Professional Documents
Culture Documents
www.elsevier.com/locate/nel
Homogenisation method for the modal analysis of tube bundle with
uidstructure interaction modelling
Jean-Franois Sigrist
a,
, Daniel Broc
b
a
DCNS Propulsion, Service Technique et Scientique, 44620 La Montagne, France
b
CEA Saclay, Laboratoire dEtudes Mcanique et Sismique, 91191 Gif-sur-Yvette, France
Received 1 August 2007; received in revised form 20 November 2007; accepted 26 November 2007
Available online 28 January 2008
Abstract
The present paper is concerned with the modal analysis of a two-dimensional tube bundle with uidstructure interaction (FSI) modelling.
The numerical modelling of FSI effects is performed with a homogenisation approach, using a method whose principles have been presented in
a previous paper for the modal analysis of a nuclear reactor with internal structures and FSI modelling [J.F. Sigrist, D. Broc, Homogenisation
method for the modal analysis of a nuclear reactor with internal structures modelling and uid structure-interaction coupling, Nucl. Eng. Des.
237 (2007) 431440]. The method is adapted here in the case of tube bundle and compared with the classical approach, based on a direct nite
element discretisation of the coupled problem with all tubes modelling. The theoretical background of the method is recalled, the numerical
implementation in a nite element code is exposed and a comparison of the homogenisation and coupled methods is proposed in the
case of a 10 10 tube bundle. Calculation of eigenmode shapes, frequencies and effective masses with the two methods is performed; it is
concluded that: (i) the computational time is signicantly lowered when using the homogenisation method instead of the coupled method, since
the problem size is reduced by 90%; (ii) the tube bundle dynamic is described in a space-averaged manner, which is sufcient to account for
the main inertial coupling effects. Extension of the method to a three-dimensional case can now be considered; implementation of the method
in a commercial nite element code is also currently investigated.
2007 Elsevier B.V. All rights reserved.
Keywords: Fluidstructure interaction; Tube bundle; Modal analysis; Homogenisation method
1. Introduction
Dynamic analysis of tube bundle is of paramount importance
in the safety assessment of nuclear installations. These analyses
require in particular the calculation of frequency, mode shape
and effective mass of the systemeigenmodes. As uidstructure
interaction (FSI) effects can signicantly affect the dynamic be-
haviour of immersed structures, the numerical modelling of the
tube bundle has to take FSI into account. A complete modelling
of heat exchangers (including pressure vessel, tubes and uid)
is not accessible to the engineer for industrial design studies.
In the past decades, homogenisation methods have been stud-
ied and developed in order to model tubes and uid through
an equivalent continuous media, thus avoiding the tedious
n
[%]
n
[
%
]
Tubes in vacuum
Tubes in fluid
Fig. 4. Cumulated effective masses versus frequency coefcient for the simple
2 2 tube system.
computed using a discretisation of the whole structure and
uid domains.
Homogenisation method: the coupled uidtubes system
is modelled with an equivalent homogenised medium. Dis-
cretisation of the homogenised equations yields the added
mass matrix.
3.1. Coupled method
Fig. 5 gives the representation of a single tube immersed in
the uid domain and denes the geometrical parameters and
m
l
= 4.5810 kg n
l
l
u
l
,k
p(x
k
)
e
k
= 1.0010
3
kg/m
3
k
l
= 1.00
10
2
N/m
2R = 2.70 10
-2
m
I
l
=j( u
(k,l)
e
k
) n
l
on I
l
, (4)
where I
l
is the tubes outer contour. The variational formulation
of the coupled tubeuid problem reads
l,k
(m
l
u
(k,l)
+k
l
u
(k,l)
) ou
(k,l)
=
l
__
I
l
pn
l
e
k
dI
l
_
ou
(k,l)
ou
l
, (5)
where ou
l
stands for any tube virtual displacement eld, and
_
O
pop
c
2
dO +
_
O
p op dO
=j
l
__
I
l
u
(k,l)
e
k
dI
l
_
n
l
op op, (6)
2
Throughout the paper, the problem is described under the assumption
that the uid is compressible. In practice, compressibility effects can be
discarded since the tube bundle eigenmodes are rather low-frequency modes,
and computation will be performed with a high value of sonic velocity in
the uid. However, the coupled and homogenisation formulations are stated
within the framework of elasto-acoustic coupling which corresponds to what
is implemented in most nite element codes.
