Escolar Documentos
Profissional Documentos
Cultura Documentos
Shirley J. Dyke
Professor
Department of Civil & Mechanical Engineering
Purdue University,
West Lafayette, IN 47907
sdyke@purdue.edu
ph: (314) 935-5695
fax: (314) 935-4338
Lawrence Bergman
Professor
Department of Aero/Astro Engineering
University of Illinois
Urbana, IL 61801
lbergman@uiuc.edu
ph: (217) 333-4970
fax: (217) 244-0720
Gursoy Turan
Assistant Professor
Department of Civil Engineering
Izmir Institute of Technology
Izmir, Turkey
gturan@likya.iyte.edu.tr
ph: +90 (232) 498 6278
Steven Hague
HNTB Corporation
715 Kirk Drive
Kansas City, Missouri, 64105
shague@hntb.com
Keywords:
Benchmark problems, structural control, cable-stayed bridges, protective systems, earthquake engineering, smart systems
Abstract
This paper presents the problem definition for the second generation of benchmark structural control problems for cable-stayed bridges. The benchmark problem is based on the Bill
Emerson Memorial Bridge that is currently under construction in Cape Girardeau, Missouri,
USA. Construction of the bridge is expected to be completed in 2003. The goal of this study is to
provide a testbed for the development of strategies for the control of cable stayed-bridges. Based
on detailed drawings of the Emerson bridge, a three-dimensional evaluation model has been
developed to represent the complex behavior of the full scale benchmark bridge. The linear evaluation model is developed using the equations of motion generated around the deformed equilibrium position. Phase II considers more complex structural behavior than Phase I, including multisupport and transverse excitations. Evaluation criteria are presented for the design problem that
are consistent with the goals of seismic response control of a cable-stayed bridge. Control constraints are also provided to ensure that the benchmark results are representative of a control
implementation on the physical structure. Each participant in this benchmark bridge control study
is given the task of defining (including devices, sensors and algorithms), evaluating and reporting
on their proposed control strategies. These strategies may be either passive, active, semi-active or
a combination thereof. Participants should also evaluate the robust stability and performance of
their resulting designs through simulation with a evaluation model which includes additional
mass due to snow loads. A simulation program has been provided to facilitate direct comparison
of the capabilities and efficiency of the various control strategies. The problem has been made
available for downloading through the internet in the form of a set of MATLAB equations. A
sample control design is included to guide participants through the benchmark problem.
Introduction
The control of long-span bridges represents a challenging and unique problem, with many
complexities in modeling, control design and implementation. Cable-stayed bridges exhibit complex behavior in which the vertical, translational and torsional motions are often strongly coupled.
Clearly, the control of very flexible bridge structures has not been studied to the same extent as
buildings have. As a result, little expertise has been accumulated. Thus, the control of seismically
excited cable-stayed bridges presents a challenging problem to the structural control community.
The first generation of benchmark problems on cable-stayed bridges focused on the Bill
Emerson Memorial Bridge under construction in Cape Girardeau, Missouri, USA (Dyke et al.,
1.
2.
3.
4.
5.
6.
Doct. Cand. & Grad. Res. Asst., Dept. of Civil Engineering, Washington University, St. Louis, MO 63130.
Associate Professor, Dept. of Civil Engineering, Washington University, St. Louis, MO 63130 (Corresp. Author).
Senior Researcher, Korea Institute of Machinery & Materials, P.O. Box 101, Yusung, Taejon, 305-600 Korea.
Professor, Dept. of Aeronautical Engineering, University of Illinois, Urbana, IL 61801.
Assistant Professor, Dept. of Civil Engineering, Izmir Institute of Technology, Izmir, Turkey.
Associate Vice President, HNTB Corporation, 715 Kirk Drive, Kansas City, Missouri, 64105.
November 5, 2002
Caicedo, et al.
2000, 2002; also see: <http://wusceel.cive.wustl.edu/>). Based on detailed drawings of the Emerson bridge, a three-dimensional evaluation model was developed to represent the complex behavior of the full scale benchmark bridge. A linear evaluation model, using the equations of motion
generated around the deformed equilibrium position, was deemed appropriate. Because the structure is attached to bedrock, the effects of soilstructure interaction were neglected. To simplify the
problem for phase I, two assumptions were made regarding the excitation. This problem focused
on a one dimensional ground acceleration applied in the longitudinal direction and uniformly and
simultaneously applied at all supports. Researchers reported their Phase I results during a theme
session devoted to this problem held at the Third World Conference on Structural Control in April
2002 in Como, Italy (Agrawal, et al., 2002; Turan et al., 2002; Moon et al., 2002a; Bakule, et al.,
2002).
Although a significant amount of expertise was accumulated during phase I, the assumptions made regarding the excitation (longitudinal and simultaneously acting) limited the extent to
which this problem modeled a realistic situation. A structures response to an earthquake is based
on the simultaneous action of three translational components of ground motion: two in the horizontal plane, and one in the vertical direction. Structures are typically analyzed for the two horizontal components of ground motion. The structural response depends on the incidence angle (the
angle between the ground motion components and the structural axes). Additionally, the excitation is expected to vary at each of the supports due to the length of these structures. A phase II
problem was developed to extend the problem to consider these issues.
This paper presents the second generation of benchmark control problems for cable-stayed
bridges. In this problem the ground acceleration may be applied in any arbitrary direction using
the two horizontal components of the historical earthquake with a specified incidence angle.
Multi-support excitation is also considered in this phase of the study. Here the prescribed ground
motion is assumed to be identical at each support, although it is not applied simultaneously. We
assume that bent 1 undergoes a specified ground motion, and the motion at the other three supports is identical to this motion but delayed based on the distance between adjacent supports and
the speed of the L-wave of a typical earthquake (3 km/sec). The total response of the structure is
obtained by superposition of the response due to each independent support input (Chopra, 2001;
Clough and Penzien, 1993).
This problem has been prepared to provide a testbed for the development of effective strategies for the control of long-span bridges. This second generation problem considers the control of
more complex motions of the bridge as compared to the first generation problem. To evaluate the
proposed control strategies in terms that are meaningful for cable-stayed bridges, appropriate
evaluation criteria and control design constraints are specified within the problem statement.
Additionally, an alternate model of the bridge is developed for evaluating the robustness of the
designs. This model includes the effects of snow loads on the bridge deck. Designers/researchers
participating in this benchmark study will define all devices, sensors and control algorithms used,
evaluate them in the context of their proposed control strategies, and report the results. These
strategies may be passive, active, semi-active or a combination thereof. The phase II problem will
be made available for downloading on the benchmark web site in the form of a set of MATLAB
equations <http://wusceel.cive.wustl.edu/quake/>. A sample control design is included.
November 5, 2002
Caicedo, et al.
350.6m
(1150)
142.7m
(468)
1
32
570.0m
(1870)
142.7m
(468)
33
64
Illinois approach
Bent 1
1
Pier 2
Pier 3
Pier 4
Cable Number
xg
November 5, 2002
Caicedo, et al.
f c' of 41.36 MPa (6000 psi). Additionally, a concrete barrier is located in the center of the bridge,
and a railing is located along the edges of the deck.
The 128 cables are made of highstrength, lowrelaxation steel (ASTM A882 grade 270).
The smallest cable area is 28.5 cm2 (4.41 in2) and the largest cable area is 76.3 cm2 (11.83 in2).
The cables are covered with a polyethylene piping to resist corrosion. The H-shaped towers have
a height of 102.4 m (336 ft) at pier 2 and 108.5 m (356 ft) at pier 3. Each tower supports a total 64
cables. The towers are constructed of reinforced concrete with a resistance, f c' , of 37.92 MPa (5.5
ksi). The cross section of each tower varies five times over the height of the tower, as shown in
Fig. 3. Section A is used in the top of the legs, section B in the middle of the legs, and section E in
the bottom of the towers. Some of these elements have variable sections. Section D shows the
cross section in the bottom strut, and section C shows the cross section of the strut located in the
29.3m
(96)
Railing
Cable
Anchorage
Barrier
Concrete Slab
Floor Beam
Steel girder
0.61 m
Section A
6.71 m
4.88 m
6.71 m
3.96 m
Section C
1.68 m
2.74 m
1.68 m
to 2.59 m
Section A
Var. 2.74 m
to 3.66 m
5.18 m
Section C
Section B
Section B
Section D
Section E
3.66 m
30.8 m
3.66 m
6.71 m
6.71 m
3.81 m
3.96 m
Deck
3.96 m
3.66 m
Section D
3.66 m
Section E
November 5, 2002
Caicedo, et al.
middle of the tower. The approach bridge from the Illinois side is supported by 11 piers and bent
15 which are made of concrete. The deck consists of a rigid diaphragm made of steel with a slab
of concrete at the top. The density of the materials as specified in the drawings are summarized in
Table 1.
Table 1: Density of Materials.
Density
(Kg/m3)
2402.77
2482.86
2306.66
2322.68
7849.08
7849.08
Material
Reinforced concrete
Prestressed concrete
Seal Concrete
Stay cable grout
Structural Steel
Cables (Steel)
Density
(PCF)
150
155
144
145
490
490
Evaluation Model
Based on the description of the Emerson bridge provided in the previous section, a threedimensional finite element model of the bridge was developed in MATLAB (1997). A linear
evaluation model is used in this benchmark study. However, the stiffness matrices used in this linear model are those of the structure determined through a nonlinear static analysis corresponding
to the deformed state of the bridge with dead loads (Wilson and Gravelle, 1991). Additionally, the
bridge is assumed to be attached to bedrock, and the effects of soilstructure interaction are
neglected.
