Professional Documents
Culture Documents
A BSTRA CT
1 INTRODUCTION
I
m
~-//~-////.-///H/.-~-//~ / /
×
x is the relative displacement of the mass m, ~;gis the ground acceleration and
Q denotes the restoring force. Using an elastic-ideal plastic material law in
Elasto-plastic oscillators subjected to Kanai-Tajimi spectra 111
the classical plastic hinge model (see Fig. 1), the total deflection x of the mass
m is d e c o m p o s e d into the linear part ~ and the plastic part x N, called the drift:
x = ~ + xN (2)
Inserting eqns (2) and (3) into (1) and putting the nonlinear part to the forcing
side, renders the equation of m o t i o n for the linear oscillator:
~'+ 2(O)o ~ + eoZo{= _(~,, + j/N) (4)
where a small viscous d a m p i n g ( and the linear eigenfrequency a) g = c/m are
introduced.
Hence, the linear oscillator is driven by the ground m o t i o n w, and the
nonlinear drift process x N. The stiffness and the d a m p i n g of the system do
not change due to the nonlinearity. Between the intervals of plastic actions,
the drift x N remains constant. The linear solution part { of the elasto-plastic
oscillator does not cross the elastic barrier a; { = a = const during yielding.
Thus, the kinetic energy (m/2)~a) at the beginning of yielding is balanced by
the dissipated work due to yielding:
e{ra} = --E{Wd,p,}
m "2
E{ Ta} = ~- E{~(a)} (5)
E{ Wd,pl } = -- F6
where the expectation operation E{ } is used because the earthquake
loading is treated as a stochastic process. ~(a) denotes the velocity of the
linear system at the barrier a, F = mco2a,is the yielding force and the mean
yield increment 6 is the average a m o u n t of plastic deformation during a
single isolated barrier crossing of the total deformation. Using
rnt°2 2
E{ra} = ~ - - a ~ (6)
S~g(O9o) is the value of the one-sided power spectral density function of the
ground acceleration at the linear eigenfrequency oJo.
It is customary to characterize the build up and decay of the ground
m o t i o n by means of a time dependent envelope function ~(t). Thus
S~(o~, t) = ~9(t)Sg,(cn) (9)
is the evolutionary power spectral density of the ground acceleration.
In eqn (8), ~O(t)= H(t); H(t) is the Heaviside step function and the linear
response is assumed to be quasi-stationary.
Equation (7) has been improved by taking into account the work of the
ground m o t i o n E{ W~} during the yield interval 4- 6--the energy balance at
the barrier becomes:
- E { L} = E{ wd..,} + E{ (10)
To approximate E{ W~} the results given by Grossmayer 9 are used:
E{ W~} = E{ Aty}mrcS,,(O9o)/2 (11)
The yielding time interval is approximately: 8
E{Aty} -~ a¢/a~o o (12)
Thus
a~
6 = ~aa (1 + ~S~(t~o)/acaco 3) (13)
where
do9
q = (1 - 22/),o22) U2 ~ 2(¢/n) U2
Note, that p~- ~ gives the mean time between clumps of plastic actions, This is
an important response measure with respect to the danger of material
degradation and collapse.
Furthermore, the variance of the accumulated drift can be approximated
by means ofp~. The total accumulated drift is the sum of the d~ o f e q n (14):
Nc
where Arc is the number o f clumps in a time interval 0 < z < t, and it is
assumed, that Arc has a Poisson distribution proportional to lq, and the time
between clumps is expected to have an exponential distribution with mean
1/p.. Hence,
E{Nc} ~-- Ia~t (21)
and
a~N(t) = E{Nc}a~, = 2!a,62t (22)
which is backed by simulations.
F r o m the above formulation it is clearly seen, that 6 and #a are stationary,
while the variance of the accumulated drift is proportional to the time t. This
is a consequence of the stationarity o f the linear solution part. Hence, the
total problem is not stationary in the stochastic sense (it is not even weakly
stationary) and therefore it will be called quasi-stationary. The extension to
a nonstationary formulation for a¢(t), 6(0 and pa(t) is given in Refs 4-6.
114 R. Hasenzagl, H. lrschik, 1~: Ziegh, r
Using the above results for the drift process, probabilistic inelastic
response spectra are calculated: The approximation for the probability ~that
the drift does not exceed a given level b during the duration s,
P = exp ( - [ e x p ( E { N c ( s ) } ) - 1] exp(-b/6)> (23)
is used to calculate the level b for a given probability, e.g. P = 0 . 5 .
Accordingly, a ductility ratio is defined:
tl(P, s ) = 1 + b/a (24)
and is interpreted as an inelastic response spectrum.
With the exception of minor improvements, formulas (5) to (24) follow
Vanmarcke et al., 1 - 3 where the fully associated linear system is used for the
calculation of the inelastic response measures, i.e. the original earthquake
loading O,(t)S,,(~o) defines the loading of the linear system.
