MS02-504 Earthquake Behavior of Natural Draft Cooling Towers - Determination of Behavior

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 275

Leuven, Belgium, 4-6 July 2011


G. De Roeck, G. Degrande, G. Lombaert, G. Müller (eds.)
ISBN 978-90-760-1931-4

Earthquake Behavior of Natural Draft Cooling Towers – Determination of Behavior


Factors with Special Regard to Different Types of Supporting Column Systems
Dr.-Ing. Christian Lang1
1
L.A.W. Ingenieure GmbH & Co. KG, Lindenstraße 13, D-67433 Neustadt, Germany
email: christian.lang@law-ing.de

ABSTRACT: In this contribution, methods for computation of behavior of large RC Natural Draft Cooling Towers subjected to
earthquake ground motion are presented. The structural response is affected by non-linear behavior due to foundation uplift,
cracking of concrete shell and mainly by non-linear behavior of concrete columns. Algorithms for computation of linear and
non-linear structural dynamic response are explained. Shell ring elements in conjunction with column macro supporting ring
elements are deployed which can be advantageously used for computation of structural response of shells of revolution
subjected to ground motion. In engineering practice, behavior factors are used in structural design to account for non-linear
structural behavior and system ductility. For design of natural draft cooling towers subjected to earthquake, the behavior of
supporting column systems plays a very important role in structural response. The relation between strength and flexibility of
different supporting column types and the corresponding influence on structural behavior are discussed within this contribution.

KEY WORDS: Cooling towers; Ground motion; Behavior factor.

1 INTRODUCTION where the structure is excited dynamically by ground motion.


Natural Draft cooling towers are commonly found in Earthquake will govern the design of columns and foundation
engineering practice at many power plant sites where process in many cases.
water has to be cooled. Currently, the largest cooling tower The dynamic behavior and earthquake response of natural
world wide with height 200 m can be found at the draft cooling towers is mainly influenced by the supporting
Niederaussem site close to Cologne in Germany. Standard structure, especially the type of columns. Different column
heights of natural draft cooling towers vary from 160 m – 180 types are commonly used which can be seen in Figure 1.
m today depending on thermodynamic requirements. These are:
Although the shell can be regarded as a typical landmark, a • A-, V-, X- truss columns
standard natural draft cooling tower consists of the following • I-beams
structural parts: foundation, columns, shell, and upper While A-, V-, X-columns behave more or less as truss
stiffening ring. elements for transmission of horizontal forces from wind or
earthquake to the ground, I-beams are beam elements where
significant bending moments must be activated for
transmission of horizontal forces.
In this paper, the earthquake behavior of natural draft
cooling towers is explained with special emphasis on non-
linear column behavior from which behavior factors can be
derived which can be easily used in structural design to
account for non-linear dissipation effects. It can be already
seen from Figure 1: A-, V-, X- truss columns provide more
strength but less ductility than I-beams which will have a
significant effect on earthquake design.

2 BASIC MECHANICAL PRINCIPLES FOR


MODELING OF SHELLS OF REVOLUTION
2.1 Geometry and displacement interpolation
For numerical analysis of shells of revolution ring elements
Figure 1. Different supporting systems of NDCTs. can be used advantageously. In Figure 2, the geometry of a
shell ring element is depicted as implemented in finite element
code ROSHE [3]. The middle surface is described by a
A natural draft cooling tower has to resist different actions. relation between height Z and radius R(Z), see [2]. With
Among those is dead weight, wind, temperature, shrinkage. exclusion of non-symmetric terms, the displacement vector is
Earthquake must be regarded as an extreme design situation
Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 276

