Professional Documents
Culture Documents
Paper Paper Paper Paper Paper
Paper Paper Paper Paper Paper
Akira T AIRA
TAIRA Masato Y ASUHARA
YASUHARA
Engineer, Technical Div.,
Steel & Hybrid Structure G., Central Japan Railway Company
Structure Technology Development Div.
Keywords : Steel column member, Steel rigid frame pier, Ductility, Earthquake resistance
performance, Hysteresis model
;
Dimension
; ;
;;;
Specimens h (mm) b f (mm) t f (mm) b w (mm) t w (mm) b s (mm) t s (mm)
A-1 1500 450 8 450 7 43 4.5
;;
A-2 1500 410 12 380 12 - -
;;;;;
A-3 1500 450 9 420 9 75 8
A-4 1500 450 9 420 9 50 4.5
Parameters
;
- -
Specimens Rr λ γ/γ* λs N/N y Section type
A-1 0.532 0.197 0.627 0.711 0.20 Stiffened section
A-2 0.604 0.227 - - 0.16 Unstiffened section
;
A-3 0.464 0.216 3.31 0.295 0.16 Stiffened section
A-4 0.464 0.211 2.92 0.147 0.18 Stiffened section
※ Yield stress...309.2(N/mm 2) [t = 7mm], 305.7(N/mm 2) [t = 8mm], 302.4(N/mm 2) [t = 9mm], 298.4(N/mm2) [t = 12mm], here t : thickness
※ Lateral diaphragm spacing...500mm (A-1 ~ A-3), 250mm (A-4)
;;;
;;
Fig. 1 Steel column specimen and loading method
Fig. 2 Horizontal load (P)- horizontal displacement ( δ )
hysteresis curve of A-3
Fig. 3 Non-dimensional P- δ envelope curves Fig. 4 Envelope curve of test results and skelton curve
non-dimensionalized at the yield point. Py and δy means mum bearing capacity is based on references 4)-6); in
the yield point load and displacement respectively. The the tests, however, local buckling was observed at this
following tendencies can be recognized based on Fig. 3. stage and it is possible that bearing capacity could sub-
(1) A-2 - A-4 showed somewhat of an improvement in sequently drop rapidly as local buckling progresses de-
bearing capacity and ductility compared to A-1. pending on column proportions. In addition, this point
(2) A-3, in which Rr is smaller and γ/γ* is greater, dem- essentially corresponds to the point at which there is a
onstrated the most outstanding ductility in the rapid increase in displacement in the vertical direction.
present series of tests with a gentle decrease in bear- Based on these, we set the ultimate point at 95% of maxi-
ing capacity also after the peak. mum load.
It is possible to categorize the steel member damage
level as shown in Table 2 based on the load-displacement
3. The steel column damage level relationship and conditions of specimen damage.
θn – 3.5
= 0.0471 (1 + N/N y)R r ⋅ λ + 2.94 5. Evaluation of steel rigid frame pier earthquake
θy
resistance performance
[unstiffened]
Where, 5. 1. A Cyclic loading test of a steel rigid frame pier
My0 : yield moment without the action of axial force specimen
Nu : strength of the central axis compressive column
NE : Euler buckling load of cantilever column In steel railway structures, there are extremely few
θy0 : rotation corresponding to My0 column piers, while, on the other hand, the rigid frame
E : Young’s modulus, G : shear modulus pier is utilized widely. We conducted the cyclic loading
I : cross-sectional secondary moment test using the model steel rigid frame pier specimen. A
Aw : web cross-sectional area summary of the specimen is given in Fig. 6 and the speci-
h : height of the column fications are shown in Table 3. We made the specimens
[Stiffened] and [unstiffened] refer to stiffened and as relatively thick unstiffened sections, assuming a scale
unstiffened sections, respectively. The method of calcu- of about 1/3rd of the actual size of the structure. Since
lation of the maximum load point and ultimate point is column-beam connections in ordinary railway structures
based on the test results and the range of applicability is commonly have haunches, haunch plates are installed on
as indicated below. the inside.
-
[Stiffened] : 0.3 -< Rr -
< 0.7, 0.2 -
< λ-< 0.5, As indicated in Fig. 6, we applied a constant load in
0.0 -
< N/Ny - < 0.3, γ/γ* -> 1.0 the vertical direction (vertical load that is 0.05 relative
-
[Unstiffened] : 0.3 - < Rr -
< 0.9, 0.2 -
< λ-< 0.5, to the axial force of the column) and a cyclic load through
0.0 -
< N/N y -
< 0.2 displacement control in the horizontal direction (in-plane
Fig. 5 shows a comparison of the proposed formulae direction), the same as column members, on load beams
and test results for deformation at the time of maximum giving pin support at two points. The yield point was
load and maximum bearing capacity and deformation at considered to be the point which strain measured in the
the ultimate point. Based on Fig. 5, although disparities column reached yield strain.
can be observed to a degree, the proposed formulae are Fig. 7 indicates the load-displacement hysteresis
thought to accurately grasp the affects on bearing capac- curve at the central point of the load beam in cyclic load-
ity and ductility of each parameter. ing. The major damage conditions of the specimen col-
umns are indicated in the diagram.