J.-F. Sigrist, D. Broc / Finite Elements in Analysis and Design 44 (2008) 323333 327
where op stands for any virtual pressure. Finite element dis-
cretisation of the above equations leads to the denition of the
FSI operator R, such that
l
__
I
l
pn
l
e
k
dI
l
_
ou
(k,l)
oU
T
RP,
l
__
I
l
u
(k,l)
e
k
dI
l
_
n
l
op oP
T
R
T
U. (7)
Discretisation of the left-hand side of Eq. (6) yields the uid
mass and stiffness matrices operator, dened as
_
O
pop
c
2
dO oP
T
M
F
P, (8)
_
O
p op dO oP
T
K
F
P. (9)
Discretisation of Eqs. (5) and (6) yields the following matrix
system [26]:
_
M
S
0
jR
T
M
F
_ _
U(t )
P(t )
_
+
_
K
S
R
0 K
F
_ _
U(t )
P(t )
_
=
_
0
0
_
. (10)
Eq. (10) above involves non-symmetric mass and stiffness ma-
trices; it is, however, possible to derive a symmetric formula-
tion by using the uid displacement potential eld as and
additional scalar variable of the uid problem. It is recalled
that is related to the pressure p within the uid domain from
the following relation p = j(j
2
/jt
2
). Using the (u, p, )
description of the coupled uidstructure problem yields the
following symmetric formulation [26]:
_
M
S
0 jR
0 0 M
F
jR
T
M
F
jK
F
__
U(t )
P(t )
(t )
_
+
_
K
S
0 0
0 1/jM
F
0
0 0 0
__
U(t )
P(t )
(t )
_
=
_
0
0
0
_
. (11)
3.2. Homogenisation method
Assembling the uid and uid/structure interaction matrices
requires a nite element mesh of the uid domain, which can
be a rather tedious task for complex geometries. Besides, for
tube bundle modelling, such an approach reaches its limits as it
is not possible to account for the 5000 tubes or so of the steam
generator with a nite element method. Homogenisation of the
uid+tubes domain provides a more straightforward approach
of the problem. The present method is based on the description
of the uid forces acting on a single tube in an elementary
uid cell, as depicted by Fig. 6. In the proposed method, is
calculated according to the following equation [11,21]:
=m
A
u
S
+(m
A
+j|O
S
|)
U
F
, (12)
where m
A
and j|O
S
| are the tubes added mass and displaced
mass, respectively. Eq. (12) states that at the local scale (i.e. in
the elementary periodic cell), uid forces on the tube depend
Fig. 6. Elementary cell O
T
: elementary tube O
S
immersed in uid elementary
domain O
F
.
only on 2 degrees of freedom, namely the tube displacement
u
S
and the uid mean displacement U
F
dened by
U
F
=
O
S
O
T
u
S
+
1
O
T
_
O
F
u
F
dO
F
. (13)
It can be demonstrated that U
F
also represents the displace-
ment of the uid at the cell boundary. The uid motion within
the uid cell is governed at the local scale by the following
equation:
_
O
F
j u
F
dO
F
=, (14)
where is the force exerted by the tube on the uid (since the
force exerted by the uid on the tube is , see Eq. (12)) and is
the pressure force exerted on the uid domain. is calculated
from the pressure gradient as =|O
T
|p where |O
T
| is the
elementary cell volume. Combining this latter equation with
Eqs. (12) and (13) yields
_
O
F
j u
F
dO
F
=j|O
T
|
U
F
j|O
S
| u
S
(15)
and
_
O
F
j u
F
dO
F
=|O
T
|p +m
A
u
S
(m
A
+j|O
S
|)
U
F
. (16)
From Eqs. (14) and (15), it can be inferred that the pressure
gradient in the cell is calculated according to
j
U
F
=(1 J)p +jJ u
S
(17)
where J is the connement ratio, dened as
J =
m
A
+j|O
S
|
m
A
+j(|O
S
| +|O
T
|)
. (18)
328 J.-F. Sigrist, D. Broc / Finite Elements in Analysis and Design 44 (2008) 323333
Fig. 7. Homogenised uid (uid +tubes) and elastic structure.