The finite element model employs beam elements, cable elements and rigid links. The nonlinear static analysis is performed in ABAQUS (1998), and the element mass and stiffness
matrices are output to MATLAB for assembly. Subsequently, the constraints are applied, and a
reduction is performed to reduce the size of the model to something more manageable. These
steps are described in the following sections and summarized in Fig. 4. The first ten undamped
frequencies of the evaluation model are 0.2899, 0.3699, 0.4683, 0.5158, 0.5812, 0.6490, 0.6687,
0.6970, 0.7102, and 0.7203 Hz.
To make it possible for designers/researchers to place devices acting longitudinally between
the deck and the tower, a modified evaluation model is formed in which the connections between
the tower and the deck are disconnected. If a designers/researcher specifies devices at these
nodes, the second model will be formed as the evaluation model, and the control devices should
connect the deck to the tower. As one would expect, the frequencies of this model are much lower
than those of the nominal bridge model. The first ten undamped frequencies of this second model
are 0.1619, 0.2667, 0.3725, 0.4547, 0.5017, 0.5653, 0.6190, 0.6489, 0.6968, and 0.7097 Hz. Note
that the uncontrolled structure used as a basis of comparison for the controlled system, corresponds to the former model in which the deck-tower connections are fixed (the dynamically stiff
shock transmission devices are present).
Description of Finite Element Model
The finite element model, shown in Fig. 5, has a total of 579 nodes, 420 rigid links, 162
beam elements, 134 nodal masses and 128 cable elements. The towers are modeled using 50
nodes, 43 beam elements and 74 rigid links. Constraints are applied to restrict the deck from mov-
November 5, 2002
Caicedo, et al.
ing in the lateral direction at piers 2, 3 and 4. Boundary conditions restrict the motion at pier 1 to
allow only longitudinal displacement (X) and rotations about the Y and Z axes. Because the attachment points of the cables to the deck are above the neutral axis of the deck, and the attachment
points of the cables to the tower are outside the neutral axis of the tower, rigid links are used to
connect the cables to the tower and to the deck (see Fig. 6). The use of the rigid links ensures that
the length and inclination angle of the cables in the model agree with the drawings. Additionally,
the moment induced in the towers by the movement of the cables is taken into consideration in
this approach. In the case of variable sections, the average of the section is used for the finite element. The cables are modeled with truss elements. In the finite element model the nominal tension is assigned to each cable.
The FEM model described above is used directly in cases when the control devices are
employed in the longitudinal direction between the deck and tower. If the designer/researcher
employs no control device at these locations (in which case the shock transmission devices are
included), the model is modified by including four longitudinally-directed, axially-stiff beam elements that force the deck to move with the tower in the longitudinal direction. The uncontrolled
structure used as a basis of comparison corresponds to this second case. Note that the program
included with the benchmark files determines if the designer/researcher has placed devices in this
location and builds the appropriate FEM model.
Note that the Illinois approach is not included in this model (see Fig. 1). Because the bearing at pier 4 does not restrict longitudinal motion and rotation about the X axis of the bridge, the
Illinois approach was found to have a negligible effect on the dynamics of the cable-stayed portion of the bridge.
Nonlinear Static Analysis
Cable-stayed bridges exhibit nonlinear behavior due to variations of the catenary shape of
the inclined cables, cable tensions that induce compression forces in the deck and towers, and
MATLAB
MATLAB
ABAQUS
Modify FEM
Assemble
K Kg M , Mg
Static
condensation
Output
Element
el
el
,
,M
,
K
Compute Damping
Matrices
,C
C
g
Figure 4: Flow Chart of the Procedure Used to Develop the Evaluation Model.
November 5, 2002
Caicedo, et al.
Element Type
Beam elements
Cable elements
Rigid links
Pier 4
Pier 3
Retained DOF
x, x
All DOF
x
Y
X
Pier 2
Bent 1
Cable anchorage
work point
Beam element
Rigid links
Rigid links
Rigid Link
Deck
Rigid Link
Cables
Y
Tower
Z
November 5, 2002
Caicedo, et al.
Cable element
Cable element
Beam element
Rigid links
Rigid links
Lumped mass
Lumped mass
large displacements (Wang et al., 1993). A nonlinear static analysis was performed using the commercial finite element program ABAQUS, giving the model tangent stiffness matrix at the
(deformed) equilibrium position. In ABAQUS, the B31 beam element was used for the structural beam element, and the element T3D2 was used for the cable elements.
In modeling the cables, the catenary shape and its variation with the axial force in the cable
are modeled using an equivalent elastic modulus (Ernst, 1965). The cable element is a large-displacement truss element that has a modified modulus of elasticity, E eq , given by
Ec
E eq = -------------------------------------------2
( wL x ) A c E c
1 + ---------------------------3
12T c
(1)
where A c is area of the cross-section, T c is the tension in the cable, w is its unit weight, L x is the
projected length in the X-Z plane, and E c is the modulus of elasticity of the material. The cable
stiffness contribution to the global stiffness matrix is only applied when the cable is under tension
and is omitted otherwise. The cable elements are modeled as truss elements in ABAQUS, and
their equivalent elastic moduli are used in the nonlinear static analysis.
The deck was modeled using the method described by Wilson and Gravelle (1991). In this
approach the deck is modeled as a central beam (the spine) which has no mass. Lumped masses
are employed to model the mass of the deck, which are connected to the spine using rigid links
(see Fig. 7). The masses are included to more realistically model the torsional response of the
deck to lateral loads, and have been shown to be important in the modeling of this structure
(Caicedo, et al., 2000).
Shear center
The deck is comprised of two main steel
girders along each longitudinal edge of the deck
Y
supporting the concrete slab (see Fig. 2). Thus,
the deck is treated as a C-shaped section as shown
Mass center
Z
in Fig. 8 (Wilson and Gravelle, 1991). Here the
steel beams are represented by the flanges of the
section, and the concrete slab is represented by
Figure 8: C-Shaped Section Used to
the web of the C-shaped section. The axial stiffDetermine Properties of Deck.
ness of the deck is calculated by converting the
area of the concrete slab into an equivalent area of
steel using the ratio of the two elastic moduli. The area of the equivalent section is 1.844m2. The
moments of inertia about the vertical and transverse axes are also obtained converting the conNovember 5, 2002
Caicedo, et al.
crete slab to an equivalent steel structure. The barriers and railings were not taken into consideration because they are not structural elements. The inertia of the typical deck section has values of
I yy =160.67m4, I zz =0.6077m4, and J eq =0.0677 m4. The neutral axis is located at 1.77m above the
bottom of the steel beams.
The calculation of the torsional stiffness of the deck section takes into consideration both
pure and warping torsional constants. The pure torsion constant is determined by (Wilson and
Gravelle, 1991)
n
Jt =
i=1
bi ti
---------3
(2)
where b i and t i are the length and thickness of thin sections which make up the deck cross-section. The warping constant is calculated as (Bleich, 1952)
2
2
d
2
d A
w = ----- I zz + e A 1 --------
4
4I yy
(3)
where d is the distance between the webs of the two steel beams located along the edges of the
deck, e is the distance between the neutral axis and the middle of the concrete slab, and A is the
equivalent cross sectional area. I yy and I zz are the moments of inertia of the deck about the Y and
Z axes, as determined previously. The torsional stiffness of the deck was obtained using the formula (Wilson and Gravelle, 1991)
2
G s J eq
Es w
= G s J + -----------------2
Gs L
(4)
where G s is the steel shear modulus of elasticity, J eq is the equivalent torsional constant, J is the
pure torsion constant, E s is the modulus of elasticity of steel, and L is the length of the main
span.
Calculation of the mass of the deck considers the steel beams, rigid concrete slab, barriers
and railings. The total mass of the deck per unit length was determined to be 2,645.7 kg/m
(55.2 slug/ft). To portray the behavior of the C-shaped section, the deck is represented as two
lumped masses, each having a mass equal to half of the total mass of the deck. The masses are
joined to the beam element by a rigid link as shown in Fig. 7. The vertical distance between the
lumped mass and the center of the beam is equal to the distance between the shear center and the
mass center of the C-shaped section shown in Fig. 8.
Because the mass moment of inertia of the main deck is different than the one induced by
the lumped masses, it is necessary to make corrections to those quantities. In the calculation, the
correction consists of finding the difference between the mass moment of inertia of the lumped
masses and that of the actual deck sections mass moment of inertia. This difference in the mass
moment of inertia is added to the node at the center of the deck to achieve the correct value of
mass moment of inertia in the section model. The mass moment of inertia of the lumped masses
with respect to the jth axis (either the X, Y, or Z axis), I j , is calculated using the formula
November 5, 2002
Caicedo, et al.
I j = 2M l r 2
(5)
where M l is the mass of each lumped mass, and r is the perpendicular distance from the mass to
each axis. The actual mass moment of inertia of the deck with respect to the jth axis, I mj , is calculated using the equation
n
I mj =
( Imi + mi ri2 ) .
(6)
i=1
where I mi is the mass moment of inertia of each of the component of the deck with respect to its
own centroidal axis, m i is the mass of each component, and r i is the perpendicular distance
between the centroid of each component and the jth axis. Thus, the corrected mass moment of
inertia of the section becomes
j = I mj I j
(7)
The value of this parameter about each axis for a typical section of the deck are X = 4.43 10 6
kg m2, Y = 4.45 10 6 kg m2, and Z = 18.3 10 3 kg m2. Negative values indicate that
the contribution of the lumped masses to the mass moment of inertia of the section is larger than
the mass moment of inertia of the actual section. Thus, to achieve the correct mass moment of
inertia for the section, a negative value is assigned to the spine to balance the larger value
included by the lumped masses when the rigid links are condensed out.
