice raised to 1.5 times the design pressure (i.e. up to 0.975 MPa) with dry air and at
ambient temperature to enable monitoring of new cracks after the first and second pressure rise
(Figure 26).
For the benefit of CONMOD and in order to apply the CONMOD methodology, a
continuous concrete crack detection system (to detect new crack generation and growth of
existing cracks) was set. This was the Sound Print acoustic detection system performed by
ADVITAM , using 20 sensors set on the inner wall and another 20 on the outer wall of the
mock-up. This NDE technology enables detection of degraded areas during pressure rise, or a
new loading, beyond that which the structure has previously undergone. The results show a
number of acoustic events, which are interpreted as the formation and/or growth of cracks as
being high during the first pressure rise (ODPT-1) and low during the second (ODPT-2). The
crack maps constructed with the Sound Print system and used as the basis for this work are
shown in Figures 29, 30 and 31. Several displacement or deformation sensors have also given
useful information on the local and global behaviour of the structure during ODPT, including
crack opening.
5.2 Finite Elements Analysis using CONMOD methodology
The consulting company SOCOTEC Industrie was chosen as sub-contractor by EDF to
perform the Finite Elements Analysis (FEA) concerning the ODPT on the MAEVA mock up,
with Finite Elements code, Code_Aster® developed by EDF. The technical specifications for
the FEA calculations of the MAEVA mock up have been defined and edited on the basis of the
modelling feedback provided previously by CESA Project. All calculations are based on
Mazars damage model. Two geometrical models have been specified:

An axisymmetrical model with truss elements, which represents a standard zone
(far from any singularity such as hatch or buttress)

A 3D model simulating one half of the structure that takes into account the
symmetry of the structure, boundary conditions and loading
The meshes are specified in order to account for the different structural elements: concrete,
pre-stressing tendons, reinforcement etc. Calculation methodologies and data are specified in
order to allow for linear as well as non-linear calculations in different situations: initial state,
service life and testing over the design pressure. Loss of pre-stress due to creep and shrinkage
is assessed in each situation. It is assumed that the structure is supported by rigid soil.
On the mock-up, a lot of local deterioration and cracks on the external face of the cylinder
have been observed during visual inspections made before the first pressure rise, ODPT-1. The
origin of these cracks was probably linked with damage induced by the previous test
campaigns. Nevertheless, the FEA did not attempt to simulate the effects of these. This should
be one point of possible improvement for further modelling. The full CONMOD methodology,
19
as presented on Chapter 2 of this report, has not been applied in the case of the ODPT
undergone by the MAEVA mock up. Pre-stress loss was assessed according to French design
code, taking into consideration:
Elastic losses corresponding to the level of a tendon to the combined effects of other tendons
Friction losses, due to friction between tendon and its duct, depending on
curvature
Slip that is inevitable when pre-stressing force is transferred to the anchorage set
Steel relaxation (cables)
Shrinkage/creep of concrete
5.3 Axisymmetrical modelling
The mesh looks like a slice of the mock up, and is intended to be representative of the
behaviour of a standard area of the structure (quadrant 3) without singularities such as
penetrations or buttresses. Elastic as well as non-linear calculations based on Mazars damage
model have been performed. The results of elastic calculations in terms of global
displacements have been used in order to refine the modelling of the boundary conditions
between the cylindrical wall and the bottom and upper slabs. The results of the non-linear
calculations up to the final test pressure indicate no damage occurring in the cylinder wall, as
long as the area in question is remote from any singularity of the structure (pipe entry,
buttress…).
5.4 3D modelling
Symmetry conditions are applied to take into account the unmeshed half of the structure.
The main load of the ODPT is applied as a uniform internal pressure. It was decided that the
calculations would be limited to the first rise ODPT-1 and decrease of pressure. The
distribution of tension along the tendons has been calibrated with available monitoring data.