Eq. (12) is the starting point of the proposed homogenisation
method, which can be applied in the general frame depicted
by Fig. 7. Let O be the uid and tubes domain, which will
be described with their following degrees of freedom: p stands
for the mean pressure eld in the uid and u
S
for the tube
displacement tube, now supposed to vary continuously in the
homogenised domain. m
S
and k
S
are the mass and stiffness
characteristics of the tubes, j and c the density and sound wave
speed in the uid. Let 2 be an elastic structure in contact with
the tubes + tubes domain and let u
2
denote the displacement
eld in 2.
In order to derive a symmetric formulation of the ho-
mogenised tubes + uid/structure system, the displacement
potential in the homogenised uid domain [ is introduced and
dened as
p =j
j
2
[
jt
2
, (19)
so that Eq. (17) now reads
j
U
F
=j(1 J)
[ +jJ u
S
. (20)
The constitutive law of the uid is on the other hand
p =jc
2
U
F
. (21)
The variational formulation of the uid problem is derived from
Eqs. (19) to (21) and reads
1
j
_
O
pop
c
2
dO +
_
O
[op
c
2
dO =0, (22)
for any virtual pressure eld op and
_
O
po[
c
2
dO j
_
O
(1 J)[ o[dO
+j
_
I
o[ u
2
n dI jJ
_
O
u
S
o[dO =0 (23)
for any virtual displacement potential eld o[.
The variational formulation of the structure problem is de-
rived from the equation of motion of the tubes and of the elastic
structure 2 and reads
_
2
j
2
u
2
ou
2
d2 +
_
2
o(u
2
) c(ou
2
) d2
+j
_
I
n ou
2
[dI =0 (24)
for any virtual displacement of the structure ou
2
and
_
O
m
S
u
S
ou
S
dO +
_
O
k
S
u
S
ou
S
dO
=
_
O
j(J |O
S
|/|O
T
|) u
S
ou
S
dO
+jJ
_
O
[ ou
S
dO (25)
for any virtual displacement of the tubes ou
S
.
Finite element discretisation of Eqs. (22)(25) yields
M
S
+M
S
0 0 jJC
0 M
2
0 jR
0 0 0 M
F
jJC
T
jR
T
M
F
j(1 J)K
F
U
S
(t )
U
2
(t )
P(t )
(t )
K
S
0 0 0
0 K
2
0 0
0 0 1/jM
F
0
0 0 0 0
U
S
(t )
U
2
(t )
P(t )
(t )
0
0
0
0
. (26)
In Eq. (26), the various mass and stiffness operators for uid
and structure problem are dened in a classical manner; see, for
instance, [27]. The following matrices are dened in a specic
manner for the homogenisation formulation:
Structure/uid coupling matrix:
_
I
[n ou
2
dI oU
2
R
,
_
I
jo[ u
2
n dI o
T
jR
T
U
2
. (27)
Tube/uid coupling matrix:
_
O
u
S
o[dO o
T
C
T
U
S
,
_
O
[ ou
S
dO oU
T
S
C
. (28)
Tube added mass matrix:
_
O
j(J |O
S
|/|O
T
|) u
S
ou
S
dO oU
T
S
M
U
S
. (29)
R and C are assembled with elementary matrices r
e
and
c
e
obtained from uid and structure element coupling (see
Fig. 8) and uid and tube elements coupling (see Fig. 9).
The homogenisation formulation does not need modelling all
the tubes of the bundle and requires a coarser nite element
mesh than the coupled method, as will be highlighted in the
J.-F. Sigrist, D. Broc / Finite Elements in Analysis and Design 44 (2008) 323333 329
e
p,
u
e
Coupling element
Structure element
Fluid element
Fig. 8. Fluid and structure elements coupling.
p,
u
S
e
e
Coupling element Tube element
Fuid element
Fig. 9. Fluid and tube elements coupling.
next subsection. As a consequence, the method is expected
to be efcient from the computational point of view and
convenient from the engineering point of view.