Problem Formulation
Consider the general equation of motion for a structural system subjected to seismic loads
MU + CU + KU = M x g + f
(8)
where U is the second time derivative of the displacement response vector U , M, C, and K are
the mass, damping and stiffness matrices of the structure, f [N] is the vector of control force
inputs, xg [m/sec2] is the longitudinal ground acceleration, is a vector of zeros and ones relating the ground acceleration to the bridge degrees of freedom (DOF), and is a vector relating the
force(s) produced by the control device(s) to the bridge DOFs. This equation is appropriate when
the excitation has a single component or when the excitation is uniformly applied at all supports
of the structure. The formulation of the equations expressing the response to multiple input component is somewhat different, however, from that described for the system having only a single
input component. For multiple supports, the resulting motion of the supports relative to each other
induces pseudo-static responses in the structure that must be considered in addition to the
dynamic responses.
For the analysis of the bridge with multiple-support excitation, the model must include the
degrees of freedom at the supports. The equation of dynamic equilibrium for all the DOFs is written in partitioned form (Chopra, 2001; Clough and Penzien, 1993)
November 5, 2002
10
Caicedo, et al.
M Mg
T
M g M gg
t
t
t
U + C C g U + K K g U = 0 + f
T
T
Pg
0
C g C gg U g
K g K gg U g
U g
(9)
where U and U g are the absolute displacement vector of the superstructure and the displacement
vector of the supports, respectively. Matrices M , C , and K are the system matrices of the structural model. Matrices M g , C g , and K g are the mass, damping and elastic-coupling matrices
expressing the forces developed in the active DOFs by the motion of the supports. Matrices M gg ,
C gg , and K gg are the mass, damping and stiffness matrices of the supports, respectively. It is
t
desired to determine the displacement vector U in the superstructure DOFs and the support
forces P g . Observe that control forces f are only applied to the superstructure DOFs.
t
s
The total displacement U is expressed as its displacement U due to static application of
the ground motion, plus the dynamic displacement U relative to the quasi-static displacement.
Therefore, the relationship between these displacement components is expressed by
t
s
U = U + U
0
Ug
Ug
(10)
K g K gg
s
U = 0
s
Ug
Pg
(11)
where P g are the support forces necessary to statically impose displacements U g that vary with
s
time; obviously, U varies with time and is therefore known as the vector of quasi-static displaces
ments. Observe that P g = 0 if the structure is statically determinate, or if the support system
undergoes rigid body motion; for the latter condition, an obvious example is identical horizontal
motion of all supports. The remainder U of the structural displacements is known as the dynamic
displacement vector because the quasi-static displacements cannot be evaluated by rigid-body
kinematics.
With the total structural displacements split into quasi-static and dynamic displacements,
we return to the first of the two partitioned equations
t
t
t
MU + M g U g + CU + C g U g + KU + K g U g = f
(12)
Substituting Eq. (10) and transferring all terms involving U g and U to the right side, leads to
s
s
s
MU + CU + KU = f ( MU + M g U g ) ( CU + C g U g ) ( KU + K g U g )
(13)
where the right-side term is called the vector of effective earthquake forces. The third term of the
right side in Eq. (13) drops out because Eq. (11) gives
November 5, 2002
11
Caicedo, et al.
KU + K g U g = 0
(14)
in which the displacement U by definition is the pseudo-static vector. Solving for these displacements leads to
s
U = K Kg Ug
(15)
Rs = K Kg .
(16)
The influence vector describes the influence of support displacements on the structural displacements. Finally, substituting Eq. (15) and (16) into Eq. (13) gives
MU + CU + KU = f ( MR s + M g )U g ( CR s + C g )U g
(17)
If the ground accelerations and velocities are prescribed at each support, this completes the
formulation of the governing equation.
For many practical applications, further simplification of the effective force vector is possible. The damping term is zero if the damping matrices are proportional to the stiffness matrices
because of Eq. (14); this stiffness-proportional damping may be unrealistic, however. While the
damping term is not zero for arbitrary forms of damping, it is usually small relative to the inertia
term and may therefore be dropped. Nevertheless, the damping term is included in the governing
equation in this benchmark problem.
The element mass and tangent stiffness matrices generated in ABAQUS are summed at
each node to assemble the global stiffness and mass matrices within MATLAB. The equations
are partitioned into active and constrained DOFs, and constraints were applied by condensing out
rigid links (applying kinematic constraints). The resulting model has 909 DOF for the superstructure plus 45 DOF for the supports.
Model Reduction
The model resulting from the finite element formulation has a large number of degrees-offreedom and high frequency dynamics. Thus, some assumptions are made regarding the behavior
of the bridge to make the model more manageable for dynamic simulation while retaining the fundamental behavior of the bridge. The active DOF retained in the model include: i) the nodes at the
top of each tower; ii) the lowest nodes at which cables are connected on each tower; iii) nodes at
the joints of the towers; iv) nodes or DOFs of elements whose shear and overturning moments are
among the design criteria; and v) approximately every third node of the bridge deck; and vi) rotational DOFs about the longitudinal and vertical axis of all spinal deck nodes. These locations are
indicated in the finite element model in Fig. 5. Note that the support DOF are not eliminated
because they are required for determination of the response of the structure to multiple support
excitation.
Static condensation is performed by first partitioning the mass and stiffness matrices corresponding to the structure DOF into active and dependent DOF as in
November 5, 2002
12
Caicedo, et al.
M =
M aa M ad
M da M dd
K aa K ad
K =
K da K dd
(18)
Assuming that no loads are applied to the dependent DOFs, the system equation for static equilibrium is written as
K aa K ad U
Pa
=
,
K da K dd U
0
(19)
is the active, and U is the dependent displacement vector. Using the second row of Eq.
where U
(19), the transformation matrix is obtained as
TR =
I
1
K dd K da
(20)
where T R is the static transformation matrix, and I is an identity matrix of appropriate size, such
that
U
.
= TR U
U
(21)
= T T KT .
K
R
R
(22)
The corresponding coefficient matrices for the ground excitation and the control forces are given
by
T
= TR .
= T R M and
(23)
Note that prior to making this transformation, and must be reordered into active and dependent DOF. Application of this reduction scheme to the full model of the bridge resulted in a 419
DOF reduced order model. The first 100 natural frequencies of the reduced model (up to 3.5 Hz)
were compared and are in good agreement with those of the 909 DOF structure.
The damping in the system is defined based on the assumption of modal damping. The
damping matrix was developed by assigning 3% of critical damping to each mode. This value was
selected to be consistent with assumptions made during the design of the bridge. The reduced system was used to construct the damping matrix using
November 5, 2002
13
Caicedo, et al.
= M
2 1 1 0
0
0
0
0 2 n n
(24)
where is the modal matrix, and i and i are the natural frequency [rad/sec] and modal damping ratio of the ith mode, respectively.
After model reduction the resulting equation of motion for the damped structural system
from Eq. (17) is
U
R
+C
+K
U
= f ( M
+ M )U g ( C
+ C )U g
U
R
M
g
s
g
(25)
= K
1 K and U
is the displacement vector of active structural DOFs. This model is
where R
s
g
designated the evaluation model. It is considered to portray the actual dynamics of the bridge and
will be used to evaluate various control systems. Note that this model always includes the effects
of the shock transmission devices, which constraint longitudinal motion. The evaluation model
and earthquake inputs are fixed for this benchmark problem. A representative sample of the mode
shapes is shown in Fig. 9 and two representative transfer functions of the model are shown in Fig.
10.
November 5, 2002
14
Caicedo, et al.
100
(a)
Magnitude (dB)
Magnitude (dB)
50
-50 -1
10
10
10
Phase (deg)
Phase (deg)
50
0
-50 -1
10
10
10
1000
0
-1000
-2000 -1
10
(b)
10
-1000
-2000
-3000 -1
10
10
10
10
Frequency (Hz)
Frequency (Hz)
An alternate model is developed to evaluate the robust performance of the proposed control
designs. To develop the alternate model, the effects of snow loads on the model are considered.
UBC (1991) codes indicate that the ground snow load in the Cape Girardeau region is 73.3 kg/m2
(15 psf) for 50-year mean recurrence interval. In addition to the snow load, 24.4 kg/m2 (5 psf)
should be added for zones with frequent rain and snow. Snow with a mass per length of 97.7 kg/
m2 (20 psf) are added to the deck, and its effect perturbs the mass matrix of the model. This
amount increases the mass of the bridge by approximately 3.5%. The nominal evaluation model
has the following first ten undamped natural frequencies: 0.2899, 0.3699, 0.4683, 0.5158, 0.5812,
0.6490, 0.6687, 0.6970, 0.7102, and 0.7203 Hz. The second evaluation model, considering snow
loads, has the following first ten undamped natural frequencies: 0.2790, 0.3559, 0.4489, 0.4969,
0.5589, 0.6230, 0.6406, 0.6809, 0.6816, 0.6971 Hz. The maximum variation in these frequencies
is 4.38%. The second model is to be used to investigate the robustness of the control system with
respect to the uncertainties in mass.
Analysis Tool
The linear model of the bridge system is simulated using a version of the analysis tool
developed by Ohtori and Spencer (1999) for linear systems. This tool allows the user to implement the compiled C code from within the MATLAB environment through a SIMULINK
block to simulate the responses of a seismically excited structural system. This tool solved the
incremental equations of motion using the Newmark- method in combination with the pseudo
force method. The program has been validated through a comparison with IDARC2D, a commercial structural analysis program. Although the full version of the program may be used for
nonlinear analysis, the version included in this benchmark problem is only applicable to linear
systems.