Since only one half of the real structure has been considered, this entailed several calculations
to estimate what tension to apply at the “fictitious” extremities of the cable that was cut by the
plane of symmetry. The elastic modelling had revealed the areas where cracks may occur,
considering the values of computed stresses in concrete. The results of the non-linear
modelling, taking into account damage and cracking, have been compared with the results of
the acoustic emission survey performed with the Sound Print system, as well as the visual
inspection. When the non-linear model was made, the damage model was activated in the
concrete elements. However, behaviour has remained under the yield threshold in the structural
steel elements (reinforcement, pre-stressing bar and tendon). Considering the global
deformation of the structure (Figure 27 and 28), it can be said that it is qualitatively consistent
with that shown by the instrumentation measurements. This qualitative agreement is also true
at a local level. From a quantitative point of view, it appears that the FEA gives lower
20
displacements than monitoring. One can infer that the, global rigidity of the mock-up is
overestimated by the model. It must be kept in mind that the structure is considered to be
“healthy”, i.e. no cracks, at the beginning of the calculations. However, previously induced
cracks exist before the ODPT. This confirms that the structure has been assumed to be stiffer
than it was during the pressure test.
The comparison between the damage variable of the model (which stands for cracking) and
the locations of acoustic events enables the reliability of our FEA tools to be assessed. This
comparison has been done near a penetration, near the buttress and in the area that was
opposite to the penetration. Around the penetration (quadrant 1), both FEA and NDE exhibit
crack initiation and development. The crack patterns are quite similar (Figure 29). Near the
buttress (quadrant 2), FEA prediction and NDE measurements are also in good agreement. For
the purpose of a more in-depth comparison, it would have been of great interest to compare
acoustic intensity (or energy dissipated by the opening of cracks) with the amplitude of damage
evaluated by FEA. Nevertheless, FEA did not manage to detect any crack initiation remote
from buttress or pipe entry, at 180° azimuth angle of the penetration. This is consistent with the
conclusions stated previously for the axisymmetrical modelling. In fact, acoustic events could,
be assumed to have been created by cracks already generated by previous tests on the mock-up
that have grown with ODPT loading. A link could be made with the previous observation on
displacement and the actual stiffness in relation to cracking.
5.5 Demonstrator 2 – Conclusions (EDF)
Numerical modelling for reinforced and pre-stressed concrete structures (containment vessel
alike) has been carried out. These calculations allow a comprehensive prediction for the
behaviour of the structure, as well as the prediction of initiation of new cracks (generated by
simulated loading) together with their localisation, as shown by comparisons with NDE and
monitoring data.
The initial state seems to have an important influence on local behaviour (opening/closure of
existing cracks under applied loading). If the purpose is to estimate leak rate through a prestressed reinforced concrete wall, knowledge of crack configuration and more sophisticated
analysis are required (see CESA report) to get an overview of such analyses. Current EDF
studies are dedicated to initial state assessment in structures that have undergone complex
loads for several years. A Bayesian approach can help in this way. Another point to stress is
the use of smeared crack concepts. It is well known that engineers using them should be very
careful with mesh dependency or localisation, once convergence difficulties are overcome (see
chapter 5). A simple isotropic damage model such as Mazars’ one is only a first step to better
understanding and predicting the behaviour of structures. More realistic models are also less
easy to handle and easily become unsuitable for engineering purposes.
Because of the restricted time allocated for CONMOD, it was not possible to apply the
whole CONMOD Methodology at the MAEVA mock up case with the ODPT loadings. A
second run of calculations would be useful, for example, it would have been interesting to
adjust the elastic modulus of HPC to better fit with displacement measurements. Also, it would
have been interesting to attempt some modelling on the long term effects of initial state
induced stresses, e.g. cracks formed during the first test campaigns or at the early age, not
21
taken into account in the present modelling. It includes being able to simulate the effect of
different phenomena in the same modelling, which is currently a specific issue for EDF.
6 CONMOD – Conclusions and recommendations
During the course of this project, which has spanned a relatively short time, four major
advances have been made on the course to producing workable solutions to the problem of
ageing management of concrete containment structures:
1) co-operation on a trans-european scale
2) increased awareness among the theoretical sphere of the possibilities and necessity
of NDE
3) increased awareness in the practical NDE sphere of the possibilities and necessity
of FEA
4) reduced reticence and in some cases full co-operation from the nuclear utilities on
several important issues
This together with the fact that CONMOD has become an example to the industry on a much
wider scale are a measure of the success of the project.
On the purely technical side the project has been a demonstrator of the important issues
applicable to Scandinavian containment structures and other structures in general. It has been
shown that the tools are available today to obtain the required information about the structures
and that this information can be effectively used with the available analysis methods. A
methodology has been defined for condition assessment and ageing management using FEA
and NDE.
The objective of finding a means of predicting the condition and behaviour of concrete
containments under loading and the effects of ageing has not been fully achieved and validated.
