Professional Documents
Culture Documents
a
Kajima Architectural and Engineering Design, 6-5-30, Akasaka, Minato-ku, Tokyo 107, Japan
b
Kajima Technical Research Institute, 2-19-1, Tobitakyu, Chofu-shi, Tokyo 182, Japan
Abstract
The Nagano Olympic Memorial Arena, located in Nagano City, Japan, is where the 18th winter Olympic games will be held in
1998. It is one of the world’s largest speed skating arenas, covering an 80 m ⫻ 216 m free space containing a 400 m speed skate
oval and seating for 10,000 spectators.
One of the main features of the structure is its semi-rigid hanging roof spanning 80 m. Each hanging member is composed of
two glulams sandwiching a steel plate.
Hanging structures generally incorporate bracing members to increase roof stability and to resist wind up-lift and lateral load.
This structure instead utilizes the bending stiffness of the glulams and the in-plane stiffness of plywood panels attached to roofing
members. However, few studies have been carried out to evaluate the dynamic structural characteristics of this kind of structure.
This paper describes experimental and analytical investigations conducted to verify this structure’s wind-resistant design and confirm
its structural safety against strong wind load. 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Hanging roof; Composite beam; Glulam and steel plate; Dynamic behavior; Dynamic wind load
0141-0296/99/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 9 8 ) 0 0 0 3 0 - 3
S. Ban et al. / Engineering Structures 21 (1999) 770–785 771
and confirm its structural safety against strong wind of 15 structural units. Each unit has two reinforced con-
load. crete counterweights, two leaning walls, and a hanging
First, dynamic loading tests were conducted on a 1/4- roof panel pin connected to the top edges of the leaning
scale hanging roof panel model to determine its funda- walls. The center roof unit is the highest, and roof unit
mental dynamic characteristics. Next, dynamic loading height reduces in 3-m gaps toward the outside units. The
tests were conducted on an actual hanging roof panel to configuration of the hanging roof portion is the same for
confirm its dynamic characteristics. In these tests, the all units, and is as follows: span, 80 m; sag, 5 m; and
effects of oil-dampers attached to the hanging roof width, 18 m. The basic load bearing mechanism of the
panels to increase damping and reduce its response were roof unit is shown in Fig. 2. The details of each structural
also investigated. Three-dimensional wind-induced portion are summarized in follows.
response analyses considering geometrical nonlinearity
were carried out to confirm its structural safety against 2.1. Hanging roof
fluctuating wind force.
This paper summarizes these experimental and ana- A section of the roof panel is shown in Fig. 3. Each
lytical investigations. main beam is a composite member composed of glulams
and a steel plate. Two glulam beams (300 mm ⫻
125 mm ⫻ 2) sandwich a steel plate (200 mm ⫻ 12 mm)
2. Outline of structural system and they are connected to each other with steel bolts at
2 m pitch. The beams are arranged at 600 mm intervals.
The structural system of this speed skate oval is out-
Thus, one roof panel consists of 30 main beams pin sup-
lined in Fig. 1. As shown, the overall structure consists
ported at the ends of the steel plate, and connected to
each other at 10-m intervals by steel tie plates.
Plywood panels 12 mm thick are nailed to the top sur-
Table 1
Similarity law for scale model test
Parameter Dimension Full scale value Scale model value Typical quantity
Table 2
Comparison of frequencies and dampings
No. Mode shape Frq. (Hz) Free vibration Forced vibration Microtremor
1st Vertical 1st Frq. (Hz) 1.20 1.22 1.23 1.24 1.23
Damping (%) 1.22 1.29 1.18 0.56 0.58
2nd Torsional 1st Frq. (Hz) 1.79 1.80 1.81 1.82 1.82
Damping (%) 1.11 0.92 0.90 0.55 0.23
3rd Torsional 2nd Frq. (Hz) 1.96 1.97 2.00 2.03 2.02
Damping (%) 1.34 0.94 0.72 0.43 0.44
4th Vertical 2nd Frq. (Hz) 2.03 2.06 2.07 2.07 2.08
Damping (%) 1.06 1.00 1.11 0.50 0.42
Table 3
Comparison of frequencies and dampings (damper case)
Without damper 1.20 苲 1.24 0.56 苲 1.29 2.03 苲 2.08 0.42 苲 1.11
4-dampers at corner 3 kgf/kine 1.18 3.40 2.01 2.73
10 kgf/kine 1.22 4.48 2.02 2.58
7-dampers at side 3 kgf/kine 1.31 5.53 1.94 1.51
10 kgf/kine 1.32 3.43 1.93 1.32
mode (torsional 2nd) for the 1/2-edge case and sym- 4.2.2. Microtremor measurement test
metric vertical vibration mode (vertical 2nd) for the 1/2- Microtremors were measured when there was a gentle
mid case. wind and no construction noise. One measurement frame
The average natural frequencies and damping factors lasted 200 s, and transfer functions were calculated by
deduced from the waveforms measured in the free averaging ten measurement frames.