November 5, 2002
15
Caicedo, et al.
To use the code, one must define the mass, stiffness, and damping matrices for the evalua , and K
,C
in Eq. (25)), as well as the matrices defining the inputs and outputs
tion structure ( M
of the structural system. The input and output matrices are found using the state space form of Eq.
(25) given by
U g
x = A e x + B e U
U g
y = C e x + D e U
(26)
T U
T ] is the state vector, A is the state matrix, and B , C , and D are deterwhere x = [ U
e
e
e
e
mined by the inputs and outputs selected by the designer/researcher.
These matrices must be available in the MATLAB workspace to perform the simulation.
Note that the analysis tool does not require use of the A e matrix although it is available to the
researcher/designer for possible use in developing a control-oriented model of the bridge.
The sensors must be defined to measure the outputs of the evaluation model. Researchers/
designers must develop models for the sensors which must take the following form
s
s
x = g 1 ( x , y m, y f, t )
..
xg
(27)
ye
Evaluation model
yc
f
Control device(s)
yf
Control
Algorithm
ys
ym
Sensors
November 5, 2002
16
Caicedo, et al.
y = g 2 ( x , y m , y f, t )
s
(28)
s
where x is the continuous-time state vector of the sensor(s) and y is the continuous-time output
of the sensor(s) [Volts]. y f is the continuous-time output vector from the control device model
(see Eqs. (3335)), which may include forces produced by individual control devices, device
stroke, device acceleration, is used for evaluation of the control strategy and is available for feedback in the control algorithm.
Passive, active, and semi-active control devices (or combinations thereof) may be used in
designing control systems. For active/semi-active control systems, the associated discrete-time
control algorithm must take the form
c
x k + 1 = g 3 ( x k, y k, k )
c
u k = g 4 ( x k, y k, k )
(29)
(30)
where x k is the discrete-time state vector of the control algorithm at each sampling time t = kT ,
s
y k is the discrete-time input to the control algorithm from the sensors (which should be discretized in time and quantized to represent an A/D converter), and u k is the discrete-time control
command from the control algorithm.
Dynamic models of the control devices selected by the researcher/designer are not required
for this benchmark study. Ideal control devices may be assumed. Note that the program allows
designers/researchers to place control devices at constrained nodes although errors will result in
the simulated responses. To interface with the benchmark bridge model the control device
model(s) must take the form
f = g 5 ( y c, u k, t )
(31)
y f = g 6 ( y c, u k, t )
(32)
where y c contains the continuous-time responses from the evaluation model that influence the
control forces, and f is the continuous-time force output of the control device(s) applied to the
structure (in units of [kN]). Researchers/designers who choose to employ dynamic models of their
control devices should use the form
d
d
x = g 7 ( x , y c, u k, t )
(33)
f = g 8 ( x , y c, u k, t )
(34)
(35)
y f = g 9 ( x , y c, u k, t )
d
where x is the continuous-time state vector of the control device. Figure 12 provides the SIMULINK model used for evaluation of proposed control strategies.
November 5, 2002
17
Caicedo, et al.
Form Evaluation
Model Using
Code Provided
Define Device(s):
f = g 5 ( y c, u k, t )
y f = g 6 ( y c, u k, t )
Define Algorithm:
Define Sensor(s):
s
x
s
=
y =
s
g1 ( x ,
s
g2 ( x ,
y m, y f , t )
y m, y f, t )
c
xk + 1
c s
= g 3 ( x k, y k, k )
c s
u k = g 4 ( x k, y k, k )
or
d
x = g 7 ( x , y c, u k , t )
d
f = g 8 ( x , y c, u k , t )
d
y f = g 9 ( x , y c, u k, t )
Evaluate
Controller
Submit Codes
to Web Site
GUI Tool
A MATLAB-based graphical user interface (GUI) tool has been developed to aid the
researcher/designer in generating the input/output information for the evaluation model. The
graphical user interface allows the user to select the node numbers defining the evaluation outputs, y e , the connection outputs, y c , and the measured outputs, y m , for use in each control strategy. The location of the control devices may also be specified within the GUI. Once the control
system setup is specified, the user may choose to generate the evaluation model from within the
GUI or from the MATLAB command window directly.
November 5, 2002
18
Caicedo, et al.
Evaluation Criteria
For cable-stayed bridges subjected to earthquake loading, critical responses are related to
the structural integrity of the bridge rather than to serviceability issues. Thus, in evaluating the
performance of each control algorithm, the shear forces and moments in the towers at key locations (see Fig. 3) must be considered. Additionally, the tension in the cables should never
approach zero, and should remain close to the nominal pretension.
A set of eighteen criteria have been developed to evaluate the capabilities of each control
strategy. Because the earthquake is assumed to have two horizontal components at a specified
incidence angle, several of these criteria are evaluated in both the X (longitudinal) and Z (transverse) directions. The first six evaluation criteria consider the ability of the controller to reduce
peak responses, the second five criteria consider normed responses over the entire time record,
and the last seven criteria consider the requirements of the control system itself.
For each control design, the evaluation criteria should be evaluated for each of three earthquake records provided in the benchmark problem: i) El Centro. Recorded at the Imperial Valley
Irrigation District substation in El Centro, California, during the Imperial Valley, California earthquake of May, 18, 1940; ii) Mexico City. Recorded at the Galeta de Campos station with site Geology of Meta-Andesite Breccia in September 19, 1985; iii) Gebze, Turkey. The Kocaeli earthquake
recorded at the Gebze Tubitak Marmara Arastirma Merkezi on Aug. 17, 1999. The Mexico City
earthquake is selected because geological studies have indicated that the Cape Girardeau region is
similar to Mexico City. The El Centro and Gebze earthquakes allow for the researcher/designer to
test his/her control strategies on earthquakes with different characteristics. These three earthquakes are each at or below the design peak ground acceleration level for the bridge of 0.36 gs.
To consider multi-support excitations, the prescribed ground motion is assumed to be identical at each support, although it is not simultaneous. Bent 1 is assumed to experience a specified
ground motion, and the motion at the other three supports is delayed based on the distance
between adjacent supports and the speed of the L-wave of a typical earthquake (3 km/sec). For
each control design, the evaluation criteria should be evaluated for each specified excitation case:
Case A) an incidence angle of 15 with arrival times of [0 0.05 0.16 0.20] sec, and, Case B) an
November 5, 2002
19
Caicedo, et al.
incidence angle of 45 with arrival times of [0 0.03 0.12 0.15] sec. The incidence angle is defined
between the longitudinal direction of the bridge and the N-S wave of the earthquake. Researchers/
designers may evaluate their control designs for additional incidence angles by modifying the
input file.
The first two evaluation criteria are non-dimensionalized measures of the shear force at key
locations in the towers. The elevation of these key locations correspond to the tower base and the
deck level (see Fig. 3). The latter criterion was selected because this elevation corresponds to a
drastic reduction in the cross-sectional area of the towers. Evaluation criteria one and two are
given by
max F ( t )
i,t bi
-
J 1 = max --------------------------max
El Centro
F 0b
Mexico City
(36)
max F ( t )
i,t di
-
J 2 = max --------------------------max
El Centro
F 0d
Mexico City
(37)
Gebze
Gebze
max
where F bi ( t ) is the base shear at the ith tower, F 0b = max F 0bi ( t ) is the maximum unconi,t
trolled base shear (of the values at the two towers), F di ( t ) is the shear at the deck level in the ith
max
tower (see Fig. 3), F 0d = max F 0di ( t ) is the maximum uncontrolled shear at the deck level,
i,t
max
max
and
indicates absolute value. The values of F 0b , F 0d , and all other values used to normalize the evaluation criteria, are provided in Tables 78 of Appendix II.
The second set of evaluation criterion are non-dimensionalized measures of the moments in
the towers at the same key locations, given by
max M bi ( t )
i,t
-
J 3 = max ---------------------------max
El Centro
M 0b
Mexico City
(38)
max M di ( t )
i,t
-
J 4 = max ---------------------------max
El Centro
M 0d
Mexico City
(39)
Gebze
Gebze
max
where M bi ( t ) is the moment at the base of the ith tower, M 0b = max M 0bi ( t ) is the maximum
i,t
uncontrolled moment at the base of the two towers, M di ( t ) is the moment at the deck level in the
max
ith tower, and M 0d = max M 0di ( t ) is the maximum uncontrolled moment at the deck level in
i,t
the two towers.
The fifth evaluation criterion is a non-dimensionalized measure of the deviation of the tension in the stay cables from the nominal pretension, given by
November 5, 2002
20
Caicedo, et al.
T ai ( t ) T 0i T 0i
0c
Mexico City
(40)
Gebze
where T 0i is the nominal pretension in the ith cable, T ai ( t ) is the actual tension in the cable as a
max
function of time, and T 0c = max ( T a0i ( t ) T 0i T 0i ) is the normalized actual cable tension of
i,t
the uncontrolled system. This criterion is selected to reduce the likelihood of failure or unseating
of the cables.
The sixth evaluation criterion is a measure of the peak deck displacement at piers 1 and 4.
x bi ( t )
J 6 = max max ------------
i,t
El Centro
x 0b
Mexico City
(41)
Gebze
where x bi ( t ) is the displacement of the bridge deck at the ith location and x 0b is the maximum of
the uncontrolled deck response at these locations. This criterion is included to consider the likelihood of impact of the deck at these locations.