It has been shown in eqn (4), however, that the ~-process is driven by the
earthquake excitation as well as by the drift process and that an effective
earthquake loading should be used. To account for the effective loading
including the drift, an effective envelope function 0e(t, ~) with frequency
dependence is introduced. Hence,
S~g(o~, t) = ~9~(t)~e(t, ~o)Sg~(e~) (25)
is the effective evolutionary input power spectral density. Again restricting
to the quasi-stationary case, ff~ becomes time independent and ~b~(og)S~g(e~)
has to be used in eqns (8) and (16) instead of Sg~(~). In order to calculate ~%,
an energy consideration is introduced: It is required, that the change in the
1000
l~t,3o ///
500
{=0,05 9'
200
100
50
/ 3 2
H= ~ o Fig. 2. Average eigentime between clumps
Sgg of plastic-action #£1tOo; • simulation re-
15 ,,,, .... I .... ~ .... I .... , .... I ........ sults and - - . - - theory of Vanmarcke et
al.1-3 for ~ = 0-02, - - present theory for
0 100 200 300 /+00 = 0-02 and 0"05.
Elasto-plastic oscillators subjected to Kanai-Tajimi spectra 115
average input power of the elastic system (without and including the drift) is
balanced by the mean power dissipated due to yielding:
E{L,}(1 - ~Oe(o~))=/~aF0 (26)
E{ L,} = ½rcmS,,(Ogo)
/~a and 6 being nonlinear functions of ~,o(~oo)= ~'e via tr¢ of eqn (8), eqn (26)
becomes a nonlinear equation for ~O, which can be solved on a pocket
computer using standard iteration procedures. Thus, contrary to Van-
marcke et al., nonlinearity explicitly enters the mathematical formulations,
which is appropriate to elasto-plastic problems.
An an example for this improvement, Fig. 2 shows the average time
between plastic clumps compared to theoretical and simulated results of
Vanmarcke et al. ~-3 It is seen, that the present theory fits perfectly well to
the simulation results.
3 D I M E N S I O N A L ANALYSIS A N D N U M E R I C A L RESULTS
v,
1.0-
.9 / 11
8. ff ,-o.o, o;..,
0.5
0.1
0.05 0.1 0.5
Fig. 4. Dimensionless mean single yield increment 6/a as a function of 1 / x / ~ ; © •
simulation results and - - - theory of Grossmayer 9 for ( = 0'02, present theory for
( = 0"02 and 0"05.
.20 ,\ \\
\ \
.15 g: 0.05
.10 =Q02
.05
, ,
0 '
10 20 50 100 200 500
Fig. 5. Nondimensional r.m.s, value axN/ax/~toto f accumulated drift for a viscous damping
= 0.02 and 0"05.
118 R. Hasenzagl, H. Irschik, F. Ziegler
the method. In order to study the influence of the structural and loading
parameters on the elasto-plastic response in more detail, however, the
ground excitation is characterized by the well established Kanai-Tajimi
representation
[I + 4C~z(~o/cog)2]So
S,,(o~) = [ 1 - (co/cog)2] 2 _~_ 4(2(co/0~,)2 (33)
co~ and ~g denote characteristic frequency and damping values of the ground
motion, respectively, So is a measure of the intensity of the earthquake. In
the following we use o~g = 2 0 s - t and ~ = 0.8 which has been reported by
Grossmayer et al. 13 for the 1976 Friuli earthquake. Choosing a r.m.s value of
the g r o u n d acceleration of a,;: = 0.127g, gives an intensity factor
So = 260"10 -4 m2/sec 3, see Ref. 12, g is the acceleration due to gravity. The
power spectral density function S,g(o~) with the above values of coS, ~, and So
is shown in Fig. 6. Also shown is a power spectral density function
corresponding to the earthquake code of Portugal 14 for rock. Note the
remarkable coincidence in the frequency content. To take into account a
variation of the load intensity, S~(co) of Fig. 6 is scaled by a loading factor 2
and is introduced into the similarity number
H = a2co~/2S~(O~o) (34)
The similarity number H thus is splitted into the linear eigenfrequency
fo = ~Oo/2~ and into the reduced yielding barrier a/x/-2~: Elasto-plastic
oscillators with the same value of fo and a/x/~ result in the same
I,-21"31
500
t,O0
300 / \\.