A methodlogy and suitable toolbox of NDE techniques has however been identified. This
includes mergence of NDE and FEA in a several stage process. These have been partly
demonstrated in the work carried-out at Barsebäck and MAEVA. The Barsebäck and MAEVA
projects were somewhat ”out of phase” in terms of time and planning, which has restricted the
mutual benefits of these objects. On the other hand, the independance of these two projects has
given a certain amount of objectivity. This may not have been the case if the projects were
conceived and planned together.
In terms of important issues relating to concrete structures like nuclear containments, it has
been confirmed very clearly that as-built compliance, original quality and condition, as well as
ageing are all significant. It could be said that as-built compliance and quality (original
condition) are the major issues, at least in many cases. For structures that are a certain age
before inspection, then it is vital to determine both as-built compliance, condition and
geometry. A dependance on NDE and FEA in managing this kind of structure requires that
22
both geometry and detail be known before condition and time-dependant changes in conditon
can be predicted and characterised.
The Barsebäck and MAEVA work were different in the respect that little was known about
the actual conditon of Barsebäck, while the MAEVA mock-up was very well documented. The
MAEVA mock-up does not have unknown characteristics such as those inevitably introduced
by workmanship, nor does it have trends associated with ageing. The MAEVA mock-up can
therefore be dealt with in a simpler manner, perhaps not requiring all the stages in the
CONMOD methodology. An older structure like the Barsebäck containment is on the other
hand a veritable pandoras box of different conditions and unknowns, requiring the full
CONMOD procedure and methodology. The MAEVA project would have benefitted from a
combination of seismic profiling, acoustic monitoring and FEA-modelling. One obvious
benefit would have been to confirm and map the existence of cracks in the structure prior to
ODPT using seismic techniques. The potential of seismic techniques has perhaps been
underestimated for this purpose.
The techniques available today in combination with advanced FEA-modelling of seismic
responses, can determine and characterise condition change, such as cracking. If cracking can
be predicted in a FEA-model, as demonstrated in MAEVA, then the seismic response changes
associated with the growth of the cracks can also be predicted using FEA-modelling
techniques. These changes can also be measured, preferably using a combination of seismic
techniques. If the cracks can be predicted than they must follow a pattern and such patterns
should be recognisable using NDE. The pattern and extent of mechanical changes to the
concrete and the increasing frequency and growth of cracks will affect the amount of detail that
seismic techniques can determine about them.
The use of seismic techniques for crack characterisation has not been fully demonstrated in
this project due to lack of time and suitable objects. The technique finally adopted, MASW,
has shown itself to be a robust method with great repeatability. Also, the method provides a
spectrum of different data, e.g. both surface and standing wave patterns. In terms of crack
growth monitoring then there is potential in exploring the possibilities of using this combined
information with frequency-dependant attenuation of surface and compression waves.
A clearer definition is required of the term ”ageing” of concrete. The different kinds of
ageing should be clarified, e.g. long-term and progressive changes such as strength gain and
drying or crack initiation and growth due to periodic load testing. These need to be defined
more clearly in their context in order for a management methodology to be defined for each
case. The ageing problem should also be recognised as being very much dependant on original
quality and compliance.
Recommendations
There is a degree of urgency in developing ageing management programmes that can be
usefully applied in practice considering the present age of many concrete containments and
their expected remaining lifetimes. Also, given that all concrete structures are unique then there
is no universal solution to the problem of ageing management and action should not be delayed
in expectance of this. There may be a tendency for some utilities to stand off until it is
considered that a sufficiently useful and reliable technology has been developed. However,
such a technology cannot be fully developed and validated that has universal application. A
logical starting point towards real progress would therefore be to initiate a broad programme of
23
investigative studies at many sites focussed on a several stage process. This would involve an
inventory of the structures and construction of their profiles along the lines outlined in
CONMOD as a first step. Parallel studies of ageing mechanisms specific to these structures
could then be made, using methods that give best effect, i.e. with less focus on the (NDE)
method but more focus on the problem and solving it correctly. The important issues specific
to each structure could then be identified and dealt with using the best available methods. The
ageing management programmes could therefore be built on the principle of CONMOD and
customised for specific needs. This would be the first and necessary major step in introducing
real investigative concrete management to nuclear containments and hopefully a release from
what is today largely a theoretical discussion forum without visible solution.