It is recalled that the homogenisation method provides a con-
sistent calculation of the effective masses since it complies with
the mass conservation condition, which reads
n
j
n
=
T
M =m (30)
for any direction , with m being the total mass of the sys-
tem (structure, tubes and uid). Demonstration of such prop-
erty with the mass matrix M of the homogenised tubes +
uid/structure system (see Eq. (26)), can be performed for the
case of tubes in the same manner as for the case of reactor
internals [21].
4. Application to a 2D tube bundle
4.1. Application example: L L tube bundle
A numerical application of the methods exposed in the pre-
vious subsection is now considered. The method is tested with
the CASTEM code [28], which is a research-oriented nite ele-
ment code. However, as the method has been designed in order
to calculate the uid/tubes interaction matrix and the modied
uid matrices in a straightforward manner using existing cou-
pling operators, implementation of the method within industrial
nite element codes, such as the ANSYS code, is possible and
will be investigated for future industrial applications.
In the present study, a simple LL tube bundle embedded in
a circular structure (as sketched by Fig. 10) is studied. The tube
and uid geometry and physical properties are dened in Fig. 5:
m
l
= 4.58 kg, k
l
= 100 N/m, R = 1.35 cm, j = 1000 kg/m
3
.
The radius of the outer structure is R
= 1.05 (LR/
2) =
2.23 m (L = 10 is the number of tubes, R is the radius of
a single tube). The pitch to diameter ratio is T/2R = 90%
(where T = 3 cm is the tube spacing), which correspond to a
rather conned tube bundle. The outer bundle structure is of
negligible mass (m
2
>m
l
) and of large stiffness (k
2
?k
l
) in
comparison to the tube mass and stiffness. Since all tubes have
identical properties, the eigenfrequencies of any tube in the
two spatial directions are identical: f
0
=(2)
1
k
l
/m
l
, i.e.
f
0
0.75 Hz.
Computation of the frequencies and effective masses for each
eigenmode with the coupled and homogenisation methods are
now compared. Fig. 11 gives a representation of the nite el-
ement mesh of the uid domain for the 10 10 tube bundle
with each method.
On the one hand, calculation of the added mass operator with
the coupled method requires the discretisation of the uid do-
main, taking into account the geometrical details related to the
presence of the tubes. The nite element mesh is composed of
4949 uid nodes with 2 degrees of freedom per node,
1600 tube nodes (tube outer boundary) and 160 structure
nodes of 3 degrees of freedom per node,
100 tube nodes (tube centre) of 2 degrees of freedom,
5480 =(1600 +160) 3 +100 2 linear constraints which
model the tube and the outer envelope as rigid structures.
The linear constraints weight for 5480 additional degrees of
freedom (Lagrange multipliers).
The problem is thus dened by 20 858 degrees of freedom and
6709 nodes.
On the other hand, the homogenisation method requires:
A nite element mesh of the homogenised uid +tubes do-
main using uid nite elements coupled with tube punctual
elements (see Fig. 9) and uid and structure elements for
the outer bundle envelop (see Fig. 8). Fig. 11 sketches the
nite element mesh of the homogenised domain, using one
homogenised element for one tube.
A nite element mesh for the elementary cell, in order to
evaluate the connement ratio (see Fig. 12). This latter is
calculated by solving the elementary problem p =0 in O
F
,
with boundary conditions jp/jn=j u
S
n on jO
S
(coupling
with the tube) and jp/jn = 0 on jO
F
(outer boundary of
the elementary cell with symmetry condition) and p = 0 on
jO
F
o
(outer boundary of the elementary cell with periodicity
condition), see Fig. 6. In this case, uid forces are calculated
from =m
A
u
S
+(m
A
+j|O
S
|)
U
F
with U
F
deduced from
Eq. (12) since the uid acceleration eld is u
F
= 1/jp.
m
A
is then deduced from the uid forces, yielding J using
Eq. (18).
330 J.-F. Sigrist, D. Broc / Finite Elements in Analysis and Design 44 (2008) 323333
Fig. 10. Tube bundle conguration: LL tubes embedded in circular elastic structure, coupled with a compressible uid and subjected to a dynamic loading.