vibration tests are listed in Table 4. However, the values Transfer functions at representative measurement
of the 1/4-edge and 1/2-edge cases have uncertainties points (1/4-mid point, 1/2-mid point) to the supporting
because of their waveforms. point are shown in Fig. 19. In the transfer functions,
778 S. Ban et al. / Engineering Structures 21 (1999) 770–785
Fig. 16. Measuring points for actual roof panel test. 4.3. Comparison of test results
Table 4
Natural frequencies and damping factors (free vibration)
cal approach. However, the torsional 2nd and vertical actual hanging roof panel, which is supported by less-
2nd mode shapes are not so clear as those of the analyti- rigid posts and stays; and (2) greater material damping
cal results and show coupled vertical and torsional of the glulam is expected in the actual panel, while single
modes. These tendencies also appear in the 1/4-scale timbers were used in the 1/4-scale model.
model test results. This is considered to be because of
possible variations in the stiffness and mass distributions 4.4. Considerations
and the initial shape of the hung roof, for such non-uni-
form material as the timber. Basic vibration modes of the actual hanging roof panel
The natural frequencies and damping factors are com- were determined through the tests. The test results for
pared in Table 7. Here, the natural frequencies for the the actual panel generally agree with the previously
1/4-scale model are converted to actual-size values based obtained scale-model test and analytical results. How-
on the similarity law. ever, the second-order torsional and second-order verti-
The natural frequencies of microtremors of the actual cal vibration modes in the actual panel tests are not as
panel exhibit slightly reversed values between the tor- clear as those of the analytical results. The same trend
sional 1st and torsional 2nd modes. This is inferred to is also found in the 1/4-scale model test results.
be because of the imbalance of stiffness and mass among The natural frequencies obtained from the actual panel
actual hung members, and because the torsional 1st, tor- tests generally agree with the previous results. However,
sional 2nd, and vertical 2nd vibrations originally have those of the first-order and second-order torsional
close natural frequencies. Meanwhile, the natural fre- vibrations are not in the same order of magnitude. This
quency of the vertical 1st mode of the actual panel is in can be explained by the following: the second-order tor-
good agreement with the previous results, and this is the sional and second-order vertical modes originally have
main vibration mode for wind-induced responses. close natural frequencies, and their modes are coupled
The damping factors in the actual panel tests are due to the variations and imbalance in the mass and the
greater than 1%, which is slightly larger than those of stiffness of the actual hung members.