It must be shown by example that concrete structures are rarely as design, that the quality is
heavily influenced by workmanship and that condition and changes to condition cannot be
described by what can be seen on the surface. Nor can these changes in condition always be
detected by traditional in-service inspection methods, such as leak-rate tests and deformation
measurements. It must also be recognised that ageing of concrete is not necessarily a
detrimental thing, since concrete normally gains in strength with age and a natural drying
process can, for example, reduce the risk of potentially damaging mechanisms such as
corrosion and asr. It must be shown that each structure, particularly older structures, must be
given a diagnosis in order that their continued safe and efficient operation can be guaranteed.
This can only be possible if there is an open willingness of the operator to allow thorough
investigations to be made at their site. This would have the effect of averting a new culture of
ageing prediction based on theory, which we have seen tends to circumnavigate the really
important issues in pursuit of a subject which may be both abstract and irrelevant to the
structure in question. The introduction of a new culture to site management of concrete
structures would have many beneficial effects, not least in the form of establishing expertise at
the plants.
The problem of concrete ageing is one that is described by mechanical changes to the
concrete, be it slow and progressive or sudden. The changes apply both to the concrete as well
as the steel embedded in the concrete and how the concrete and steel interact. These conditions
and changes can be described by seismic testing methods, if they are backed up by advanced
scientific modelling techniques. The potential of these techniques in characterising concrete
changes needs to be further investigated and procedures developed that adapt the technology to
specific types, i.e. patterns of concrete change, such as natural changes or deterioration. A
bridge between ageing prediction using FEM modelling techniques and prediction of seismic
response to these changes using similar FEM techniques is required. There is potential in
examining wave velocity profiles, reflection patterns and frequency-dependant attenuation in
recognising and characterising crack growth. The ageing mechanisms need however to be
defined more clearly and modified to include the effects of original quality and compliance.
The initial or passive state of the structures should be modelled to determine the long-term
effects of initial state-induced stresses. This would include cracks that have previously been
formed and other factors relating to (original) condition and geometry.
The methods used to investigate such as NDE are based on principles which have been
known for a very long time. A concrete structure such as a containment wall is not particularily
complicated, at least considering the large areas of plane section that are undisturbed by edges
and changes in thickness. Yet, several observations have been made in the course of
CONMOD that highlight the fact that little has been done to understand basic seismic wave
24
motion in these structures. Examples are the effects on wave dispersion dependant on the
relationship between wavelength and wall thickness, various wave modes dependant on the
interaction between concrete and steel and the effects of low-velocity surface layers. In order
that seismic methods can be more widely introduced as a reliable method of monitoring
concrete condition, then these phenomena must be more clearly defined as reference to the
user. This can be achieved through the use of digital mock-ups combined with the collection of
empirical data on both physical mock-ups as well as actual structures. This could for example
include, a data-bank of seismic responses for several containment structures of similar and
varying ages. A Handbook of Seismic Responses together with a description of testing
procedures and procedures for secondary checks is required. This Handbook should include
finite element modelling techniques to plan test set-ups for specific structures and problem
cases with sensitivity analyses to determine which parameters can be determined using a given
technique.
Seismic testing offers the possibility of periodic checks of concrete condition or even
continual monitoring. This requires a very high degree of test repeatability, particularily if, for
example, wave attenuation is an important parameter. New methods of sensor technology
should be tried to improve test repeatability, for example the use of non-contact microphones
as an alternative to transducers attached to the concrete surface.
The diagnosis that will enable lifetime predictions of concrete structures and effective
management of these is to large extent dependant on non-intrusive investigation, NDE. The
physical condition of concrete and how this, changes with time can be monitored using seismic
methods, and it seems that this is the only NDE method that is capable of providing this
information. Seismic methods are in turn dependant and limited by the geometry, size,
accessibility and material properties of the concrete. For example, the amount and quality of
data that can be obtained using ultrasonic pulse echo techniques is greatly dependant on the
maximum size of the aggregates used in the concrete – smaller aggregates allowing greater
penetration and resolution. The workmanship quality of concrete also has an enormous impact
on the amount and quality of information that can be obtained non-destructively. Indeed there
is a general rule that the better the concrete then the better non-destructive testing techniques
will perform. If we strive towards better concrete that make, structures more amenable to NDE
then we have achieved two goals – improved concrete and improved NDE. There are further
improvements that can be made in the design and detailing of a structure to make the process
of extracting information easier, for example reinforcement configurations and surface
treatments and coverings. It is a novel concept yet a very practical one to build components
into a structure that can be used as targets and references when later studying crack movement
and seismic wave velocities. These are as yet unheard of yet would be very simple and quite
practical.
