Академический Документы
Профессиональный Документы
Культура Документы
long as the boundary flexibilities are known. Also, rotational member end
stiffnesses are readily available in any textbook on structural analysis, in cases
where rotational DOFs are considered in the analysis.
For squat bridge piers, where the clear column height is no longer
significantly larger then the column depth D, shear deformations can become
tion for a unit load, or the shear flexibility, in Fig. 4.7(c) can be expressed as
fv H e (4.21)
AG
where A ve represents the effective shear area (to be discussed later) and G the
shear modulus of the pier cross section. As a general rule of thumb, shear
deformations can become significant when the shear span MN of the pier is
less than three times the pier depth D, or
3D
V (4.22)
= 1 = H e
fb
kb aEIe (4.23)
1 1
k
3
fb H elaEIe + HelAveG
(4.24)
Since bridge columns are, as outlined in Chapter 1, expected to respond
under the design earthquake inelastically, effective member properties He, le,
and A ye , which reflect the extent of concrete cracking and reinforcement
yielding, should be used in the modeling and analytical member characteriza -
tion to obtain realistic seismic response quantification.
From Eq. (4.17) it can be seen that rather than gross-section-based moments of
inertia I g and shear areas A, effective properties I e and Ave are specified, and for
the height H in the stiffness formulation an effective height H e is used.
Constitutive parameters E and G are assumed to be constant for most types of
bridge analyses and can be determined for concrete piers from the nominal
concrete compression strength f: based on standard ACI or CEB procedures.
The modulus of elasticity E can be determined as outlined in
FUNDAMENTALS OF SEISMIC BRIDGE BEHAVIOR 171
Section 5.1, Eqs. (5.1) and (5.2), and the shear modulus G with a Poisson ratio
v for concrete between 0.15 and 0.2 for homogeneous material assumptions as
G E
2(1 + v) (4.25)
This yield penetration length 41 can be added to the clear column height
as outlined in Fig. 4.8 for longitudinal and transverse response, respectively,
to form the effective column height 1-1,. In addition, footing springs, as shown
in Fig. 4.8, modeling the effects of soil deformations (discussed in Section
4.4.2) should be employed to obtain the correct stiffness characteristics for
bridge bent models.
To reflect the cracked state of a concrete bridge column in the seismic
response analysis, an effective or cracked-section moment of inertia 4 should
be employed. The effective stiffness El, does not reflect only the effect of
cracking but also the state of the bridge column determined at first theoretical
yield of the reinforcement and can be determined from sectional moment
curvature analyses as
M
El, = (4.27)
cloyi
where Myi and (13 yi represent the ideal yield moment and curvature for a
a) Prototype b) Model
Mass Center
FIG. 4.8 Effective column height.
1
1_1
2
t
5.3.2. The effective stiffness I e depends on the axial load ratio Paxiail(A gfc) and
3
the longitudinal reinforcement ratio A st/Ag, where A, and A represent the
gross concrete area and the total longitudinal reinforcement areas, respec -
tively. These effective stiffnesses are represented in graphical form in Fig. 4.9
for typical circular and square column cross sections and show that for typical
column reinforcement ratios between 1 and 3% and axial load ratios between
10 and 30%, a reduction in effective section moment of inertia to between 35
ofnetration
Yield Pe T
and 60% the gross section moment of inertia HIg is not L uncommon.
He A similar
He
reduction in effective stiffness applies to other concrete
e bridge members, such
as cap beams and superstructure girders, and appropriate values for I e can
also be determined from Eq. (4.27) or Fig. 4.9 for corresponding axial load
levels and reinforcement ratios.
Finally, an effective shear stiffness GA rather than the shear stiffness
based on the shear area A, should
4 be employed to reflect the increased shear
deformations in flexurally cracked concrete
/ / / / /members.
/ // Again, a dependency
similar to the effective flexural stiffness EI, on the axial load and the
reinforcement ratio can be expected prior to significant shear distress, which
0.20 ______
FUNDAMENTALS
OF SEISMIC BRIDGE BEHAVIOR 173
11
4;
0.70
cS.
17-1, Ast/Ag = .04
0.60=
rn
C13
0.50
Ast/Ag .03
0.40 i
ra
C.) = .02
0.30-
Ast/Ag = .01
40,
0.80-
cp
0. 70
PA 0.50
A 0.60 AstiAg = .03
Ast/Ag = .02
Ast /A 5 = .
01
Fa 0.40
L)
g0.30
0.20.......................I 1 1 1111 l .