The seventh and eighth evaluation criteria are non-dimensionalized measures of the normed
values of the base shear and shear at the deck level in each of the towers, respectively, given by
max F bi ( t )
i
J 7 = max -----------------------------
F 0b ( t )
El Centro
(42)
max F di ( t )
i
J 8 = max -----------------------------
F 0d ( t )
El Centro
(43)
Mexico City
Gebze
Mexico City
Gebze
where F 0b ( t ) is the maximum of the normed value of the uncontrolled base shear of the two
towers, and F 0d ( t ) is the maximum of the normed value of the uncontrolled shear at the deck
level of the tower. The normed value of the response, denoted
, is defined as
tf
2
1
--- ( ) dt
tf
(44)
J 9 = max -------------------------------
M 0b ( t )
El Centro
(45)
Mexico City
Gebze
November 5, 2002
21
Caicedo, et al.
J 10
max M di ( t )
i
= max -------------------------------
M 0d ( t )
El Centro
(46)
Mexico City
Gebze
where M 0b ( t ) is the maximum of the normed value of the uncontrolled moment at the base of
the two towers, and M 0d ( t ) is the maximum of the normed value of the uncontrolled moment at
the deck level of the two towers.
The eleventh evaluation criterion is a non-dimensionalized measure of the normed value of
the deviation of the tension in the stay cables from the nominal pretension, given by
T ai ( t ) T 0i T 0i
(47)
Mexico City
Gebze
where T 0i is the existing pretension in the ith cable, T ai ( t ) is the actual tension in the ith cable as
a function of time, and T 0c = max ( T a0i ( t ) T 0i T 0i ) is the maximum of the normed value
i,t
of the actual cable tension for the uncontrolled
system.
The twelfth evaluation criterion deals with the maximum force generated by the control
device(s) and is described as
fi ( t )
(48)
Mexico City
Gebze
where f i ( t ) is the force generated by the ith control device over the time history of each earthquake, and W = 510, 000 kN (114,640 kips) is the seismic weight of bridge based on the mass of
the superstructure (not including the foundation).
The thirteenth criterion is based on the maximum stroke of the control device(s). This performance measure is given as
d
J 13
yi ( t )
(49)
Mexico City
Gebze
where y i ( t ) is the stroke of the ith control device over the time history of each earthquake, and
x 0max is the maximum uncontrolled displacement at the top of the towers relative to the ground.
When devices are used that do not have an associated stroke (e.g., tuned liquid dampers), the
researcher/designer should assume this evaluation constraint is zero.
The fourteenth evaluation criterion is a non-dimensionalized measure of the maximum
instantaneous power required to control the bridge, and is defined as
November 5, 2002
22
Caicedo, et al.
J 14
max P i ( t )
t
i
= max ----------------------------------
El Centro
max
x 0 W
Mexico City
Gebze
(50)
max
where P i ( t ) is a measure of the instantaneous power required by the ith control device, and x 0 max
is the peak uncontrolled velocity at the top of the towers relative to the ground. Values for x 0
are provided in Appendix II for each of the earthquakes specified. For active control devices,
d
d
P i ( t ) y i ( t )f i ( t ) , where y i ( t ) is the velocity of the ith control device. When semi-active
devices are employed, P i ( t ) is the actual power required to operate the device. For passive control devices, this criterion is zero.
The fifteenth evaluation criterion is a non-dimensionalized measure of the total power
required to control the bridge, and is defined as
J 15
tf
P ( t ) dt
i
= max i 0
.
El Centro --------------------------------
max
Mexico City
x0 W
Gebze
(51)
(52)
The seventeenth evaluation criterion is a measure of the total number of sensors required for
the proposed control strategy.
J 17 = number of sensors
(53)
The final evaluation criterion provides a measure of the resources required to implement the
control algorithm and is given by
c
J 18 = dim ( x k )
(54)
where x k is the discrete-time state vector of the control algorithm given in Eq. (29).
All criteria except J 5 , J 6 , J 11 , J 16 J 18 are dependent on the direction and should be evaluated for both the X and Z directions.
November 5, 2002
23
Caicedo, et al.
NORMED RESPONSES
CONTROL STRATEGY
Base Shear*
Base Shear*
Peak Force*
max F ( t )
i,t bi
-
J 1 = max --------------------------max
El Centro
F 0b
Mexico City
max F bi ( t )
i
J 7 = max -----------------------------
F 0b ( t )
El Centro
fi ( t )
Device Stroke*
max F ( t )
i,t di
-
J 2 = max --------------------------max
El Centro
F 0d
Mexico City
max F di ( t )
i
J 8 = max -----------------------------
F 0d ( t )
El Centro
yi ( t )
0
Mexico City
Overturning Moment*
Overturning Moment*
max M bi ( t )
i,t
-
J 3 = max ---------------------------max
El Centro
M 0b
Mexico City
max M bi ( t )
i
J 9 = max -------------------------------
M 0b ( t )
El Centro
Mexico City
Gebze
Gebze
Mexico City
Gebze
Gebze
Mexico City
Gebze
Gebze
Peak Power*
Mexico City
Gebze
Gebze
max M di ( t )
i,t
-
J 4 = max ---------------------------max
El Centro
M 0d
Mexico City
Gebze
Cable Tension
J 10
max M di ( t )
i
= max -------------------------------
M 0d ( t )
El Centro
Mexico City
Gebze
Cable Tension
T ai ( t ) T 0i
T ai ( t ) T 0i
-------------------------------
J 5 = max max -------------------------- J 11 = max max
i,t
T
El Centro i,t
El
Centro
T 0i
0i
Mexico City
Gebze
Mexico City
Gebze
max P i ( t )
i
J 14 = max t
---------------------------------El Centro
max
Mexico City
x0 W
Gebze
Total Power*
tf
P ( t ) dt
i
J 15 = max i 0
El Centro --------------------------------
max
Mexico City
x0 W
Gebze
J 18 = dim ( x k )
x bi ( t )
J 6 = max max ------------
i,t
El Centro
x 0b
Mexico City
Gebze
* Twelve evaluation criteria should be evaluated for both the X (longitudinal) and Z (transverse) directions.
A summary of the evaluation criteria is provided in Table 2. The values of the uncontrolled
responses for the three earthquakes required to calculate the evaluation criteria are provided in
Tables 78 of Appendix II. All criteria should be reported for each proposed controller and for
both Case A and B. The Mexico City, El Centro, and Gebze earthquakes should all be considered
in determining the evaluation criteria. However, designers/researchers are encouraged to include
November 5, 2002
24
Caicedo, et al.
additional criteria in their results if, through these criteria, their results demonstrate an overall
desirable quality. An example of such a situation might be a control system that performs well for
one type of earthquake but marginally for other earthquakes used to evaluate the control strategy.
Control Strategy Implementation Constraints and Procedures
November 5, 2002
25
Caicedo, et al.
10. Tension in the stay cables should remain within a recommended range of allowable values. A
lower bound is necessary to ensure that unseating of a cable does not occur, and an upper
bound provides a factor of safety to prevent failure of the cable. The tension in the ith cable
may not exceed 0.7 T fi or fall below 0.2 T fi , where T fi is the tension that would cause failure
of the ith cable. Values for T fi are provided in Table 9 (Appendix III).
11. Because the D/A converters have a range of 10 Volts, the command signal to each control
k
k
device has a constraint of max u i ( t ) 10 Volts, where u i ( t ) is the ith component of the
t
control signal.
12. Each control device employed should be described in terms of the maximum force that can be
generated. Researchers/designers must demonstrate that this force constraint is met during
each of the earthquakes.
13. Any additional constraints that are unique to each control scheme should also be reported (i.e.,
maximum stroke of control device, maximum velocity of control device, etc.). Control devices should be selected to allow for expansion of the briege due to temperature effects.
In the control of civil engineering structures absolute acceleration measurements are readily
available. Additionally, measurements of the control devices themselves are typically available.
Fourteen accelerometers and four displacement sensors are employed in the sample control system. Six accelerometers are located on top of the tower legs, including four oriented to measure
longitudinal (X) acceleration (nodes 240, 248, 353, 361), and two to measure transverse (Z) accelerations (nodes 240, 353). Eight accelerometers are located on the deck, including one at mid span
(node 34) oriented to measure longitudinal accelerations, and seven oriented to measure transverse accelerations (nodes 8, 11, 25, 34, 43, 57, 60). The natural frequency of the selected accelerometers are assumed to have a value that is at least an order of magnitude higher than the highest
natural frequency we are interested in controlling. Thus the selected accelerometers have a flat
response to approximately 3000 rad/sec (i.e., a constant magnitude and phase), and sensor dynamNovember 5, 2002
26
Caicedo, et al.
Selector
Selector
Selector
ics can be neglected. Two displacement sensors are positioned between the deck and pier 2 (node
pairs (84, 313), (151, 314)) and two displacement sensors are located between the deck and pier 3
(node pairs (118, 428), (185, 429)). These sensors are also assumed to have a flat frequency
response to 3000 rad/sec or greater. All displacement measurements are obtained in the longitudinal direction to the bridge (global X-direction).
