1. INTRODUCTION
1.1 Steel Building Industry and Past Development
Steel materials are widely used in Japan due to high strength and ductility for relatively small
volume, ease for fabrication of members and joints, well-established hysteretic behaviour, and
economical cost. From 1986 to 1991, Japan was in an economic bubble, and much technical
investment was made in developing special steel materials and members1). Fire resistant steel, SN
steel, low-yield point steel, and buckling-restrained brace were developed, which promote more use
of the steel materials/members for seismic design.
The 1995 Kobe earthquake indicated problems of steel beam-column connections of moment-
resisting frames, and resulted in improved moment connection details and welding procedures, as
well as use of the newly developed SN steel for the beam section where full yielding is expected.
On the other hand, the earthquake also indicated importance of keeping damage small enough to
allow post-quake use of the building, and promoted accelerated use of base isolation or
supplemental damping schemes2).
Steel buildings have advantages of easiest installation of dampers as well as well-defined elastic
(undamaged) state of the frame, and are qualified to be value-added buildings, performing
excellently against major earthquakes. For tall buildings, steel frame is typically used, and they are
combined with dampers in Japan especially after the 1995 quake2). On the other hand, the dampers
have been rarely used for majority of multi-story steel buildings of 6-story or shorter, although they
were recognized to be much more susceptible to earthquakes3).
Seismic Effects on Indoor situations Outside situations Buildings other than houses PGA
Intensity people
6-lower Difficult to A lot of heavy and Strongly and Less earthquake-resistant 2.50–
(5.5–5.9) keep unfixed furniture severely felt buildings easily receive heavy 3.15
standing. moves and falls. It outside. Light posts damage and may be destroyed. m/s²
is impossible to swing, and electric Even highly earthquake-
open the door in poles can fall down, resistant buildings have large
many cases. All causing fires. cracks in walls and will be
objects will shake moderately damaged, at least. In
violently. some buildings, wall tiles and
windowpanes are damaged and
fall.
6-upper Impossible Most heavy and Trees can fall down Many walls collapse, or at least 3.15–
(6.0–6.4) to keep unfixed furniture due to violent are severely damaged. Some 4.00
standing and moves and becomes shaking. Bridges less earthquake-resistant m/s²
to move displaced. and roads suffer buildings collapse. Even highly
without moderate to severe earthquake-resistant buildings
crawling. damage. suffer severe damage.
A typical gymnasium is a steel structure (S-type) composed of moment resistant frames in the Y-
direction and braced frames in the X-direction (Figs. 1 and 2). Other types combine a reinforced
concrete frame with either a steel frame (RS-type) or a steel roof (R-type), as shown in Fig. 2.
Since Japan’s seismic design code was drastically improved in 1981, the 216 steel school
buildings except 15 buildings (construction year unidentified) are categorized as follows.
1) Post-1981 group: 64 buildings (30% of total) satisfying the code.
2) Stronger Pre-1981 group: 40 buildings (19% of total) constructed in or before 1981 with
seismic resistance equivalent to 1), or retrofitted to the same level.
3) Weaker Pre-1981 group: 97 buildings (45% of total) constructed in or before 1981, who
were neither retrofitted nor seismically evaluated.
steel rigid frame steel rigid frame
RC frame
foundation foundation
S-type RS-type
steel roof
anchor
Note:
Connection between steel
member and concrete frame or RC frame
foundations is of an exposed
type using anchor bolts foundation
R-type
Fig. 1. Typical structure of a school gymnasium (S-type) Fig. 2. Three structure types
As Table 3 indicates, the majority of the post-1981 and stronger pre-1981 groups showed either
slight or minor damage, whereas 34% of the weaker pre-1981 group showed major damage. Table 4
indicates statistics for structural components of damage rank IVs or Vs only (Table 1) in the
buildings with major damage. In post-1981 group, the most severe damage occurred at column
bases with high rate of 73%, and in both stronger pre-1981 and weaker pre-1981 groups, at diagonal
braces with high rates of 50% and 61%, respectively. Note that “column base” includes the roof-
column anchor connection of the R-type gymnasiums shown in Fig. 2.
Table 3. Building categories and damage statistics (%) Table 4. Building categories and component damage (%)
Mod- Tsu- Col. Diag. Foun-
Group * Slight Minor erate Major nami Group ** base brace dation Col. Beam
Post-1981 7.8 45.3 12.5 15.6 18.8 Post-1981 72.7 18.2 9.1 0.0 0.0
Stronger pre-1981 5.0 50.0 20.0 17.5 7.5 Stronger pre-1981 33.3 50.0 8.3 0.0 8.3
Weaker pre-1981 8.2 23.7 18.6 34.0 15.5 Weaker pre-1981 23.3 60.5 11.6 4.7 0.0
* 64, 40, and 97 buildings from top. ** 11, 12, and 43 major damage buildings from top.
(structural components with damage rank IVs or Vs only)
2.3 Damage to Diagonal Braces
Diagonal braces were in 127 buildings, and 79 (62%) and 42 (33%) buildings used angle section
and round bar with turnbuckle, respectively. The latter was mainly used in weak pre-1981 group.