Further development for new containment structures should focus on establishing some
simple rules, designs and novel ideas on how to vastly improve the “accessibility” of concrete
structures to future diagnostic investigation.
Finally
The CONMOD project has been a pioneering work and has shown the way to a new
approach to the problem of condition assessment and ageing management. It is however,
largely a feasibility study and their remains a lot to do in the process of qualification, validation
25
and standardisation of investigative methods and diagnostic tools and analysis. A continuation
of this work has great potential in providing a realistic and pragmatic solution to this issue and
the time is critical in order to maintain the momentum that has been created, not least the
awareness developed amongst utilities.
7. References
[1] Shaw P, Rasmussen J, Pedersen T.K, Force Technology report – A Practical Guide to Non
Destructive Examination of Concrete (2004)
[2] Rydén N, Lund Institute of Technology, Doctoral Thesis – Surface Wave Testing of
Pavements (2004)
[3] Rydén N, Lund Institute of Technology, Licentiate Thesis – A Novel Concept for Seismic
Pavement Testing (2002)
[4] Shaw P, Rasmussen J, Pedersen, Force Technology, CONMOD Interim Project reports –
Non-Destructive Testing of Concrete Containment (2002-2005)
[5] Rasmussen J, Non Destructive Testing methods for Integrity Determination of Concrete
Structures – Licentiate Thesis, Luleå University, Division of Applied Geophysics (2003)
[6] Shaw P, Force Technology, A Study of NDE Performance in the Condition Assessment of
Concrete Structures – rilem proceedings PRO 16 Life prediction and Ageing Management of
Concrete structures (2001)
[03-01-EDF]
EDF technical specification ENESMS/02-2103-A – Spécifications
techniques pour la modélisation du comportement de la maquette MAEVA lors d’un essai en
air audelà de la pression de dimensionnement – Author : G. Heinfling – 17/03/2003 – In
French –
(Technical specifications for the modelling on MAEVA mock up behaviour during ODPT)
[03-02-EDF]
ADVITAM report F050 OE DOC 03 – Maquette MAEVA – Essais en air à
9.75 bars – Séquence 7 – Auscultation acoustique – Compte rendu d’essai – Author : Advitam
–28/04/2003 – In French –
(MAEVA Mock Up – Pressure Test with dry air – 7th sequence – acoustic survey – Test
Report)
[03-03-EDF]
SOCOTEC Industrie report AME/KVR/03.01.019/N2 Calcul tranche
élastique – Author : S. Kevorkian – 20/11/2003 – In French –
(Axisymmetrical elastic modelling)
[03-04-EDF]
SOCOTEC Industrie report AME/KVR/03.01.019/N3 Calcul tranche
Mazars – Author: S. Kevorkian – 25/11/2003 – In French –
26
(Axisymmetrical non linear modelling)
[04-01-EDF]
SOCOTEC Industrie report AME/KVR/03.01.019/N5 Calcul élastique du
demimodèle 3D – Author : S. Kevorkian – 13/01/2004 – In French
(3D Elastic modelling of the ODPT)
[04-02-EDF]
SOCOTEC Industrie report AME/KVR/03.01.019/N6 - Modélisation 3D de
l’essai en air de la séquence 7 en considérant un comportement non linéaire du béton (modèle
d’endommagement de Mazars) - Author : S. Kevorkian – 18/02/2004 – In French –
(3D modelling of the ODPT taking into account non linear behaviour of the concrete
(Mazars isotropic damage model))
[04-03-EDF]
SOCOTEC Industrie report AME/KVR/03.01.019/N7 – Synthèse des études
du comportement mécanique de la maquette MAEVA lors d'un ess
ai en air au-delà de la
pression de dimensionnement - Author : S. Kevorkian – 15/03/2004 – In French –
(Synthesis of F.E. modelling studies of the MAEVA mock-up mechanical behaviour during
ODPT)
[05-01-EDF]
EDF report ENGSDS050092, Synthèse des contributions SEPTEN/DS au
Rapport de Synthèse du Projet Européen CONMOD - Author : Alexis Courtois – 2005
(Synthesis of contribution of SEPTEN/DS group to CONMOD final reports)
[02-01-SCT]
Scanscot Technology report 02401/TR-01 – CONMOD-Evaluation of the
containment at Barsebäck Unit 1. Step 1: Initial structural analysis – Authors: O. Jovall – et
al.– 20/12/2002 – In English
[02-02-SCT]
Scanscot Technology report no. 02403/TR-01 – Barsebäck NPP 1.