To ensure that the accelerations and displacement measured on the bridge are within the
range of the A/D converters sensors are selected with a sensitivity of 7 V/g (i.e. 7 Volts = 9.81 m/
2
s ) for the accelerometers and a sensitivity of 30 V/m (i.e., 10 V = 0.33 m) for the displacement
sensors. Thus the sensor system is defined in the form of Eqs. (2728) as
s
y = Ds ym + v
(55)
where y is a vector of the measured absolute accelerations and device displacements in Volts,
y m is the vector of measured continuous-time absolute accelerations and device displacements in
physical units (i.e., [m/sec2] for accelerations and [m] for displacements), and v is the measurement noise, and
Ds =
I 14 14 G a
I8 8 Gd
(56)
where G a = 0.714 V/(m/sec2) is the sensor gain for acceleration and G d = 30.30 V/m. The sensor
block is represented in the SIMULINK block shown in Fig. 15. Note that in the sample controller
the device outputs are not measured, and therefore the corresponding signal y f is not connected to
the system, although it is available for participants to use. The gain block converts the continuoustime acceleration measurements from physical units to Volts. Finally, noise with an rms value of
0.03 V is included, as is specified in the control constraints, is added to the acceleration signal.
November 5, 2002
27
Caicedo, et al.
Control Devices
Tower
A total of 24 hydraulic actuators are
Control
employed as control devices and oriented
Devices
to apply forces longitudinally (X-axis), 8
Central Spine
between the deck and pier 2, 8 between the
Girder
deck and pier 3, 4 between the deck and
bent 1, and 4 between the deck and pier 4.
Four actuators are located between each of
the following pairs of nodes on pier 2 and
3: (84, 313), (151, 314), (118, 428), (185,
Edge Girder
429); two actuators are located between
each of the following pairs of nodes on
Figure 16: Schematic of Control System
bent 1 and pier 4: (68, ground), (135,
Design: Typical Tower Actuator
ground), (134, 444), (201, 440). The actuaImplementation.
tors have a capacity of 1,250 kN. Figure 16
shows the typical device layout at the
tower.
For this sample control design actuator dynamics are neglected and hydraulic actuators are
considered to be ideal. The equation governing the dynamics of the longitudinal actuators in the
form of Eqs. (3132) are
f = Kf u
(57)
yf = Dd u
(58)
where D d =125 kN/V (10V=1,250 kN) is the gain of the actuator, and K f is a matrix that accounts
for the gain of the actuator (i.e., the relationship between the input voltage and the desired control
force) as well as the fact that multiple actuators are used at each actuator location. For the sample
control design K f takes the form
2I 2 2
Kf =
0
D d = G dev D d .
4I 4 4
0
(59)
2I 2 2
(60)
November 5, 2002
28
Caicedo, et al.
Figure 17 shows the SIMULINK control device block. For the sample control there are no
connection outputs in the model of the devices because the actuator dynamics are neglected and
this ideal control device model does not require any inputs from the structure. Connection outputs
are available for other type of device implementations.
Control Design Model
A reduced order model of the system is developed for control design. This model, designated the design model, is formed from the evaluation model and has 60 states. The resulting
model has the same outputs as the evaluation model (see Eq. (26)). The reduced order model is
formed in MATLAB by forming a balanced realization of the system and condensing out the
states with relatively small controllability and observability grammians. The resulting state space
system is represented as follows
x r = A x r + B U g + E u
r
r
r
Ug
(61)
U g
z
+ Fr u
(62)
z r
z = Cr x + Dr
y r
ym = Cr x + Dr
U g
y
+ Fr u
U
(63)
where x is the design model state vector, A r and B r are the system matrices, z is the regulated
z
z
z
output vector, which is obtained from the mapping matrices, C r , D r and F r . Similarly, y m is the
y
y
y
measurement vector, which is obtained from the mapping matrices, C r , D r and F r . The gains of
the sensors and control devices (i.e., D s in Eq. (56) and D d in Eq. (59), respectively), as well as
the matrix defining the number of control devices ( G dev in Eq. (59)), are incorporated into this
model for control design. Thus, the inputs to the design model are the ground excitation at the
base of the structure and the control signals to the devices, whereas the inputs to the evaluation
model include the ground excitation and the applied control forces to the structure as in Eq. (25).
Control Algorithm
The sample controller employs a linear quadratic Gaussian (LQG) control design. For the
T
control system designs, the disturbances to the system (i.e., [ U Tg U Tg ] ), are assumed to be identically distributed, statistically independent stationary white noise processes, and S w = S ww I 8 8 ,
and an infinite horizon performance index is chosen, i.e.,
1
J = lim --- E
z r
T
z T
z r
z
(
C
x
+
D
u
)
Q
(
C
x
+
D
u
)
+
u
Ru
dt
r
r
r
r
November 5, 2002
29
(64)
Caicedo, et al.
q1 I4 4 0
0
q2
(65)
R is an [ 8 8 ] identity matrix. Further, for each case, the measurement noise is assumed to be
identically distributed, statistically independent Gaussian white noise processes, and
S ww S vi v i = = 25 .
The control and estimation problems are considered separately according to the separation
principle (Stengel 1986; Skelton 1988), yielding a controller of the form
u = K u x
(66)
where x is the Kalman Filter estimate of the state vector based on the reduced order model. By
the certainty equivalence principle (Stengel 1986; Skelton 1988), K u is the full state feedback
gain matrix for the deterministic regulator problem given by
1 ( N
+ BT P )
Ku = R
d
(67)
+A
TP PB R
1 B Td P + Q
0 = PA
(68)
= C z T QC z N
R
1 N
T
Q
d
d
(69)
= C z T QD z
N
d
d
(70)
= R + D z T QD z
R
d
d
(71)
= A B R
1 N
T.
A
d
d
(72)
and
Calculations to determine K u were done using the MATLAB (1997) routine lqry.m within the
control toolbox.
The Kalman Filter optimal estimator is given by
y r
y
r
r
x = A r x + B r u + L ( y m C r x D r u )
November 5, 2002
30
(73)
Caicedo, et al.
L = [R ( F r E r + C r S )]
(74)
0 = SA + A S SG S + H
(75)
and
yT
yT
A = A r C r R ( F r E r )
yT
(76)
G = Cr R Cr
yT
(77)
1 y
H = E r E r E r F r R F r E r
y yT
R = I + F r F r
(78)
(79)
Calculations to determine L were done using the MATLAB routine lqew.m within the control
toolbox.
For implementation on a digital computer, the controller is put in the form of Eqs. (2930)
using the bilinear transformation (Antoniou, 1993) to yield the following compensator
c
(80)
uk = Cc xk + Dc yk .
(81)
xk + 1 = Ac xk + Bc yk
c
Calculations to determine the discrete-time compensator were performed in MATLAB using the
c2dm.m routine within the control toolbox.
The SIMULINK block shown in Fig. 18 is used to represent the sample control algorithm
in the simulation. To represent the hardware used to implement this algorithm on a digital computer, the input signal passes through a model of an analog-to-digital converter (A/D) and the output control signal passes through a model of a digital-to-analog converter (D/A). The model
consists of a quantizer and a saturator as described in the Control Strategy Implementation Constraints and Procedures.
November 5, 2002
31
Caicedo, et al.
10000
Acceptable
Region
.7Tfi
8000
Actual
Region
6000
2000
(a)
0
0
.2Tfi
50
100
Cable Number
x 10
4000
10000
8000
6000
The closed-loop response is evaluated for the three earthquakes specified. Table 3 shows the
values of the evaluation criteria in Eqs. (3654). The responses of the controlled bridge are compared to those of the uncontrolled bridge for the El Centro, Mexico City, and Gebze earthquakes
in Figs. 1921. In each figure, the left plots show the maximum and minimum cable tension as a
function of cable number. The dark region provides the acceptable range of cable tensions as
specified in the control constraints (between the 0.2 T fi and 0.7 T fi ), and the lighter region provides a graphical description of the actual minimum and maximum cable tension. Note that the
uncontrolled cable tension falls below the lower bound in cables near the tower when subjected to
the El Centro earthquake. However, in each case the controlled cable tension is well within the
bounds. Additionally each figure provides a graph of the base shear at pier 2 in the longitudinal
direction. Note that the controller is able to achieve a significant reduction in the base shear forces
as compared to the uncontrolled system.
To demonstrate the feasibility of this controller, peak values of the force, stroke, and velocity are provided for each earthquake in Table 5. Note that the force requirement as well as the
velocity and displacement requirements are feasible in a device of this size.
4
2
0
-2
-4
4000
-6
0
2000
(b)
0
0
50
100
20
40
60
time (sec)
Cable Number
Figure 19: Simulated Responses to El Centro Earthquake (no snow, case A):
(a) Uncontrolled Cable Tensions, (b) Controlled Cable Tensions, and
(c) Uncontrolled and Controlled Base Shear Force Record (Pier 2).
November 5, 2002
32
Caicedo, et al.
8000
Actual
Region
6000
2000
(a)
0
0
.2Tfi
50
100
Cable Number
Cable Tension (kN)
x 10
4000
10000
8000
6000
Acceptable
Region
.7Tfi
10000
1
0
-1
-2
4000
2000
(c)
(b)
0
0
50
20
100
40
60
time (sec)
Cable Number
.7Tfi
10000
8000
Actual
Region
6000
4000
2000
(a)
0
0
Acceptable
Region
.2Tfi
50
100
Cable Number
10000
8000
6000
Figure 20: Simulated Responses to Mexico City Earthquake (no snow, case A):
(a) Uncontrolled Cable Tensions, (b) Controlled Cable Tensions, and
(c) Uncontrolled and Controlled Base Shear Force Record (Pier 2).
x 10
2
1
0
-1
-2
-3
4000
(c)
-4
0
2000
(b)
0
0
50
20
40
60
time (sec)
100
Cable Number
November 5, 2002
33
Caicedo, et al.