1 " 1 ____________________________________ll________I
(c) Damping. The third term in the general equation of motion is a viscous
or velocity proportional damping force fd(t) = cas(t), which is used primarily
174 MODELING AND ANALYSIS
for mathematical or numerical convenience and stability rather than for the
phenomenological modeling of actual bridge damping characteristics. Viscous
damping is physically correct only for an oil-filled dashpot and is difficult to
rationalize for other forms of damping actually encountered in bridge or other
structural systems. More common damping types in bridges are (1) Coulomb
damping, (2) radiation damping, and (3) hysteretic damping. Coulomb or
friction damping occurs primarily in bridge superstructure bearings and move-
ment joints and is independent of velocity or displacement. To a lesser degree,
friction damping can occur in cracks of reinforced concrete structures. Radia-
tion damping in bridges occurs due to soil structure interaction (SSI) and
energy dissipated by waves radiating out into the half-space of soil surrounding
the bridge footings. The most common and physically most obvious form of
damping or energy dissipation in bridge structures is in the form of hysteresis
of the forcedeformation response.
To conform to the simple mathematical form of the equation of motion,
other forms of damping, in particular, hysteretic damping, encountered in
bridge systems are conveniently expressed in the form of an equivalent viscous
damping coefficient ceq. The equivalent viscous damping coefficient ceq is
commonly expressed by the equivalent damping ratio t eq and the critical
damping coefficient c,, which is the smallest amount of damping for which
no oscillation occurs in free dynamic response [CO]:
C e q = teqCcr (4.29)
The hysteretic damping or energy loss per cycle, represented by the area
Ah in Fig. 4.10 for one complete idealized loaddisplacement hysteresis loop,
can then be converted for the same displacement amplitude to an equivalent
viscous damping ratio:
A h A h
eq = (4.30)
27TV,A, 477 -A,
where V, and Am represent the average peak force and displacement values
[C10]. The area A, represents the elastic strain energy stored in an equivalent
linear elastic system under static conditions with effective stiffness
V
keff = --'n (4.31)
Am
The equivalent viscous damping coefficient can then be obtained from Eq.
(4.29). To assess the magnitude of equivalent viscous damping from hysteretic
damping, typical forcedisplacement hysteresis loop shapes for various bridge
members are depicted in Fig. 4.11.
FUNDAMENTALS OF SEISMIC BRIDGE BEHAVIOR 175
FORCE
k eff
vnia.
Oliarj
Vm=1/2( Ivi,1+
1
A = Elastic Strain
am=1/2( Irma m i n e
A llEMN M E n e rg y
' a l min
A A 1114:112441. DISPLACEMENT
6
` max
= E n e rg y D i s s i p a t i o n
Per Cycle (Shaded Area)
min
(d) Column Hinge with (e) Rocking of Pier (f) Knee Joint
High Axial Load
FIG. 4.11 Typical hysteretic response in bridge components.
in Fig. 4.10, can still be employed. The same difficulty exists with any dynamic
analysis where equivalent viscous damping is employed since the damping
values are derived for the maximum amplitudes and are subsequently applied
in the analytical model also to all smaller-amplitude cycles.
From the discussions above it is obvious that determination of a correct
equivalent viscous damping coefficient for analytical bridge models is difficult
at best. Thus empirical damping values are frequently employed to reflect the
sum of all possible damping contributions, as well as the fact that most of the
cyclic dynamic response in an earthquake is expected to occur at smaller -
than-expected maximum displacement levels.
Thus, for steel structures, damping values between 2 and 5% of critical
damping are commonly assumed, while for concrete structures a range from
2 to 7% is used to reflect the most representative dynamic response range.
In light of these uncertainties, the commonly assumed 5% viscous damping
coefficient in structural dynamic analysis can hardly be argued with. Only in
FUNDAMENTALS OF SEISMIC BRIDGE BEHAVIOR 177
cases where (1) soilstructure interaction plays an important role, (2) special
energy absorption devices are employed, and (3) high hysteretic energy dissi -
pation is relied upon should higher damping coefficients be employed. It
should also be noted that most analytical models will be based on initial elastic
stiffness, as discussed in the preceding section, and the damping adopted
should represent the elastic phase of response. Where inelastic time-history
analysis is used, the hysteretic energy dissipation will be directly modelled by
the force-displacement hysteresis rules adopted in the analysis. Only with the
substitute structure analysis procedure, discussed further in Section 4.5.2(b),
where the effective stiffness represents that at maximum displacement, rather
than at yield, should the effective damping be increased to include the effects
of hysteretic damping.
Assuming that the displacement u(t) with time follows a harmonic motion,
as shown in Fig. 4.12, of the form
where o.) is the circular natural frequency, a a phase shift for the sinusoidal
response, and A a scaling factor that determines the amplitude of the harmonic
motion, Eq. (4.33) and its second time derivative can be substituted into Eq.