the 1/4-scale model test results. This can be explained The damping factors for the basic vibration modes
by the following: (1) dissipation damping occurs in the range between 1 and 2%, and are slightly greater than
780 S. Ban et al. / Engineering Structures 21 (1999) 770–785
Table 5
Natural frequencies and damping factors (microtremors)
Table 6
Comparison of vibration modes
Vertical 1st
[0.62 Hz]
0.56 Hz 0.57 Hz
0.63 Hz
Torsional 1st
0.77 Hz
Torsional 2nd
[1.02 Hz]
(0.90 Hz) 0.86 Hz
0.92 Hz
Vertical 2nd
[1.04 Hz]
0.92 Hz 0.96 Hz
0.95 Hz
Table 7
Comparison of natural frequencies and damping factors
Vibration mode Frequency Actual panel test 1/4-scale model test Analysis
damping factor
Vertical 1st Frequency (Hz) 0.56 0.57 0.60 苲 0.62 0.62 0.62 0.63
Damping F. (%) 1.06 1.06 1.18 苲 1.29 0.56 0.58 -
Torsional 1st Frequency (Hz) (0.85) 0.88 0.90 苲 0.91 0.91 0.91 0.77
Damping F. (%) (4.37 苲 4.70) 1.46 0.90 苲 1.11 0.55 0.23 -
Torsional 2nd Frequency (Hz) (0.9) 0.86 0.98 苲 1.00 1.02 1.01 0.92
Damping F. (%) (2.32 苲 2.35) 1.12 0.72 苲 1.34 0.43 0.44 -
Vertical 2nd Frequency (Hz) 0.92 0.96 1.02 苲 1.04 1.04 1.04 0.95
Damping F. (%) 1.01 苲 1.16 1.70 1.00 苲 1.11 0.50 0.42 -
782 S. Ban et al. / Engineering Structures 21 (1999) 770–785
⑀a 1/Ea ⫺ ba/Eb 0 0 0 a
冦冧冤 冥冦 冧
⑀b ⫺ ab/Ea 1/Ea 0 0 0 b
Fig. 21. Modeling of shell with orthotropic material. ␥ab ⫽ 0 0 1/Gab 0 0 ab (1)
␥bc 0 0 0 1/Gbc 0 bc
for the full-size composite beams [1] were carried out ␥ac 0 0 0 0 1/Gca ac
to evaluate the structural characteristics of the hanging
beam by the authors. In the test, a general part and a where the subscripts (a, b, c) indicate the principal
glulam joint part of them were tested. The test results material directions.
showed that the axial and bending stiffnesses of the glu- ⑀a, ⑀b and a, b Normal strains and
lam joint part of the beam were a little lower than those stresses in the a, b
of the general part. Furthermore, the test results showed directions, and
that the axial stiffness can be evaluated considering only ␥ab, ␥bc, ␥ca and ab, bc, Shear strains and stresses
the steel plate section of the composite hanging beam, ca
and the bending stiffness can be evaluated as the effec- Ea: a-direction elastic
tive stiffness by averaging the values for the general and modulus, set equal to
the joint part. Fig. 22 shows the cross-section of the 25.41 (tf/cm2)
equivalent steel beam element employed in the analysis, Eb: b-direction elastic
whose axial tensile stiffness and out-of-plane bending modulus, set equal to 1.75
stiffness were equivalent to those of the composite beam. (tf/cm2)
ab: Poisson ratio, set equal to
5.1.2. Orthotropic material model 0.1 (Eb ab ⫽ Eaba),
As the second step in the modeling, the parameters Gab: Shear modulus, set equal
for an orthotropic linear elastic material representing the to 0.70 (tf/cm2)
hanging roof panel must be determined. The stress–strain Gbc: Shear modulus, set equal
relationship for the orthotropic linear elastic material is to 0.84 (tf/cm2)
defined by Eq. (1). Gca: Shear modulus, set equal
The additional bending moment owing to the plywood to 9.86 (tf/cm2)
panels attached to the top surface of the glulams was
neglected. This membrane member was assumed to be
located in the center of the equivalent steel hanging
beam, as shown in Fig. 21. The thickness of a shell 5.2. Free vibration analysis
element with the orthotropic material is equal to the
height (356.7 mm) of the equivalent steel hanging beam. A free vibration analysis is conducted for one of the
The a-direction elastic modulus Ea parallel to the hanging roof panels. The computer program ‘ADINA’
composite hanging beam is set equal to the axial stiff- [8] is employed. The analytical model is shown in Fig.
ness and the bending stiffness of the equivalent hanging 23, although the oil dampers are not considered in the
beam, including the plywood panel. Eb, the modulus at free vibration analysis. The steel tie beam is modeled as
right angles to the hanging beam, is equal to the axial the truss element. In the geometric nonlinearity problem,
the total tangential stiffness matrix [KT] is composed of
the large displacement matrix [KL], the initial stress
matrix [KS] and the ordinary stiffness matrix [K0].