expressed using a Fourier series with order nmax according to ∫∫ p


α
δuα + p3 δu3 dA
Equation (1). A

+ ∫ nα δuα + n3 δu3 dC
C

− ∫∫ ~
n αβ δα αβ + mαβ δβαβ dA
A

− ∫∫ ρ h ⋅ [aαβ u
&&α δuβ + u
&&3 δu3 ] dA
A

− ∫∫ [c t aαβ u& α δuβ + c n u& 3 δu3 ] dA = 0


A (2)
2.3 Solution of equations of motion
In non-linear dynamics, the system of equations of motion is
expressed by Equation (2). However, this equation cannot be
solved directly due to the dependence on the current state but
has to be linearized. Incremental iterative solution strategies
have to be applied to the linearized system of equations as
given in Equation (3).
+ + +
Figure 2. Body forces and stress resultants. K t ⋅ V + C ⋅ V& + M ⋅ V
&& = P - P - P - P (3)
e i ce m
For non-linear structural behavior, direct integration
schemes work almost the same as for linear behavior. The
difference consists in the formulation in displacement,
velocity and acceleration increments. From a given state of
equilibrium of the last time step, the time is increased which
results in a change of effective load due to the change in the
external forces and due to the motion. The first solution of the
new variables of motion for a new time step is calculated in
the increment step (predictor step). After that, equilibrium is
checked in the updated state. However, it is most likely not to
be satisfied due to the linearization error. Iteration steps (or
corrector steps) are applied until the convergence criterion is
satisfied (out of balance forces must become zero). Actually,
in every increment or corrector step the problem is reduced to
the problem of a linear static analysis using an effective
stiffness which consists of a combination of stiffness, mass
and damping and an effective load vector.
Figure 3. Ring element concept.
3 DYNAMIC EARTHQUAKE ANALYSIS: APPLICA-
For the unknown Fourier coefficients of displacements nui, a TION STUDY NDCT ON V-COLUMNS
finite element approximation is introduced. Cubical
polynomials in meridian direction θ² are used for each The first application study will deal with a natural draft
component of the displacement vector. Hence, the proposed cooling tower supported on V-truss columns as depicted in
ring element possesses 12 DOFs per Fourier term n as Figure 4. The geometry of this tower is described by two
depicted in Figure 3. hyperbolic shape functions below and above the throat which
⎡ u1 ⎤ n ⎡ u1(θ ) ⋅ sin(nθ ) ⎤
n 2 1 follow the following relation:
⎢ ⎥
(1)
u(θ1, θ 2 ) = ⎢⎢u2 ⎥⎥ =
2
(4)
max

∑ ⎢ u (θ n
2
2
) ⋅ cos(nθ1 )⎥ ⎛ θ 2 − θ 02 ⎞
R(θ ) = R 0 + a ⋅ 1 + ⎜
2 ⎟
⎢⎣u3 ⎥⎦ n =0
⎢ nu3 (θ2 ) ⋅ cos(nθ1 )⎥ ⎜ b ⎟
⎣ ⎦ ⎝ ⎠
The following parameters apply below and above throat at
2.2 Principle of virtual work for dynamic problems θ0² = 115.83 m:
Below throat: R0 = -15.3644 a = 51.9644 b = 113.9896
The principle of virtual work implies equilibrium equations in Above throat: R0 = +36.3422 a = 0.25780 b = 8.0293
weak formulation and is given in Equation (2) for a shell. For
dynamic problems, additional inertia and damping forces have The cooling tower is subjected to dead weight and
to be considered as described by the last two expressions. horizontal ground motion (Kern County NS component Taft
Lincoln School USGS Station 1095, [8]) with maximum peak
ground acceleration a = 0.35 g. For this time history, non-
linear dynamic transient analysis has been carried out [5].
Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 277

Figure 5. Comparison of column tensile forces.