Remaining 6 (5%) buildings include tubes and flat bars. The following focuses on fracture of the
braces (defined as damage rank IIIs or above, Table 2):
Fracture of the angle brace occurred in 15 buildings mostly in the weak pre-1981 group, and the
failure rate out of the 79 buildings was 19%. Figs. 3a-c for instance, show pull-out failure at the
brace end, brace net section fracture, and pull-out fracture at the gusset plate, respectively.
On the other hand, fracture of round bar brace with turnbuckle occurred in 28 buildings almost
exclusively in the weak pre-1981 group, and the failure rate out of the 42 buildings was 67%, much
higher than the angle brace case mentioned above. Figs. 4a-c for instance, show fracture at the
threaded ends of the rod, fracture at the bolts, and fracture of turnbuckle, respectively. Note that the
turnbuckle brace in Fig. 4c was constructed in the 1990s, when the “Japan Industrial Standard
turnbuckle braces for buildings” was not commonly used, and the ductility of the brace not always
ensured.
Since 1981, the brace connection, including the bolt, weld, and gusset plate, has been required
to have a higher ultimate strength than the yield strength of the brace’s effective cross section.
Fracture of the angle brace, round bar with turn buckle, and connections occurred in the pre-1981
group indicate the brace and connection conceivably did not meet the requirement.
(a) Pull-out fracture at brace end (b) Net section fracture (c) Pull-out fracture at gusset plate
Fig. 3. Fractures at angle brace
(a) Fracture at threaded end (b) Fracture of bolt (c) Fracture of turnbuckle
Fig. 4. Fractures at round bar brace with turnbuckle
(a) Anchor yielding caused cracks (b) Pull-out fracture of anchor bolts (c) Anchor & concrete damage (R-type)
Fig. 5. Failures at the base of steel column or steel roof (R-type gymnasium)
(c) (d)
NS EW
(a) 29-story building (b) NS (x-) and EW (y-) frames Falling of books at story level 25 Elevator cable damage
(other building)
Fig. 6. Super-tall building in central Tokyo Fig. 7. Acceleration-induced damage
mode has been obtained with excellent accuracy4), 5). Fig. 8b shows that the acceleration of each
mode at the top floor is dominated by the 2nd mode for the first 90 seconds. For the later 110
seconds, the 1st mode response increased and became dominant. Although not shown, for the 16th
floor the 2nd mode was much more dominant, producing almost the same acceleration as top floor.
This is due to the 2nd mode vector giving the largest modal responses around the mid-height of the
building5).
Top acceleration
Base acceleration
From available response records of other super-tall buildings and some interviews to the
occupants, it appears that behavior of various elements (non-structural components, equipment, and
contents) can be explained by their vibration periods and building modal properties as discussed
above. Previously, little attention was paid to the acceleration-induced damage which, however, can
become enormous for the future shaking 3 to 4 times stronger ((2) of Sec. 1.2).
PGA = 595cm/s2
El-Centro eq.
(NS)
El-Centro earthquake (NS)
Metropolitan near-field eq.
Acc. (cm/s2)
(a) Time histories of ground motions (b) Pseudo-velocity response spectrum (5%damping)
Fig. 9. Ground motions used for design validations
cm/s. The scaled El Centro earthquake is one of the major earthquakes considered for seismic
design of tall buildings in Japan.
Limiting the story drift ratio (drift angle) within 0.01 rad is considered to minimize damage to
structural and some non-structural components. Other non-structural components, equipment, and
contents are checked against accelerations, and their reinforcements, if needed, are being explored
experimentally and analytically. In this manner, efforts are being made to assure safety of many
students and staff as well as continuity of research and education, and to utilize the building as a
haven for students who, like many commuters (Sec. 1.2), would lose transportations to their homes
25FL 25FL
20FL 20FL
15FL 15FL
10FL 10FL
0 0.5 1 1.5 2 0 0.2 0.4 0.6 0.8 Case (a) Case (b) 0 0.5 1 1.5 2 0 0.2 0.4 0.6 0.8
Fig. 10. αNi and αRi indicators for x-frame effectiveness Fig. 11. αNi and αRi indicators for y-frame effectiveness
Fig. 11 left shows two cases of locations in y-direction frame, and Fig. 11 right plots αNi and αRi
for the two cases. Case (b) damper locations give much higher αNi and αRi than Case (a). The
building has very slender braced frames that tend to deform in a cantilever mode. Inserting
dampers to the bay between such frames is effective, since shear deformation of the bay is
amplified by relative vertical displacements of the columns on the opposite sides of the bay. Case
(b) is more effective, since the relative displacements are larger.
In this manner, αNi and αRi can indicate appropriate locations of dampers and effectiveness of the
existing frame. Their use for various checks associated with design and analysis of supplementally
damped steel buildings is recommended in the AIJ specifications (Chap. 4).
