MATHEMATICAL METHODS, COMPUTATIONAL TECHNIQUES, INTELLIGENT SYSTEMS

Evaluation of damping in dynamic analysis of structures

TEPES ONEA FLORIN, GELMAMBET SUNAI
Faculty of Civil Engineering
University Ovidius of Constantza
Constantza, 22B, Unirii Street,
ROMANIA
tflorin@univ-ovidius.ro; gelmambets@univ_ovidius.ro

Abstract: From physical point of view, the dumping represents the soil seismic excitation energy taken over process
through internal absorption, rubbed between existent layers, as and cracks on rocky foundations. Generally, on heavy
dams dynamic analysis a viscous dump is considered, proportional to the deformation speed.
The dumping can be evaluated on experimental bases or on environmental conditions measurements. The latest
determine higher values of dumping elements. This could be explained by the local factors influence, which is not
possible to be modelled as backlash treatment, foundation ground characteristics, and the concrete technology. This
represents atypical dissipate phenomenon.
A major influence is done by the excitation level as real seism or experimental excitation.
In the present work we establish the influence of the dissipative effect of the backlash on concrete blocks. The
backlash finite elements modeling make it possible, studying different situations, as rub effect, cohesion effect, seismic
action on varying directions with the same accelerogram of 0.4g.
The studied blocks have the same dimensions, the relative displacement being obtained by foundation stiffness
modified under two block parts.

Key-Words: - Rayleigh, spatial mesh, excitation, dumping, dissipation

1 Introduction was pick Eb  300000daN / cm 2 and for the

The damping matrix is obtained from the Cauchy
foundation E f  150000daN / cm , with the dam’s
2
sequence:
p 1 high of 30 m and the slope 1    0,5 . The dam is
C  M  ak ( M 1 K ) k made by two plots of 15 m width, separated by a
k 0
backlash of 1 mm. The two plots adjacent nodes,
where the coefficients ak , k =1, 2, ..., p are obtain from corresponding to the space, have the same quota on x
p simultaneous equations: and z axis. These nodes can be connected with by
springs in order to model the friction. We can notice that
1  a0  in the case of plane mesh:
i    a1i  a2i 2  ...  a p 12 p 3 
2  i  - the fundamental vibration mode is flexural on
upstream-downstream;
For p = 2: - the second mode is flexural on the high of the dam;
C M  K , - the others modes are of torsion.

where α and β are constants that can be obtained from

two dumping ratios of two different frequencies.
This study is made for an idealized symmetric
concrete dam. It was used two calculus models for the
dam-foundation ensemble; a plain one and spatial model,
with simultaneous calculation.
2 Problem formulation
The plane finite element mesh is made by 80
quadrilateral elements for the foundation and 56
elements for the dam. The elasticity modulus for the dam
2 Problem Solution
Between the two considered block parts there is a
relative motion relieved by superior mode shapes. The
phenomenon is more complex because on the separation
dam parts surface, the rub and strike also appear.
This study points out only the rub phenomenon.
To simulate the rub energy dissipation, the interface
nods were connected with Truss finite elements. These
elements withstand to the block parts relative motion
working like springs.
The relative motion phenomenon is due to the

ISSN: 1790-2769 67 ISBN: 978-960-474-188-5

 MATHEMATICAL METHODS, COMPUTATIONAL TECHNIQUES, INTELLIGENT SYSTEMS

difference in phase resulted by different high, modified vibration mode.

excitation and different structural properties. The calculus was made in both cases by finite
For relative motion between dam parts the calculus is elements, for the first 10 vibration modes. If we couple
simpler, the foundation elasticity modulus being 1  i and solve the systems of equation,  and 
changed. coefficient are obtained.
For comparison, a numerical integration without So for plane discrete mesh
springs limitation was done. The results are presented in
the following figures.   1, 44,   1,51E  3 , for ω1 and 3
The rub between the dam parts is uniformly and
distributed on backlash surfaces. This distributed force is   1.97,   5.66 E  4 , for ω1 and ω10 .
considered hypothetically concentrate in nods. We can observe that the effect of the mass matrix
As it was presented, Truss elements model the rub increases and the effect of stiffness matrix decreases in
phenomenon. Changing the springs stiffness in accord the same direction with the increasing of the second
with feedback structural response, different results were frequency taken into account.
obtained as are presented in following tables (for a In the case of spatial mesh, the fundamental vibration
crowning node). It is noticed that the structural response mode is reduced, 1  13, 48rad / s and for the plane
is almost identical for a large interval of the springs
stiffness. It was chosen a 0.1m2 springs area. mesh, 1  21.3rad / s .
It is noticed that rub force could not overtake a limit This difference appears because in spatial mesh we
value, and the x direction maximal displacement of the also take into account of the torsion vibration modes.
crowning node number 323 become 0.429E-2 in For the spatial mesh and for the frequency ω1 and
comparison with 0.438E-2, which is the value 3 , α = 0,893 and   2,5 E  3 , while if use the
corresponding to no backlash energy dissipation
frequency 1 and 9 ,   1, 063 and   1.56 E  3 .
hypothesis. These displacements are measured by
comparison to reference base. In the case of spatial discrete mesh we can notice a
For the spatial mesh the first vibration mode implies a growing influence of mass matrix and a decresind
symmetrical displacement and flexural on upstream- influence of stiffness matrix, in the same time as the
downstream direction of the two plots, and for the pulsation value growing.
second mode an anti - symmetrical displacement. The variation of α and β factors is much reduced
To start with the 6’s vibration mode, it appear also a when the spatial mesh is used.
rotation of the two plots, implying a relative moving of For the spatial mesh case we obtain a mean value
the plots surfaces in the space between them.  Ed  0,99 and for the plane mesh,  med  1.72 . All
The purpose of this work is to study the effect of the the results are presented in Tables 1 and 2.
superior vibration modes on the energy dissipation in the
backlash between the two plots. Table 1 Plane mesh
Since both masses and stiffness matrix are Num. Ωi (rad/s)
orthogonal, the damping matrix is also orthogonal. From 1 21.3
orthogonality condition we obtain: 2 38.59
3 44.43
iT ( M   K ) j  2 , 4 79.44
5 87.59
where i ,  j are eigen vectors, ωi is circular frequency, ξi 6 96.43
- fraction of critical dumping. 7 109.0
The equation becomes: 8 118.8
9 121.3
  i 2  2ii . 10 155.1
1  i α β
A parametric study for plane and spatial dam-
1+3 1,44 1,519E-3
foundation discrete mesh was done, in order to
1+5 1,71 9,17E-4
determine the influence of the coefficients α and β.
1+7 1,78 7,67E-4
Critical damp fraction was taken as constant,   0, 05 , 1+10 1,97 5,66E-4
for all the vibration modes, because a massive structure
as a concrete dam could obtain, after the structure
excitation (with a value lower that the seismic one), only
a fraction of critical dumping corresponding to the first