Pressurization and leakage test – Authors: T. Lindberg – et al.– 08/03/2002 – In English
[02-03-SCT]
Scanscot Technology report CON/3.1/SCTE/09 – CONMOD WP No. 3:
Pressure and leak-tightness programme Barsebäck NPP 1: Questionnarie: Pressure test and
concrete testing – Authors : O, Jovall – 12/07/2002 – In English
[02-04-SCT]
Scanscot Technology report CON/3.1/SCTE/01 – CONMOD WP No. 3:
Pressure and leak-tightness programme Barsebäck NPP 1: Övergripande aktivitetslista för
planering och genomförande av täthets- och deformationsmätning vid Barsebäck 1 – Authors :
O. Jovall – 05/04/2002 – In Swedish
(CONMOD WP No. 3: Pressure and leak-tightness programme Barsebäck NPP 1: Planning
of leak-tightness and deformation measurements, general activity list)
27
[04-01-SCT]
Division of Structural Engineering, Lund University report TVBK-3051 –
Assessment of material property data for structural analysis of nuclear containments – Authors:
S. Thelandersson – 20/12/2002 – In English
[05-01-SCT]
Scanscot Technology report 03402/TR-01 – ISP 48: Posttest analysis of the
NUPEC/NRC 1:4-scale prestressed concrete containment vessl model – Authors: O. Jovall –
M. Pålsson – B. Svärd – 23/02/2005 – In English
[05-05-SCT/EDF] Scanscot Technology / Electricité de Francenreport 02401/TR-05 CONMOD-Finite element analysis – Authors O. Jovall - A. Courtois - G. Heinfling 29/04/2005 - In English
Table I
Part of structure
Type of
Object of inspection
inspection
Pipe
entry x-ray
Voids in concrete
(concrete above)
Surrounding wall
x-ray
Radar
Impact
Echo
MASW
Weisele
probe
Cores
Wall thickness
Result of inspection
Positive (void  350 x 200 x
30 mm)
Positive ( 15 % greater than
design)
condition No voids observed
Concrete
(voids)
Cable duct filler (voids)
Cables inside ducts
Reinforcing details (size,
position (lateral), type)
Reinforcing condition
Reinforcing details (depth
and position)
No voids observed
No damage observed
As design
No damage observed
Depth greater than design
(200 contra 40 mm).
Position as design
Cable duct details (depth Depth greater than design
and position)
(360 mm contra 160 mm).
Position as design.
Concrete thickness
Suspect > design
Concrete condition
OK. Some near-surface
anomalies
Bond between liner and Yes
concrete
Concrete thickness
Suspect > design
Concrete condition
Typical LCW
Concrete wave velocities Typical LCW
Bond between liner and Yes
concrete
Concrete rH
Typical LCW
Concrete condition
Typical LCW. Near surface
(<250 mm) anomalies
28
Pipe entry
Peep hole Suspect void in concrete
& fibre- above
optic
Thickness of pre-stressed
wall
Condition of liner plate
(visual)
Weisele
RF in air-filled void
probe
Confirmed void
140 mm greater than design
OK.
82.1%
Table II
Observation
Void above Pipe
entry*
Category
CQ
Effect on ageing
Possible
Wall thickness*
AC
Affects drying
process
No
Specify possible damage
Corrosion to liner plate.
Unlikely to occur in the
future if no corrosion
observed after 30 years.
None
Cable duct
AC
position*
Reinforcing
AC
No
position*
Concrete
CQ
None identified
condition near
surface (slipforming)*
Concrete
A
Affects drying
Risk of cracking and
cracking (dryingprocess
increased creep and
shrinkage)*
shrinkage
Local high
AC, CQ
Affects drying
Increased risk of local
moisture content
process
corrosion and/or asr
CQ  Condition and Quality; AC  As-Built Compliance; A  Ageing
List of observations (B1) and possible effects on ageing. Note that these also have an
effect on structural behaviour* which is not defined here.