Mexico
Gebze
Max
Mexico
Gebze
Max
J1
J2
J3
J4
J5
J6
J7
J8
0.3314
0.4149
0.4736
0.4736
J9
1.0219
1.1169
1.0366
1.1169
0.8100
0.8284
0.9406
0.9406
0.9673
0.9984
0.9942
0.9984
J 10
0.3241
0.3964
0.4556
0.4556
J 11
1.0968
1.0793
1.0458
1.0968
0.6116
0.7658
0.9547
0.9547
1.0092
0.9924
1.0012
1.0092
0.2484
0.1208
0.1817
0.2484
J 12
J 13
J 14
1.0278
1.7826
2.4027
2.4027
0.2670
0.3270
0.3201
0.3270
J 15
1.0138
1.0560
1.0528
1.0560
0.8692
0.9641
0.9555
0.9779
0.9962
0.9908
0.2480
0.3193
0.3997
0.3997
1.0057
1.0526
1.0393
1.0526
0.6358
0.7930
0.7805
0.7930
1.0018
1.0044
1.0033
1.0044
0.6295
0.9707
1.0484
1.0484
X
Z
X
6.633e-
J 16
24
24
24
24
0.9641
J 17
14
14
14
14
0.9962
J 18
60
60
60
60
0.3714
0.4148
0.4276
0.4276
0.9908
1.0442
0.9979
1.0442
0.6899
1.0280
0.9366
1.0280
0.9836
0.9695
0.9888
0.9888
J 10
0.3616
0.4035
0.4908
0.4908
J 11
0.9793
1.0270
0.9928
1.0270
0.5997
0.6806
0.7362
0.7362
0.9968
1.0004
1.0022
1.0022
J5
0.2854
0.1220
0.1845
0.2854
J6
1.2433
2.1255
2.7008
2.7008
0.2771
0.3424
0.3536
0.3536
J 15
0.9823
1.0032
1.0255
1.0255
0.8537
0.9207
0.9869
0.9920
0.9716
1.0010
J1
J2
J3
J4
J7
J8
J9
J 12
J 13
J 14
November 5, 2002
0.2632
0.3469
0.4725
0.4725
0.9858
0.9976
1.0221
1.0221
0.6809
0.8387
0.9101
0.9101
1.0033
1.0032
1.0048
1.0048
X
Z
X
Z
X
J 16
24
24
24
24
0.9869
J 17
14
14
14
14
1.0010
J 18
60
60
60
60
34
Caicedo, et al.
Mexico
Gebze
Max
Mexico
Gebze
Max
J1
J2
J3
J4
J5
J6
J7
J8
0.3584
0.4847
0.4800
0.4847
J9
1.0218
1.1152
1.0366
1.1152
0.8531
0.8611
0.9734
0.9734
0.9674
0.9987
0.9939
0.9987
J 10
0.3358
0.4158
0.4633
0.4633
J 11
1.0954
1.0930
1.0454
1.0954
0.6424
0.7705
0.9789
0.9789
1.0094
0.9927
1.0011
1.0094
0.2461
0.1128
0.1780
0.2461
J 12
J 13
J 14
1.0535
1.8218
2.5460
2.5460
0.3035
0.3789
0.3262
0.3789
J 15
1.0137
1.0557
1.0528
1.0557
0.8759
0.9681
0.9755
0.9783
0.9968
0.9907
0.2619
0.3439
0.4179
0.4179
1.0057
1.0524
1.0393
1.0524
0.6427
0.8033
0.8047
0.8047
1.0018
1.0043
1.0033
1.0043
0.6522
1.0113
1.1110
1.1110
X
Z
X
J 16
24
24
24
24
0.9755
J 17
14
14
14
14
0.9968
J 18
60
60
60
60
J1
J2
J3
J4
0.4310
0.4935
0.4330
0.4935
0.9871
1.0360
0.9979
1.0360
0.7352
1.0520
0.9637
1.0520
0.9822
0.9735
0.9886
0.9886
J 10
0.3836
0.4295
0.5046
0.5046
J 11
0.9802
1.0270
0.9929
1.0270
0.6325
0.6956
0.7627
0.7627
0.9969
0.9999
1.0022
1.0022
J9
J 12
J 13
J5
J6
1.2584
2.1521
2.8727
2.8727
0.3191
0.4028
0.3628
0.4028
J 15
0.9827
1.0029
1.0256
1.0256
0.8672
0.9426
1.0058
0.9921
0.9717
1.0011
J7
J8
0.2820
0.1190
0.1791
0.2820
J 14
November 5, 2002
0.2796
0.3751
0.4932
0.4932
0.9859
0.9975
1.0221
1.0221
0.6887
0.8544
0.9392
0.9392
1.0032
1.0033
1.0048
1.0048
0.6583
1.0023
1.2756
1.2756
X
Z
X
J 16
24
24
24
24
1.0058
J 17
14
14
14
14
1.0011
J 18
60
60
60
60
35
Caicedo, et al.
Dir
El Centro
Mexico
Gebze
Max
1358
854.1
1442
1442
Stroke (m)
0.104
0.093
0.167
0.167
Vel (m/s)
0.809
0.394
0.527
0.809
1215
803.9
1131
1215
Stroke (m)
0.119
0.094
0.152
0.152
Vel (m/s)
0.633
0.441
0.449
0.633
Table 6: Maximum Actuator Requirements for Sample Control Strategy with Snow Loads.
Response
Dir
El Centro
Mexico
Gebze
Max
1421
930.4
1500
1500
Stroke (m)
0.106
0.096
0.177
0.177
Vel (m/s)
0.822
0.395
0.539
0.822
1279
844.7
1208
1279
Stroke (m)
0.120
0.096
0.161
0.161
Vel (m/s)
0.641
0.436
0.463
0.641
Closure
A second generation benchmark problem on the seismic control of cable-stayed bridges has
been developed. Phase II of this benchmark problem extends the problem to consider ground
motions with two horizontal components, as well as multi-support excitation. The evaluation
model of the Bill Emerson Memorial Bridge, the MATLAB (1997) files used for the sample
control design, and the simulation model, are available at:
http://wusceel.cive.wustl.edu/quake/
If you cannot access the World Wide Web or have questions regarding the benchmark problem
please contact Dr. Shirley Dyke via e-mail at: sdyke@seas.wustl.edu. Participants in the benchmark study will be expected to submit their control designs and supporting MATLAB files electronically for inclusion on the benchmark homepage, as described in the information distributed
with the benchmark cable-stayed bridge problem.
Acknowledgments
This research is supported in part by the National Science Foundation Grant No. CMS 9733272 (Dr. S.C. Liu, Program Director). The authors would like to thank Mr. Shyam Gupta of the
Missouri Department of Transportation and Mr. Bill Strossener from FHWA for pertinent information on the Cape Girardeau Bridge. The helpful advice of Prof. Yozo Fujino (University of
Tokyo), Prof. Masato Abe (University of Tokyo), Prof. Hirokazu Iemura (Kyoto University),
Prof. Joel Conte (University of California, Los Angeles), Prof. Petros Voulgaris (University of
November 5, 2002
36
Caicedo, et al.
Illinois), Prof. Fabio Biondini (Politecnico di Milano), and Gerry Pollok (ABAQUS), as well as
additional comments provided by members of the ASCE/IASC Task Group on Benchmark Structural Control Problems and other members of the structural control community, are gratefully
acknowledged.
References
ABAQUS (1998). Hibbitt, Karlsson & Sorensen Inc. Pawtucket, RI.
Agrawal, A., Yang, J.N., and He, W.L., (2002). Performance Evaluation of Some Semi-active
Control Systems for Benchmark Cable-Stayed Bridge, Proceedings of the Third World
Conference on Structural Control, Como Italy, April 711, 2002.
Antoniou, A. (1993). Digital Filters: Analysis, Design, and Applications, McGraw-Hill, Inc.,
New York, pp. 444446.
Bakule, L., Paulet-Crainiceanu, F., Rodellar, J., Rossell, J. (2002). Decentralized Overlapping
Control Design for a Cable-Stayed Bridge Benchmark, Proceedings of the Third World
Conference on Structural Control, Como Italy, April 711, 2002.
Bleich, F. (1952). Bucking Strength of Metal Structures, McGrawHill, New York.
Caicedo, J.M., Dyke, S.J., Turan, G. and Bergman, L.A., (2000). Comparison of Modeling Techniques for Dynamic Analysis of a Cable-Stayed Bridge. Proceedings of the Engineering
Mechanics Conference, ASCE, Austin, Texas, May 2123.
Caughey, T. K., (1998). The Benchmark Problem. Earthquake Engineering and Structural Dynamics, Vol. 27, pp.1125.
Chen, J-C. Ed. (1996) Proceedings of the 2nd International Workshop on Structural Control: Next
Generation of Intelligent Structures, Research Centre, The Hong Kong University of Science and Technology (see http://cwis.usc.edu/dept/civil_eng/structural/welcome.html).
elebi, M., (1998), Final Proposal for Seismic Instrumentation of the Cable-Stayed Girardeau
(MO) Bridge, U.S. Geological Survey.
Chopra A.K. (2001). Dynamics of Structures: Theory and Applications to Earthquake Engineering, Prentice Hall, Upper Saddle River, New Jersey.
Clough, R.W. and Penzien, J. (1993). Dynamics of Structures, 2nd Edition, McGraw-Hill, New
York.
Craig, Jr., R.R. (1981). Structural Dynamics, An Introduction to Computer Methods, John Wiley
& Sons, New York.
Dyke, S.J., Caicedo, J.M., Turan, G., Bergman, L.A., and Hague, S. (2002). Phase I Benchmark
Control Problem for Seismic Response of Cable-Stayed Bridges, Journal of Structural Engineering, ASCE, (in press).