4 SIMPLIFIED STATIC EARTHQUAKE ANALYSIS:


APPLICATION STUDY NDCT ON I-BEAMS
Figure 4. Cooling tower model on V-columns. I-beams as supporting system of a cooling tower have been
preferably used during the last years. Their main advantage
It can be found that the following non-linear effects occur: can be found in the construction stage erection of lower ring
Uplift of foundation, cracking of concrete shell and especially beam and start of climbing. The number of columns depends
cracking of concrete columns. Results of column tensile on the precast elements at the shell bottom and climbing
forces are plotted in Figure 5. Due to uplift, a reduction of system used. To each column, a climbing unit is attached in
column tensile forces in regions of uplifting foundation areas the beginning of the climbing process. However, I-beam
occur at column pairs 1 – 3 and 20 – 23. However, maximum supporting systems have to transfer horizontal loads to the
column forces occur in regions around 90° (column pairs 9 – ground by bending action. The analysed system is depicted in
14) where maximum horizontal forces have to be transferred Figure 6 including the deformation pattern when subjected to
from the shell to the ground. From non-linear column force horizontal ground motion. The geometry of this tower is as
results, a behavior factor can be determined using the well described by Equation (4) using the following parameters
following design relation. below and above the throat θ0² = 107.00 m:
! below: R0 = -1.93500 a = 32.2350 b = 79.9114
S d,linear (g + γ I ⋅ e/q ) = S d,non -linear (g + γ I ⋅ e ) (5) above: R0 = +29.4081 a = 0.89190 b = 13.2920
wall thickness: θ² = 0 m 0.85 m
With consideration of Equation (5), a behavior factor θ² = 15 m 0.20 m
q ≈ 1.50 can be determined for cooling tower structures θ² = 39 m 0.20 m
supported on V-columns. Thus, V-, A- and X-column systems θ² = 44 m 0.18 m
mainly carry the forces from the shell to the foundation by θ² = 142 m 0.18 m
tension and compression. They are able to transfer high θ² = 145 m 0.30 m
horizontal forces to the ground, however, in conjunction with
little ductility. Thus, a low behavior factor of q = 1.50 is The following I–beam reinforcement ratios apply while a
recommended for such systems with low ductility. rigid foundation is considered for simplicity reasons:
Top section 0.90 x 0.85 As = 123 cm² (20∅28)
Bottom section 0.90 x 2.30 As = 123 cm² (20∅28)
Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 278

For further earthquake analysis, the following response


spectrum shall apply as depicted in Figure 8 according to DIN
4149 [6] soil property C-R is used.
T = 0.00 s – 0.05 s Sa = 3.0⋅(1+1.50⋅T/0.05) (6a)
T = 0.05 s – 0.30 s Sa = 2.5⋅3.0 = 7.65 m/s² (6b)
T = 0.30 s – 2.00 s Sa = 2.5⋅3.0⋅(0.30 / T) (6c)
T > 2.00 s Sa = 2.5⋅3.0⋅(0.30⋅2.00/T²) (6d)

Sa [m/s²]
10

9 Original Spectrum
8 Spectrum due to artificial ground motion
Spectrum due to artificial ground motion
7 (non-linear)
6

5
4
q
3

0
0.00 0.50 1.00 1.50 2.00
Figure 6. Cooling tower model on I-beams.
Period T [s]
Sv [cm/s]
The first natural eigenmode at 1.15 Hz describes a shell 100

eigenmode which is not relevant for earthquake response. The


first earthquake response relevant eigenmode can be found at
1.45 Hz which is also depicted in Figure 6. It can be found
that for systems on I-beams the relevant eigenmode is
described by an almost rigid body motion of the shell on the 10
very weak I-beam column supporting system which has to
carry the total horizontal force to the ground by activation of
bending moments. Almost the same acceleration applies for
the whole shell over total height so that an equivalent constant
static acceleration can be applied. 0.05 0.30 2.00
1
For this constant static acceleration acceleration load, a non- 0.01 0.10 1.00 10.00
linear static push-over analysis is carried out. Hereby, dead Period T [s]
weight is constant and horizontal force has been increased.
The corresponding load deflection curve can be found in Figure 8. Comparison of linear and non-linear spectrums.
Figure 7. The total dead weight of the tower (shell and
From this design spectrum, a compatible artificial
columns) is 104800 kN. Failure will occur at horizontal
acceleration time history is synthetically generated as depicted
acceleration a = 0.85 g with corresponding horizontal
in Figure 9. This time history is back transformed into
displacement 230 mm.
response spectrum diagram while quite good agreement can
be found between original spectrum and spectrum from
Horizontal Acceleration
artificial ground motion [9].
1.0 g 3

a [m/s²]
0.8 g
2

0.6 g
1
Load deflection curve
0.4 g Simplified bilinear curve
0
0 5 10 15 20
0.2 g
-1
Displacement at shell lintel [mm]
0.0 g
0 -50 -100 -150 -200 -250 -2

t [s]
Figure 7. Load deflection curve of NDCT. -3

Figure 9. Artificial spectrum compatible ground motion.