ISSN: 1790-2769 68 ISBN: 978-960-474-188-5

 MATHEMATICAL METHODS, COMPUTATIONAL TECHNIQUES, INTELLIGENT SYSTEMS

Table 2 Spatial mesh Table 4 Plane mesh-stress comparison σ(-)daN/cm2

Num. i Without daping -18.2
1 1 1  3 -10.93
2 17,44
3 26,49 1  5 -10.92
4 34,28 1  7 -10.91
5 37,21
6 42,17 1  10 -10.75
7 45,64
8 49,8 Table 5 Spatial mesh-displacements comparison
9 50,25
1  i α β Without daping 0.433E-2
1+3 0,893 2,5E-3 1  3 0.3319E-2
1+5 0,989 1,97E-3 1  5 0.3326E-2
1+7 1,041 1,689E-3
1+9 1,063 1,567E-3 1  7 0.3325E-2
1  10 0.3326E-2
The calculus was resumed for a critical dump fraction
of 8% in which case we obtain for spatial mesh and the
Table 6 Spatial mesh – stress comparison
pair ω1 and ω9 the following results: α=1.69 and
β=2.51E-3, which doesn’t differ very much to the case of Without
-12.03
critical dump fraction of 5%. daping
It is obvious that only for a spatial discrete mesh the 1  3 -9.2
obtained results are close to reality. The superior modes
used in the case of spatial mesh have no significant 1  5 -9.2
effect on the coefficients α and β, as it is presented in 1  7 -9.2
Table 2.
It is noticed that in the same time with the increase of 1  10 -9.2
the frequency the mass matrix effect increase to and also
the stiffness matrix effect decrease. So, we can say that Table 7 displacement comparative values for the spatial
the stiffness matrix effect connected with frequency is mesh for two critical dumping ratios of 5% and 8%.
major. (node 323)
After coefficients calculus, the dam response was 5% Rayleigh 8% Rayleigh Diference %
analyzed at the same excitation with and without (x)-0.3878E-3 -0.3623E-3 -6.57
damping matrix effect. It was followed by the effect of (y)-0.8799E-2 -0.7545E-2 -14.25
using vibration modes 1-3, 1-5, 1-7, 1-10 in mass end (z) -0.2340E-2 -0.2272E-2 -2.9
stiffness matrix coefficients on the stress and
displacement response. The displacement comparative graphics are
The stress and displacement values for different presented in Figures 1 and 2 and the stress calculus
pairs of coefficient α and β, for plane and spatial mesh points in Figure 3.
are presented in Tables 3 - 6.

Table 3. Plane mesh - displacements comparison

(node 161)
Without daping 0.728E-2m
1  3 0.4200E-2
1  5 0.4184E-2
1  7 0.4178E-2
1  10 0.4104E-2
Figure 1. Displacement comparison for spatial mesh

ISSN: 1790-2769 69 ISBN: 978-960-474-188-5

 MATHEMATICAL METHODS, COMPUTATIONAL TECHNIQUES, INTELLIGENT SYSTEMS

Rayleigh model, the difference, would be of 25%. If

cohesion influence is considered, the procentage
difference obtained is of 10%.

References:
[1] R.W. Clough, J. Penzien, Dynamics of Structures,
McGraw-Hill Book Co, New York, 1993
[2] K.J. Bathe and E.L. Wilson, Numerical Methods in
Finite Element Analysis, Englewood Cliffs, N.Y.,
Prentice - Hall, 1976
[3] D. Stematiu, Calculation of hydro structures by finite
Figure 2.  ( ) for plane mesh and spatial mesh element, Technical Publishing House, Bucharest,
1988 (in Romanian)

Figure 3.  ( ) for plane mesh and spatial mesh

Figure 4. Modal analyses.

Spatial mesh for a concrete dam.
The six vibration mode (Freq=42.17rad/s; T = 0.149s)

It is also noticed that as well as for the plane and

spatial discrete mesh, the stress and efforts values are
almost similar for all the pairs of coefficients α and β, if
damping matrix is used. Also, for Rayleigh models, only
the first three vibration modes are required.
Major response differences of 14% are obtained
only between 2 critical dams of 5% and 8%.

4 Conclusion
The modification of the displacement response is of 3%,
considering the dissipation through friction. Using the

ISSN: 1790-2769 70 ISBN: 978-960-474-188-5