Table III
Method
Radar
SASW
Parameter
Depth
to
reinforcing (mm)
Depth to cable
ducts (mm)
Rayleigh
wave
velocity (m/s)
Predicted value
40 (mm)
Measured value
170-200 (mm)
160 (mm)
270-300 (mm)
2500 (m/s)
2100-2400(m/s)**
29
Impact Echo
Compression wave 4500 (m/s)
3900 (m/s)**
velocity
Radiography
Wall thickness
1000 (mm)
1150-1180 (mm)
Comparison of predicted (theoretical) and measured responses for various NDE methods
w.r.t. inital site investigation.
** Apparent velocities
Fig.1: The structure and its ageing processes are first described on the basis of
already-known facts and theory. Then, with the help of NDE and FEM the Profile
(ageing and condition) of the structure is established. This information is used to
model and understand the ageing processes and their possible consequences.
Modelling is used in the final stages to predict NDE responses, which describe the
condition and expected changes in condition with time.
30
Fig.2: Preliminary FEA models of structural behaviour and identification of critical
sections are made. The reactor containment is a cylindrical construction and is a
suitable object for preliminary analysis with the help of an axi-symmetrical model to
study the global behavior in the event of pressurisation. However, the axi-symmetrical
model must be complemented with a plane-strain model and a local three-dimensional
penetration model in order to understand the effects from penetrations and vertical
buttresses respectively. In the final analysis, and in design or structural verification, a
fully three-dimensional model should be used to accuratly capture the true behaviour
of a, pre-stressed reactor containment.
Fig. 3(a): Compliance –
reinforcing and cable ducts
radar
scan
showing Fig. 3(b): Compliance and
Condition – radiography of
containment wall to describe
internal details and condition
31
Fig. 3(c): Condition and ageing – seismic response of the containment wall, which is
dependant on the geometry, internal details, condition and ageing of the concrete. In this
figure the surface wave phase velocity for wavelengths up to equivalent wall thickness are
shown.
32
Fig. 4: Establishing a Profile of the containment structure by combining NDE and FEA
techniques. A series of normal NDE responses are established which describe trends that are
typical for the structure in question. In addition, a series of special responses may be
established for conditions that are exceptional, e.g. as determined by local environmental
conditions or damage that may have occurred.
33
Fig. 5(a): Graphic presentation of radiographic beam from Betatron at pipe entry (left); Betatron in
position on outside of containment (right).
Fig. 5(b): Radiographic image of wall above pipe entry showing
void.
Fig. 5(c): Void as seen from fibre-optic (left); liner plate in void Fig. 5(d): X-section of wall at pipe
(centre) with thin covering of material; injection (black) and entry showing void and fibre-optic
drainage (light) pipes inside void (right).
inspection point (arrow).
34
Fig. 6 : Graphic profile of the containment wall. The outside of the containment is to
the right. The concrete across the wall section varies in condition due to the effects of
drying, long-term hydration and also due to some damage caused by the slip-forming
process. In the outer (approx.) 200 mm there are ”tears” with inward and inclined
orientation caused by slip-forming and plastic settlement of the concrete. The strength
and elastic modulus of the concrete increases with depth from ”normal” at the surface
to very high at depth. The relative humidity of the concrete is at ambient condition
near the surface and increases to approx. 90% near the liner (average about 75%). The
upper curve in the figure shows the variation of dynamic elastic modulus and the
lower figure shows the relative humidity of the concrete.
Fig. 7: The profile of the structure is described by the four normal seismic responses.
These are for the lower and upper cylindrical wall and corresponding horizontal and
vertical profiles.
35
Fig. 8: Expected (upper) theoretical dispersion curves for the containment wall assuming
details as drawing and no other condition trend in the concrete. Compare with normal
(lower) dispersion curve measured for the lower cylindrical wall.
Fig. 9: (Top curve) SASW dispersion curve for horizontal profiles on upper cylindrical
wall. Average VR = 2428 m/s for all profiles. (Bottom curve) SASW dispersion curve for
horizontal profiles on lower cylindrical wall. Average VR = 2083 m/s for all profiles.
36
Fig. 10: Predicted surface wave dispersion curve based on nominal data (wall thickness
1000 mm). The surface wave velocity is assumed to be 2200 m/s based on preliminary
tests. Otherwise, the expected Rayleigh wave velocity was 2400 m/s or higher (Fig 1.) Here
the A0 dispersion curve is zoomed in on the portion that is most likely to be measured with
SASW or MASW. The dispersion curve should theoretically be straight all the way down
to a wavelength of about 0.44 m if the concrete has the same stiffness all the way through
the thickness as in reference model 1. The curve will drop at greater wavelengths due to the
effect of the wall thickness.