Coupled method Homogenisation method
Fig. 11. Finite element mesh of the 10 10 tube bundle with tubes modelling (coupled method) and without tubes modelling (homogenisation method).
Elementary cell mesh Elementary calculation
Fig. 12. Elementary cell mesh and elementary pressure eld calculation.
The nite element mesh produced for the homogenised model
is composed of:
217 uid nodes with 2 degrees of freedom per node,
40 structure nodes (outer bundle envelop) and 400 tubes +
uid nodes (4 nodes per each of 100 homogenised ele-
ments) with 3 degrees of freedom per node,
100 tube nodes (tube centre) of 2 degrees of freedom,
1520 =(400 +40) 3 +100 2 linear constraints which
model the tube and the outer envelope as rigid structures.
The linear constraints weight for 1520 additional degrees
of freedom (Lagrange multipliers).
The problem is thus dened by 3474 degrees of freedom and
657 nodes, which roughly corresponds to 15% of the problem
size with the coupled method.
4.2. Comparison of the homogenisation and coupled methods:
modal analysis of the tube bundle
Calculation of the eigenmodes of the tube bundle is ob-
tained by solving the eigenvalue problem (c
2
M+ K) with
the mass and stiffness matrices given by Eq. (10) with the cou-
pled method or by Eq. (20) with the homogenisation method.
J.-F. Sigrist, D. Broc / Finite Elements in Analysis and Design 44 (2008) 323333 331
Fig. 13 compares the cumulated effective masses and num-
ber of modes versus the frequency ratio for the coupled and
homogenisation calculation (the positive scale refers to the
-50%
-40%
-30%
-20%
-10%
0%
10%
20%
30%
40%
50%
75% 80% 85% 90% 95% 100%
f
n
/f
o
[%]
n
[
%
]
-200
-150
-100
-50
0
50
100
150
200
N
n
[
-
]
Effective masses
Number of modes
Coupled
method
Homogenisation
method
Fig. 13. Cumulated modal masses and number of modes versus eigenfre-
quencies for the 10 10 tube bundle. Comparison of the coupled and ho-
mogenisation methods.
Table 1
Comparison of direct and homogenisation method
Frequency ratio Direct Homogenisation
[
min
(%) 79.28 80.28
[
max
(%) 98.75 98.50
Tube displacement field Fluid pressure field
Coupled method = 80.41%
Tubedis placement field Fluid pressure field
Homogenisation method = 80.34 %
Fig. 14. Eigenmode calculation with coupled and homogenisation methods. Example of global tube mode.
coupled method, the negative scale to the homogenisation
method). The 200 tubes vibration modes account for 50% of
the total system mass. The two outer structure vibration modes
at higher frequency account for the other half of the mass.
The homogenisation approach gives satisfactory results when
compared to the direct method: the maximum and minimum
frequency ratios are correctly evaluated with the homogenisa-
tion approach, see also Table 1.
Figs. 14 and 15 compare eigenmode shapes in terms of tube
displacement eld and uid pressure eld for a global mode
(i.e. mode with important tube motions, inducing high added
mass effects) and a local mode (i.e. mode with limited tube
motions, inducing low added mass effects), both being calcu-
lated with the coupled and homogenisation method. It can be
concluded that the homogenisation method is able to predict the
main physical features of the tube motions with FSI modelling.
It should be nonetheless noted that the homogenisation
method produces a signicant number of modes at same fre-
quency, especially for low added mass modes frequency (see
in Fig. 13 the curves that represent the number of modes
versus frequency). The corresponding eigenmodes are charac-
terised by in-phase tube motions (i.e. limited uid motion),
these modes satisfy the periodicity condition between two ad-
jacent cells. As a matter of fact, the homogenisation method
is not able to distinguish these modes and thus gives identical
frequencies for these modes, while the coupled method gives
332 J.-F. Sigrist, D. Broc / Finite Elements in Analysis and Design 44 (2008) 323333
Tube displacement field Fluid pressure field
Coupled method = 92.31%
Tube displacement field Fluid pressure field
Homogenisation method = 94.32%
Fig. 15. Eigenmode calculation with coupled and homogenisation methods. Example of local tube mode.