Proceedings of the 8th International Conference on Structural Dynamics, EURODYN 2011 279

With the following data, a non-linear spectrum can be factors apply for I-beam systems. However, more severe
derived based on the non-linear system properties of I-beam constructive requirements apply for I-beam systems to
supporting column system as depicted in Figure 7. exclude brittle shear behavior.
uel = 76 mm Methods and hints for determination of structural response
H = 0.6⋅104800 = 62880 kN and non-linear behavior in conjunction with selection of
k = 62880/0.076 = 827474 kN/m behavior factors have been presented. However, the presented
m = 104814/9,81 = 10684 t results show only trends. So, each system must be checked
f = 1/2π ⋅ (827474/10684)0.50 = 1.40 Hz and analysed individually.
μ = upl / uel
ACKNOWLEDGMENTS
kv = 0.25⋅104814 / (230-76) / 1000 = 171265 kN/m
kv ≈ 0.20⋅k L.A.W. Consulting Engineers have been involved into design,
consulting, structural design verification, strengthening and
The spectrum with consideration of non-linear column quality management of more than 30 cooling tower projects
behavior can be also found using non-linear spectrum worldwide. The co-operation with our project partners and
evaluation [9]. By comparison of linear response spectrum of clients is gratefully acknowledged.
equivalent single degree of freedom system with non-linear
REFERENCES
response spectrum, a behavior factor approximately q ≈ 3.0
can be derived, see Figure 8. [1] A. K. Chopra. Dynamics of structures. Prentice Hall, 2001.
[2] Y. Başar and W.B. Krätzig. Mechanik der Flächentragwerke. Vieweg &
Thus, for earthquake resistant design of this natural draft Sohn, 1985.
cooling tower structure supported by I-beams, the following [3] C. Lang, F. Altmeyer and J. Weigl. Finite Element Program ROSHE V
recommendations can be given: a) Higher behavior factors 2010, Internal L.A.W. Work Report (2010).
apply than for systems supported by V-, A- or X-truss [4] C. Lang, R. Meiswinkel and F.C. Filippou. Non-linear analysis of shells
of revolution with ring elements, Engineering Structures, 2002, Vol.24,
columns because systems supported by I-beams behave more No.2, 163-177.
flexible and show higher possibility of energy dissipation due [5] C. Lang. Beitrag zur Theorie, Numerik und Anwendung nichtlinearer
to bending action of columns. b) The behavior factor of such Algorithmen zur statischen und dynamischen Analyse von
systems should be limited to q = 3.0. c) Very severe Stahlbetonrotationsschalen. PhD thesis, 2003, TU Kaiserslautern.
[6] DIN 4149. Buildings in German Earthquake Areas. April 2005, Beuth
constructive requirements apply for such systems concerning Verlag Berlin
constructive longitudinal and transversal (shear) [7] EN 1998-1 Design of structures for earthquake resistance – Part 1:
reinforcement quantities, bond lengths, anchoring of General rules, seismic actions and rules for buildings; German version
reinforcement and constructive rules to ensure a ductile EN 1998-1:2004, Beuth Verlag Berlin
[8] Strongmotion database http://peer.berkeley.edu/
behavior. Brittle shear behavior of columns must be avoided [9] K. Meskouris and K.-G. Hintzen. Bauwerke und Erdbeben. Vieweg,
by fulfillment of these constructive requirements. By selection 2003.
of this high behavior factor and implicitly taking benefit of
linear elastic force reduction by this simplified design method,
the structure must be classified into class DCH (high ductility
class) including fulfillment of all constructive requirements
for this class DCH to exclude brittle column failure. This is
absolutely necessary since significant plastic energy
dissipation takes place since the structure is not designed for
the full linear elastic response (response spectrum using
q = 1.00).

5 SUMMARY
In structural design of natural draft cooling towers, it can be
chosen between different design solutions concerning the
support at the shell bottom. It has to be distinguished between
truss column solutions such as V- and A-columns or X-
columns especially for large air inlet heights to reduce
buckling length and on the other hand I-beam column
supporting systems. I-beam column systems have been
recently used in structural design due to some advantages
concerning construction stages and erection schedules.
However, while V-, A- and X-truss columns are able to
carry high horizontal forces with corresponding low
deformation, I-beam supporting systems behave much more
flexible.
Earthquake design is always an interaction between strength
and flexibility. I-beam supporting systems behave more
flexible than truss column systems. Thus, higher behavior

You might also like