Fig. 11: The effect on the A0 dispersion curve from a change in shear wave velocity of the
first 100 mm from the outer wall surface. A low velocity zone near the surface is added to
reference model 1. The shear wave velocity of the first 100 mm from the outer surface of
the wall is set to vary from 2416 m/s to 1966 m/s in steps of 50 m/s. It can be concluded
that the phase velocity for a wide range of frequencies or wavelengths are affected by this
change.
37
Fig. 12: Containment and position of test – Pipe entry and surrounding wall. The method
of seismic data collection is shown to the right.
Fig. 13: MASW data was collected at Level 4 on the outside surface of the pre-stressed
wall (data in x-t domain). A total of 100 points were measured over a horizontal line 5 m
long. The results are shown above. From this the P and R wave velocities can be
calculated.The estimated P wave velocity is shown.
38
Fig. 14: Left: Phase velocity spectrum for the containment wall. The average phase velocity
(apparent surface wave) is around 2200 m/s. The horizontal axis represents a x-section of the
wall. Right: Standing waves induced in the containment wall (P-waves). This information is
available in the frequency:phase velocity data. The prominent standing waves have frequencies
from 2.4 to 2.7 K Hz, i.e. back side and liner echoes.
Fig. 15: Measured and theoretical phase velocity spectra for the containment wall.
39
Fig. 16: (Left upper and lower): Phase velocity/wavelength of measured and best fit
synthetic data. (Upper right): Mismatch as a function of parameter value. The x-axis
represents the pre-defined search range of each parameter. Convergence to the true value is
eventually achieved – in this case the shear wave velocity of around 2500 m/s. (Lower
right): best matching layer model for containment wall.
Fig. 17: The layer model obtained matches well the true conditions.
40
Fig. 18(a): Measurement of wall
thickness
using
HECR.
Calculation of thickness is
independant of wave velocities.
Fig. 18(b): Results of wall thickness measurements
using HECR. The apparent as-built wall thickness
data was used in the model. The thickness is thought
to vary from about 1100-1200 mm.
Fig.19: FE analysis of horisontal part of Barsebäck unit 1 containment wall, including pipe
entry. Analysis based on nominal data from drawings.
41
a) Yiedling in linerat P = 720 kPa
b) Yielding in liner at 690 kPa
Fig.20: FE analysis of Barsebäck Unit 1 containment wall, including pipe entry. (a)
Initial analysis based on nominal data (b) Analysis based on up-dated data obtained
using NDE, i.e. a void is introduced in to the model.
Fig. 21: Barsebäck Unit 1. Radial deformation during over-pressurization, with and
without increased wall thickness (egg-shaped and circular section respectively)
42
Fig. 22: Assumed wall profie consisting of a low and high velocity layer in the
pre-stresses outer part. Inset: plane section of outer 250 mm core from wall
showing cracking induced by slip-forming (lower).
Fig. 23: In the phase velocity/frequency figure the surface wave plot can be seen
as can the high-amplitude standing waves from the liner (F = 2.76 k Hz). In this
case it appears that there is no bond between the liner and the concrete.
43
Fig. 24: The figure represents a concrete wall with embedded reflector. This could be a
crack or a liner plate with poor bond. The impact at the surface causes wave reflection
which has been modelled in this example. The seismic events could be registered with an
array of transducers placed on the concrete surface. At the present a single transducer is
used.
Fig. 25: MAEVA mock-up
44
Fig. 26: MAEVA – Detection of crack events during over-design pressure test
ODPT-1 and ODPT-2.
Fig. 27: Global deformed shape of the main part of the mock up at applied pressure P
> Pdesign.
Fig. 28: Global 3D deformed shape of MAEVA.
45
Fig. 29: Quadrant 1 - Comparison between damage zone deduced from FEA and
acoustic events from NDE around the penetration projected on a plane.
Fig. 30: Quadrant 2 - Comparison between damage zone deduced from FEA and
acoustic events from NDE near the butress.
Fig. 31: Quadrant 3 - Comparison between damage zone deduced from FEA and acoustic
events from NDE in a standard area of the cylinder.
46