Table 2
Total computation time for direct and indirect methods
Method Direct Homogenisation
Time (s)] 2150 295
Nodes in mesh () 6700 660
Problem size () 21000 3500
Extracted modes () 200 200
different frequencies. However, this is a minor drawback of
the homogenisation approach since the modes of interest here
have negligible or even null effective masses: seismic or shock
analysis does not require that the frequencies of these modes
be calculated in an exact manner.
It is, however, checked that the predominant mode in terms
of effective mass is correctly evaluated with the homogenisa-
tion method: this mode is calculated at [ = 85.98% with the
homogenisation approach, with an effective mass j=34.3% of
the total mass, while the predominant mode characteristics are
predicted as [ =85.69% and j =31.7% by the direct method.
In addition, efciency of the homogenisation method is
clearly highlighted when referring to computational time, see
Table 2. Total computation time for the calculation of the 200
eigenmodes of the tubes with the homogenisation method is
about 10% of the required time with the direct method, as a
result of a decrease in the problem size.
5. Conclusion
A homogenisation method has been developed in order to
performthe modal analysis of a tube bundle with FSI modelling.
The homogenisation approach is based on a description of the
uid+tubes system through an equivalent continuous medium,
characterised by a set of dynamic equations which describe the
behaviour of the tubes and the uid from a global point of view.
The nite element discretisation of the uid + tubes/structure
coupled system can be performed in a straightforward manner
using a coarser mesh than the one which would be required by a
coupled uid/structure analysis. From the engineering point of
view, the following conclusions have been drawn by applying
the method on a simple 10 10 tube bundle:
(i) The computational time is signicantly lowered when us-
ing the homogenisation method instead of the coupled
method, since the problem size is reduced by 90%.
J.-F. Sigrist, D. Broc / Finite Elements in Analysis and Design 44 (2008) 323333 333
(ii) The tube bundle dynamic is described in a space-averaged
manner, which is sufcient to account for the main inertial
coupling effects. In particular, the values of the minimum
and maximum eigenfrequencies of tubes coupled with the
uid are correctly evaluated by the homogenisation formu-
lation, as well as the frequency and effective mass of the
predominant mode.
(iii) The homogenisation approach fails somehow to make a
difference between local tubes modes with low added
mass effects. However, for seismic or shock analysis, this
has insignicant inuence since the modes of interest have
negligible contribution in the dynamic behaviour of the
system.
(iv) The simple tube bundle under study in the present analysis
is characterised by a high geometrical connement since
the pitch to diameter ratio is 0.9, but the resulting added
mass effects are rather limited, since the estimated decrease
in frequency is about 520% as consequence of inertial
coupling.
Extension of the method to a complete three-dimensional case
is currently under development: 3D coupling effects will then
be investigated. Besides, application of the homogenisation
method on an industrial structure can now be considered and
will be presented in future publications.
Acknowledgement
The authors gratefully acknowledge M.M. Guillop and
Galvicic from STXN for continuously supporting the scientic
exchanges between CEA and DCN with R&D programs on
uidstructure interaction modelling.
References
[1] A. Bensoussan, J.L. Lions, G. Papanicolaou, Asymptotic Analysis for
Periodic Structures, North-Holland, Amsterdam, 1978.
[2] E. Sanchez-Palencia, Non Homogenous Media and Vibration Theory,
Springer, Berlin, 1980.
[3] C. Conca, J. Planchard, M. Vanninathan, Fluids and Periodic Structures,
Masson, Paris, 1995.
[4] D. Brochard, F. Gantenbein, R.J. Gibert, Homogenisation of tube bundle.
Application to LMFBR core analysis, in: 9th Conference on Structural
Mechanic in Reactor Technology, Lausanne, 1987.
[5] R.J. Zhang, A unied homogenisation mode of beam bundle in uid, J.
Pressure Vessel Technol. 120 (1998) 5661.
[6] K. Cheval, Modlisation du comportement sismique de structures
multitubulaires baignes par un uide dense, Ph.D. Thesis, Universit
Paris VI, 2001.