Dyke, S.J., Turan, G., Caicedo, J.M., Bergman, L.A., and Hague, S. (2000). Benchmark Control
Problem for Seismic Response of Cable-Stayed Bridges, Proceedings of the Second European Conference on Structural Control, Paris, France, July 36.
Ernst, H.J. (1965). Der E-Modul von Seilen unter Berucksichtigung des Durchhanges, Der
Bauingenieur, Vol. 40, n. 2 (in German).
Hague, S. (1997). Composite Design for Long Span Bridges. Proceedings of the XV ASCE
Structures Congress, Portland, Oregon.
Jung, H.-J., Spencer, B.F. Jr., Lee, I.-W. (2002). Benchmark Control Problem of a Seismically
Excited Cable-stayed Bridge Using MR Dampers, Proceedings of the Third World Conference on Structural Control, Como Italy, April 711, 2002.
November 5, 2002
37
Caicedo, et al.
November 5, 2002
38
Caicedo, et al.
Appendix I - Nomenclature
A c, B c, C c, D c Discrete controller system matrices
A e, B e, C e, D e State space matrices of the evaluation model
z
A r, B r, C r , D r , C r , D r , E r, F r , F r Reduced model system matrices in which superscript z denotes the regulated outputs, and y denotes the measured outputs
A
Matrix used in the solution of the algebraic Riccati equation for Kalman estimator
Ac
bi
Long dimension of each element of the deck cross section used to obtain the pure
C gg, C g,
Ds
Dd
Ec
E eq
Es
Distance between the neutral axis of the deck and the center of the concrete slab
max
F 0b
F bi ( t )
F 0b ( t )
F di ( t )
F 0d ( t )
max
F 0d
fi ( t )
f c'
Matrix used in the solution of the algebraic Riccati equation for Kalman estimator
November 5, 2002
39
Caicedo, et al.
Gs
Ga
Gd
G dev
Gain factor to account for the number of control devices in control device model
g 1, g 2
g 3, g 4
g 5, g 6
Models for interfacing with the control device model (without device model)
g 7, g 8, g 9
Models for interfacing with the control device model (including device models)
Matrix used in the solution of the algebraic Riccati equation for Kalman estimator
Ii
Mass moment of inertia of the ith lumped mass in the calculation of the correction
Ij
I mi
axis (X,Y or Z)
Mass moment of inertia of each of the ith component of the deck with respect to its
I mj
centroidal axis
Mass moment of inertia of the section with respect to the jth axis
I zz , I yy
Moments of inertia of the bridge sections about the Z and Y axes, respectively
Jt
J eq
Equivalent torsion constant of the deck taking into consideration pure torsion and
J 1, J 2, , J 18
warping torsion
Evaluation criteria
K
K gg, K g,
Ku
Full state feedback gain matrix for the deterministic regulator problem
K el
Identity matrix
K aa, K ad, K da, K dd Partitions of the global stiffness matrix (for static condensation)
Ku
Discrete-time index
November 5, 2002
40
Caicedo, et al.
Lx
M el
max
M 0b
M bi ( t )
M 0b ( t )
max
M 0d
M di ( t )
M 0d ( t )
Ml
mi
Matrix used in the solution of the algebraic Riccati equation for LQR controller
Pa
Load vector
Pi ( t )
Pg
Support forces
P gs
Rs
Matrix used in the solution of the algebraic Riccati equation for Kalman estimator
r
ri
Distance from the lumped masses, perpendicular to the axis under consideration in
the correction of the mass moment of inertia of the deck
Distance from the ith element of the deck, perpendicular to the axis under consid-
Matrix used in the solution of the algebraic Riccati equation for LQR controller
Psuedo-static influence vector
November 5, 2002
41
Caicedo, et al.
Svi vi
Sw
S ww
TR
Discrete-time step
T 0i ( t )
T ai ( t )
T fi
Tc
Time variable
ti
Thickness of each element of the deck cross section used to obtain the pure torsion
constant of the deck, J t
tf
Ug
uk
k
ui ( t )
v
w
W
Sensor noise
Weight per unit length of cables
Seismic weight of the bridge (not including the foundation)
xk
r
x
Kalman filter estimate for the state vector of reduced order (design) model
November 5, 2002
42
Caicedo, et al.
x bi
Displacement of the bridge deck at the ith location (at pier 1and 4)
x 0b
x 0max
Maximum uncontrolled displacement at the top of the tower relative to the ground
x 0max
Maximum uncontrolled velocity at the top of the tower relative to the ground
xg
ground acceleration
yc
ye
yf
yk
ym
y id ( t )
X , Y , Z
Modal matrix
November 5, 2002
43
Caicedo, et al.
Case B
Mexico
Response
Dir
Definition
max
F 0bx (kN)
max F 0bxi ( t )
max F 0bzi ( t )
max F 0dix ( t )
max F 0dzi ( t )
max M 0bxi ( t )
max M 0bzi ( t )
max M 0dxi ( t )
M 0dz (kN-m)
max M 0dzi ( t )
x 0b (m)
max x 0bi ( t )
1.009e-1
F 0bx ( t ) (kN)
max F bxi ( t )
F 0bz ( t ) (kN)
max F bzi ( t )
F 0dx ( t ) (kN)
max F dxi ( t )
F 0dz ( t ) (kN)
max F dzi ( t )
M 0bx ( t ) (kN-m) X
max M bxi ( t )
M 0bz ( t ) (kN-m) Z
max M bzi ( t )
M 0dx ( t ) (kN-m) X
max M dxi ( t )
M 0dz ( t ) (kN-m) Z
max M dzi ( t )
max
F 0bz (kN)
max
F 0dx (kN)
max
F 0dz (kN)
max
M 0bx (kN-m)
max
M 0bz (kN-m)
max
M 0dx (kN-m)
max
i,t
i,t
i,t
i,t
i,t
i,t
i,t
i,t
i,t
i
i
i
i
i
i
i
i
El Centro
5.204e-2
Gebze
6.950e-2
El Centro
9.558e-2
4.419e-2
Gebze
5.618e-2
x 0max (m)
max x 0 ( t )
0.1647
0.09555
0.1593
0.1802
0.09650
0.1265
x 0max (m/s)
max x 0 ( t )
1.213
0.6948
0.7862
1.146
0.6086
0.7010
November 5, 2002
t
t
44
Caicedo, et al.
Case A
Mexico
Case B
Mexico
Dir
Definition
max F 0bxi ( t )
max F 0bzi ( t )
1.436e+4
max F 0dix ( t )
max F 0dzi ( t )
max M 0bxi ( t )
max M 0bzi ( t )
max M 0dxi ( t )
M 0dz (kN-m)
max M 0dzi ( t )
x 0b (m)
max x 0bi ( t )
1.007e-1
F 0bx ( t ) (kN)
max F bxi ( t )
F 0bz ( t ) (kN)
max F bzi ( t )
F 0dx ( t ) (kN)
max F dxi ( t )
F 0dz ( t ) (kN)
max F dzi ( t )
M 0bx ( t ) (kN-m) X
max M bxi ( t )
1.289e+5
M 0bz ( t ) (kN-m) Z
max M bzi ( t )
M 0dx ( t ) (kN-m) X
max M dxi ( t )
M 0dz ( t ) (kN-m) Z
max M dzi ( t )
max
F 0bx (kN)
max
F 0bz (kN)
max
F 0dx (kN)
max
F 0dz (kN)
max
M 0bx (kN-m)
max
M 0bz (kN-m)
max
M 0dx (kN-m)
max
i,t
i,t
i,t
i,t
i,t
i,t
i,t
i,t
i,t
i
i
i
i
i
i
i
i
El Centro
1.8e+4
5.266e-2
7.5e+4
Gebze
El Centro
Gebze
6.950e-2
9.575e-2
4.464e-2
5.618e-2
x 0max (m)
max x 0 ( t )
0.1626
0.09487
0.1593
0.1830
0.09585
0.1265
x 0max (m/s)
max x 0 ( t )
1.199
0.6935
0.7862
1.144
0.6109
0.7010
November 5, 2002
t
t
45
Caicedo, et al.
Ultimate Load
(MN)
4.94
39
12.58
158,190,222,254
4.98
39
12.84
159,191,223,255
4.59
39
11.62
160,192,224,256
4.27
39
10.88
161,193,225,257
3.97
39
10.15
162,194,226,258
3.60
37
9.78
163,195,227,259
3.30
36
9.27
164,196,228,260
3.15
34
9.20
165,197,229,261
2.83
35
8.18
166,198,230,262
2.64
34
7.67
Cable No.
(See Fig. 1)
157,189,221,253
Service Load
(MN)*
167,199,231,263
2.58
35
7.26
168,200,232,264
2.47
36
6.92
169,201,233,265
2.40
36
6.58
170,202,234,266
2.01
36
5.57
171,203,235,267
1.78
35
5.06
172,204,236,268
1.62
35
4.69
173,205,237,269
1.46
31
4.73
174,206,238,270
1.75
35
5.03
175,207,239,271
1.98
36
5.51
176,208,240,272
2.00
35
5.72
177,209,241,273
2.41
36
6.73
178,210.242,274
2.55
37
6.97
179,211,243,275
2.73
36
7.50
180,212,244,276
2.93
36
8.08
181,213,245,277
3.15
35
8.97
182,214,246,278
3.89
42
9.23
183,215,247,279
3.80
38
9.90
184,216,248,280
4.06
39
10.50
185,217,249,281
4.30
40
10.78
186,218,250,282
4.44
38
11.62
187,219,251,283
5.05
39
12.88
188,220,252,284
4.18
33
12.59
November 5, 2002
46
Caicedo, et al.