(4.32), resulting in the characteristic equation
lk w2m1 = 0 (4.35)
which occurs only for a specific circular frequency w or the eigenvalue of Eq.
(4.35). The eigenvalue solution of this scalar equation represents the circular
frequency w at which Eq. (4.35) is satisfied as
CO = (4.36)
and from the undamped natural circular frequency co in Eq. (4.36) the cyclic
natural frequency f and natural period of vibration T for a SDOF bridge
model can be found as
(4.37)
u(t)
(4.38)
T = 27r (4.39)
wd = (4.40)
From Eqs. (4.40) and (4.41) the small influence of the damping ratio on the
dynamic response characteristics is evident. For example, an equivalent viscous
damping ratio of 10% only increases the natural period of vibration by 0.5%,
and even a 50% damping ratio only results in a 15% natural period elongation.
Since typically maximum expected equivalent viscous damping ratios in bridges
are less than 15%, the influence of damping on the dynamic response
characteristics can be neglected.
As discussed above, m and k in Eq. (4.39) represent the effective seismic
mass WsIg and the effective stiffness keff with reference to the single displacement
degree of freedom at the mass centroid. This single-degree-of-freedom concept
to estimate the fundamental dynamic bridge response characteristics can also
be applied to structures with distributed parameters in the form of a
generalized single-degree-of-freedom system as outlined by Eq. (4.4).
Expressions for a generalized single-degree-of-freedom response of a dis-
tributed parameter systems can be found in [C10,N1], and only the special
case of lumped stiffness and mass characteristics along the length x of a bridge
structure are discussed below since discrete springs (modeling the stiffness of
individual bents) and discrete masses (modeling lumped superstructure inertia)
can readily be identified for typical bridge systems. In the case of discrete
translational and rotational masses m1 and j1, the principle of virtual work can
be applied to a generalized single-degree-of-freedom system in the form of
Eq. (4.4) such that a new generalized mass m* can be derived as
180 MODELING AND ANALYSIS
k* = k Ag a) (4.43)
Example 4.1. For the example of the multicolumn bent in Fig. 4.4 with an
in-plane rigid bridge deck, the generalized mass and stiffness parameters can
be found as follows, based on the idealized boundary conditions.
as
m* = K2
= 0 = TriL (4.45)
and
k* k, 1 = 3ke
with the effective stiffness k e for each bent derived from Eq. (4.18). The (4.46)
fundamental period of vibration can then be found from Eq. (4.39) as
stiffness at the abutment is added to the system and the generalized displaced
shape can be expressed as
(x) = 1 x (4.48)
where x* = L/2 represents the distance from the center of stiffness to the
generalized coordinate or displacement DOF u*(t). Lumped-mass properties m
and j can be found at x = L/2 from Eqs. (4.7) and (4.8), respectively. The
generalized mass can now be found as
*_ -
L (L/ 2 ) 2 WIL 3 ( 1 2
M _m
"IL4-
X* 12 x* = 12 (LL/23)2 = nlL (4.49)
(k co2m)u = 0 (4.53)
182 MODELING AND ANALYSIS
1k (02m1 = 0 (4.54)
and n roots or solutions o can be found that satisfy Eq. (4.54), representing
the n natural circular frequencies of motion of the bridge model.
Once the vector a) with characteristic or modal frequencies is obtained, the
corresponding mode shapes or natural modes of vibration can be obtained
by substituting the individual modal frequencies into Eq. (4.53), and arbitrarily
prescribing the magnitude of one of the DOFs for the harmonic motion, since
with time, any magnitude of displacements between zero and the maximum
values is possible.
24E/ 24E1
H3 H3 = 6E1 [ 4 4
k= (4.55)
H3 4 5
24E/
H3 H3
30E/
DOF
'ioEcoEr' 1 1.0 1.0
H Eleff
1
1
rigid
H EIS Eleff
B B MODE 1 MODE 2
The mass matrix can be expressed with the cap beam length B and the
assumed distributed constant mass Tri = m/B as
[mB 0 = m B [1 0 _ 0
(4.56)
m
1
= =m
0 mB 01 01
wl
and the mode shapes 4:1) i can be determined from Eq. (4.53) assuming that u1
= 1 as
ct, = 1.0 ci)2 = { 1.0
and
0.883 1.133
These mode shapes are depicted schematically in Fig. 4.13(c) and show
that in mode 1 the two lumped masses move in phase, whereas the sign change
in mode 2 constitutes out-of-phase motion of the two mass points.
vector r, the participation of each mode can be obtained as the modal participa-
tion coefficient
4imr
Pi= T (4.59)