Kd Cd
Kb
Fd, ud
Fig.13a shows example locations of 68 dampers in total of x- and y-directions, where average of
about one damper in each direction is considered per story. The damper price is only about 20% of
the total cost of about $100,000 per location, including structural, and largely non-structural and
cosmetic works. Thus, limiting the number of dampers (and locations) is a key to control the cost.
The stories for damper installations were selected in the order of high product αNi αRi.
Fig. 13b shows analytical results from the scaled 1940 El Centro earthquake (NS) that produce
the largest peak responses, as understood from the largest spectral response (Fig. 9b) at the
buildings fundamental vibration periods of about 3 sec. in both directions. Compared with the non-
retrofitted case, story drift ratio and acceleration are reduced to 0.74 and 0.64 times, respectively.
31
16 68
dampers εPL = 0.08
11
No
damper
εPL = 0.10
6
No
damper
1
(a) (b)
0 0.005 0.01 0.015 0 500 1000 1500
(a) (b)
Story drift ratio Accel. (cm/s2)
Fig. 13. Reduced number of oil dampers for economical Fig. 14. Existing steel brace and damage (with no
and effective retrofit (68 dampers = 2.3 dampers /story). dampers, hypothetical Tokyo near-field earthquake)
The case using 68 dampers shows almost the same performance as the 100 damper case.
Fig. 14 shows large cumulated plastic strains and deformations of an existing x-direction brace
that were obtained by using the frame response histories due to the Tokyo near-field earthquake.
Although not shown, the cumulated plastic strain is reduced by a factor of three using the dampers.
START No.
Height ≦ 60 m ?
Yes.
Set seismic level &
story drift limit Do static analysis of m-m model Do time history analysis Do time history analysis
and confirm local response of shear spring model and of m-m model and
confirm global response confirm both global &
Do preliminary design of
local responses
frame only by static Do static analysis of m-m model
analysis of m-m model and confirm local deformability Do static analysis of m-m
against 1.5 times the model and confirm
Decide damper sizes by story drift limit local response
direct displacement-
based design method
END Route 1 END Route 2 END Route 3
Fig. 15. AIJ Steel Damper and Frame Specifications (“m-m model” means the member-to-member model).
drift distributions, even when the frame tends to deform non-uniformly2), 3), 10), 11). Since the
publication 10), these methods have been extended to different types of dampers (e.g. Sec. 3.2) 2), as
well as inelastic frames including reinforced-concrete frames and timber frames. They are adopted
in several other design specifications in Japan in order to provide better seismic protections.
REFERENCES
[1] Kanno R. , Tsuji M., Hanya K., et al., 2012. “Steels, Steel Products and Steel Structures Sustaining
Growth of Society (Infrastructure Field) “, Nippon Steel Technical Report, No. 101, Nov.
[2] JSSI, 2003, 2005, 2007, 2013, Design and Construction Manual for Passively Controlled Buildings,
Japan Society of Seismic Isolation, Tokyo (in Japanese).
[3] AIJ 2014, Design Specifications on Steel Frame and Damper Nov. (in print, in Japanese)
[4] Kasai K., Mita A., Kitamura H., et al., 2013, “Performance of Seismic Protection Technologies during
the 2011 Tohoku-Oki Earthquake”, Earthquake Spectra, Vol.29, No.S1, pp.S265-S293, March.
[5] Kasai K., Pu W., and Wada A., 2012. “Responses of tall buildings in Tokyo during the 2011 Great East
Japan Earthquake”, Keynote Paper, Proc. STESSA 2012, pp. 25-35, Santiago, Chili.
[6] Matsumoto Y., Yamada S. et al. 2012, “Damage to Steel Educational Facilities in the 2011 East Japan
Earthquake: Part 1”, Proceedings, 15th World Conf. on Earthq. Eng. (15WCEE), Lisbon, Portugal.
[7] Japan Building Disaster Prevention Association, 2001. Guideline for Post-Earthquake Damage
Evaluation and Rehabilitation Technique (in Japanese).
[8] Hisada Y., Yamnashita T., et al. 2012, “Seismic Response and Damage of High-Rise Buildings in
Tokyo, Japan, during the 2011 Tohoku Earthquake”, Proceedings, 15WCEE, Lisbon, Portugal.
[9] Kasai K. and Iwasaki K., 2006. “Reduced Expression for Various Passive Control Systems and
Conversion to Shear Spring Model”, J. Str’l & Const. Eng., AIJ, No.605, pp.37-46 (in Japanese)
[10] Kasai K., Fu Y., and Watanabe A., 1998. “Passive Control Systems for Seismic Damage Mitigation,”
Journal of Structural Engineering, American Society of Civil Engineers, 124 [5], pp.501-512.
[11] Kasai K. and Ito H., 2005. “Passive Control Design Based on Tuning of Stiffness, Yield Strength, and
Ductility of Elastoplastic Damper”, J. Str’l & Const. Eng., AIJ, No.595, pp.45-55 (in Japanese)
[12] Kasai K., Nakai M., Nakamura, H., Asai H., Suzuki Y., and Ishii M., 2008. “Current Status of Building
Passive Control in Japan”, Proceedings, 14WCEE, Beijing, China.