200 i \\.,
\
100
o/~-'tom]
14
• ta/¢'~- [ cml
12 10
I0
fotHZ] 2
0.8 I 1,5 2 2 3 5 f0iHzJ
Fig. 7. Dimensionless mean single yield increment 6/a as a function of a / ~ . and of the
linear eigenfrequency fo = ~Oo/(2~) for the power spectral density of Fig. 6, ~ = 0'02.
function ofa/x/~. Equations (23) and (24) provide the ductility ratio of Fig. 9
for a given probability P - - 0 . 5 and an excitation duration s = 4 5 T 0 for
( = 0 . 0 2 , T o = 1/f0. Thus ductility is connected with probability and
earthquake (strong-) motion duration, and Fig. 9 shows the undesired effect
of plasticity, that is the occurrence of permanent deformation, crucial for the
reliability of the structure. It is emphasized that the desired effect of ductility,
namely the increasing energy dissipation, should not be considered as the
only point of view.
In the last step, the influence of the characteristic ground parameters co,
and ~, are studied. For this sake, new similarity numbers are derived. ~5
Considering the Kanai-Tajimi eqn (33) in eqn (27), the latter contains four
dimensional inputs: a, coo, cog and So. According to Buckingham's theorem,
120 R. Hasenzagl, H. Irschik, F. Ziegler
• ']0,~
,1
~[Hz}
.054-,,,,,, , I .... I'"L'~lC[Hz],o
0,g 1.5 3.5 3.5 4 5 6 7,5
Fig. 8. Dimensionless r.m.s o f a c c u m u l a t e d drift axN/ax/~ot as a function o f the reduced
yielding barrier a/x~ , Sgg((~) o f Fig. 6, ( = 0.02.
alET Ic~]
9
1/-+
12
10
8 \
2
6
t,.
1
2
**lh,,,! , i , i i I I i , 6 iiii1 ii tlllll]lllll|ll , ] II I I
these inputs are replaced by two similarity numbers, namely the frequency
ratio COo/COS and the new similarity complex.
R = coo a2/ (35)
Full physical similarity is given by the four similarity n u m b e r s ~, ~8, c%/Wo, R.
The results are calculated for various values of the ratio c%/coo and for two
values of the g r o u n d d a m p i n g factor (~. The d a m p i n g factor of the structure
is fixed at ( = 0.02 at first. Figure 10(a)shows the dimensionless m e a n single
~a .~-0.8
'i
.5
,2 ~~ ~-~°'1'3 2
.1
.05
.02
0 2OO 4O0 6OO 8OO 1000
(a)
8/a ~;g-0,24
1 L '\\
.5 ~\\\\\ ~. 41
~"
1.5
,2 2 5
2O
.1
.05
.02 " R
0 200 400 600 800 1000
(b)
Fig. 10. Dimensionless mean single yield increment t$/aas a function of R for various values
~%/O9o;~ = 0-02; (a) ~s = 0"8; (b) ~g= 0.24.
122 R. Hasenzagl, H. lrschik, F. Ziegler
q-0,o
.040
.030
/~I0,20
.020
.010
•~ -0.24
,OZ,O
~c
.030 I/ 1.5
,020
\"k ~
,010
(b)
Fig. 1 I. N o n d i m e n s i o n a l m e a n rate o f plastic clumps/~a/o~o as a f u n c t i o n o f R for v a r i o u s
ratios ~o,/~ao; ¢ = 0"02; (a) ¢~ = 0'8; (b) ¢, = 0"24.
Elasto-plastic oscillators subjected to Kanai-Tajimi spectra 123
~g-O.8
.3
.2 ,,),
\Vf--~Jo"1'3
~\ ~J
\\\\
\\"f,X z
0
~I I I ' n '1 n
~0, 20
I n I I : I
R
20 50 100 200 500 1000
(a)
.~g-0,24
---wg--I1
,3 1.5
'\\ \\ \\ /
.1 x x ~
R
20 .50 100 200 500 1000
(b)
Fig. 12. Nondimensional r.m.s value axN/ax~ot of accumulated drift as a function of R for
various values t~J~o; ( = 0'02; (a) (g = 0"8; (b) (, = 0.24.
.I ~ 1.3
0~I ~ ~
. F ~ ~ , ~ - J _5 20 R- a~3o/So
2 3
ol ....
0 50 I(30 150 200 250 300 350
.12
=1.3
.10
,06 ~3
.0~ 0
,o2 R-a2(~/So
i J | I i = I a I I
10 20 50 100 200
Fig. 13. Drift statisticsas a function of R and various values ~ojOgo; C = 0"05, C= = 0.8;(a)
Nondimensional mean single yield increment 6/a; (b) Dimensionless inverse of eigentime
between plastic clumps lq/(Oo; (c) Nondimensional r.m.s of accumulated drift axN/a~x/~o t.
Elasto-plastic oscillators subjected to Kanai-Tajimi spectra 125
improvements have been used with respect to the theory given in Section 2.
The approximation in eqn (8) has been checked by integration:
2
= 2 d~o (36)
instead of the approximation used on the left hand side ofeqn (26). In case of
a Kanai-Tajimi spectrum the filter function is:
4 CONCLUSION
ACKNOWLEDGEMENT
REFERENCES