[7] R.J. Zhang, W.Q. Wang, S.H. Hou, C.K. Chan, Seismic analysis of
reactor core, Comp. Struct. 76 (2001) 13951404.
[8] D. Broc, J.C. Queval, E. Viallet, Seismic behaviour of PWR reactor
cores: uidstructure effects, in: 17th Conference on Structural Mechanic
in Reactor Technology, Prague, 2003.
[9] D. Brochard, F. Gantenbein, R.J. Gibert, Modelling of the dynamic
behaviour of LWR internals, in: 9th Conference on Structural Mechanic
in Reactor Technology, Lausanne, 1987.
[10] D. Broc, J.F. Sigrist, Fluidstructure interaction: numerical validation of
a homogenisation method, in: Pressure Vessel and Piping, Vancouver,
2006.
[11] J.F. Sigrist, D. Broc, A homogenisation method for the modal analysis
of a nuclear reactor with uidstructure interaction, in: Pressure Vessel
and Piping, Vancouver, 2006.
[12] Y. Shinohara, T. Shimogo, Vibrations of square and hexagonal cylinders
in a liquid, J. Pressure Vessel Technol. 103 (1981) 233239.
[13] U. Schumann, Homogenised equations of motion for rod bundle in the
uid with periodic structure, Ingenieur-Archiv 50 (1981) 203216.
[14] J. Planchard, Vibration of nuclear fuel assemblies: a simplied model,
Nucl. Eng. Des. 86 (1985) 383391.
[15] J. Planchard, Modelling the dynamic behaviour of nuclear reactor fuel
assemblies, Nucl. Eng. Des. 90 (1985) 331339.
[16] J. Planchard, Global behaviour of large elastic tube bundle immersed in
a uid, Comput. Mech. 2 (1987) 105118.
[17] L. Hammami, Etude de linteraction uide/structure dans les faisceaux
de tubes par une mthode dhomognisation, Application lanalyse
sismique des cWurs RNR, Ph.D. Thesis, Universit Paris VI, 1991.
[18] E. Jacquelin, D. Brochard, C. Trollat, L. Jzquel, Homogenisation
of non-linear array of conned beams, Nucl. Eng. Des. 165 (1996)
213223.
[19] R.J. Zhang, Structural homogenized analysis for nuclear reactor core,
Nucl. Eng. Des. 183 (1998) 151156.
[20] R.J. Zhang, A beam bundle in a compressible inviscid uid, J. Appl.
Mech. 66 (1999) 546548.
[21] J.F. Sigrist, D. Broc, Homogenisation method for the modal analysis of
a nuclear reactor with internal structures modelling and uid structure-
interaction coupling, Nucl. Eng. Des. 237 (2007) 431440.
[22] J.F. Sigrist, D. Broc, C. Lain, Dynamic analysis of a nuclear reactor
with uidstructure interaction. Part I: seismic loading, uid added mass
and added stiffness effects, Nucl. Eng. Des. 236 (2006) 24312443.
[23] J.F. Sigrist, D. Broc, C. Lain, Dynamic analysis of a nuclear reactor with
uidstructure interaction. Part II: shock loading, uid compressibility
effects, Nucl. Eng. Des. 237 (2007) 289299.
[24] J. Planchard, Eigenfrequencies of a tube bundle placed in a conned
uid, Comp. Meth. Appl. Mech. Eng. 30 (1982) 7593.
[25] F. Axisa, Modelling of Mechanical Systems, Vol. 3: FluidStructure
Interaction, Elsevier, Amsterdam, 2006.
[26] H.-J.P. Morand, R. Ohayon, FluidStructure Interaction, Wiley, New
York, 1995.
[27] J.F. Sigrist, Symmetric and non-symmetric formulation for the modal
analysis of uidstructure interaction problem with a commercial nite
element code, in: Pressure Vessel and Piping, Vancouver, 2006.
[28] P. Verpeaux, A. Millard, T. Charras, A. Combescure, A modern approach
for computer codes for structural analysis, in: 15th Conference on
Structural Mechanics in Reactor Technology, Anaheim, 1989.