The degree of specimen damage is as indicated be-
low. The axial direction strain of the flange on the lower
part of the southern column reached compressive yield
strain first. We set this point as 1δy and local buckling
Fig. 8 A simple analytical model the analysis results generally coincided with the point
at which load decreased to about 95% of the maxi-
then occurred in the column base at about 4 δy, maxi- mum load. The load generally coincided.
mum load was essentially reached at about 5 δy and lo- In the case of rigid frame piers having the same struc-
cal buckling also occurred at the top of the column. Sub- ture parameters as the test specimens, it is thought pos-
sequently, the load gradually decreased as local buckling sible to evaluate the earthquake resistance performance
progressed and, at 8 δy, cracking occurred in the loca- of the structure using the simple analysis model indicated
tions where buckling had occurred in the column base. here.
This relation of load-displacement relationship and de-
gree of specimen damage are thought to be nearly the
same as in column member tests. Damage was concen- 6. Hysteresis models
trated in the column members and no external damage
was recognized in the beam and column-beam connections It is necessary when conducting dynamic analysis to
until the final loading stage. use a hysteresis model appropriately expressing the hys-
teresis characteristics of the member.
5. 2. Pushover analysis using a simple analysis model Models capable of analysis with good precision as far
as the post-peak zone after maximum load have been pro-
We conducted pushover analysis using a simple analy- posed as steel member hysteresis models 4), 7). These mod-
sis model and compared it to the load-displacement enve- els, however, are not incorporated in universal analysis
lope line in the test results. As indicated in Fig. 8, we software. Since the decline in unloading stiffness is
developed the analysis model with linear beam members thought to be small at the point at which the load de-
and beam elements with non-linear stiffness springs at creases to 95% of maximum load, we examined standard
the end of column members, taking into consideration the hysteresis models in which unloading stiffness was
damage conditions in the tests. The skeleton curve of the equivalent to initial stiffness. We examined 2 standard
member hysteresis model was formed as a tri-linear models without unloading stiffness decrease, one in which
model, as indicated in Fig. 4. The break points of the the skeleton curve is bi-linear and the other in which it is
skeleton curve were calculated based on the procedures tri-linear. The standard bi-linear model is a model that
presented in section 4 (Quantitative evaluation of steel has a bi-linear skeleton curve (Fig. 10) directed from the
member bearing capacity and ductility) above. In addi- point at which the yield point is extended by 1.3 times to
tion, regarding the column members, since the axial force the point at which the load decreases to 95% of maximum
fluctuated, we set the interaction curve of bending mo- load. The setting of this break point takes into account
ment and axial force, taking their affects into account. the fact that bearing capacity also increases a certain
Fig. 9 shows a comparison of the results of pushover degree at about initial stiffness after the yield point (Fig.
analysis and the envelope curve in the test results. The 3). The standard tri-linear model basically has a skel-
points that are labeled as damage level limiting points 1, eton curve in which the break points are joined as indi-
2 and 3 in the diagram indicate the points at which col- cated in Fig. 11; however, we set the load of break point 2
umn members reached the limiting points in the push- at 95% of maximum load and the gradient thereafter at
over analysis. In addition, the damage that appeared zero.
earliest in the envelope line is also indicated. The follow- Fig. 11 shows a comparison of the load-displacement
ing tendencies were recognized based on Fig. 9. hysteresis curve in the test results for A-3 and the hys-
(1) The displacement at damage level limiting point 2 in teresis model. Fig. 11 also shows a comparison with ac-
the analysis results essentially coincided with the cumulated hysteresis absorption capacity (area included
displacement at the maximum load point in the tests. within the load-displacement hysteresis curve). Based
The load was somewhat smaller than that in the test on Fig. 10, if we include the point at which the load de-
results. creases to 95% of maximum load, the bearing capacity is
(2) The displacement at damage level limiting point 3 in underestimated with the standard bi-linear model, though
ǛE*1.0E5(kN㨯mm)
400 2.5 tri-linear model
400
200 200 2
Ǭ(mm) Ǭ(mm)
0 0
-40 -20 0 20 40 -40 -20 0 20 40 1.5
-200 -200
1
-400 -400
-600 0.5
-600
-800 test test
-800 0
calculation calculation 0 10 20 30
-1000 -1000 Ǭ(mm)
8. Acknowledgement