Professional Documents
Culture Documents
Journal of Constructional Steel Research: Kami Imani, Hossein Showkati, Tohid Ghanbari Ghazijahani
Journal of Constructional Steel Research: Kami Imani, Hossein Showkati, Tohid Ghanbari Ghazijahani
Journal of Constructional Steel Research: Kami Imani, Hossein Showkati, Tohid Ghanbari Ghazijahani
a r t i c l e i n f o a b s t r a c t
Article history: Circular thin plates are extensively found in numerous types of engineering applications. Nonetheless, there have
Received 17 May 2011 been few experimental studies investigating the real behavior of reinforced circular plates under uniform loading.
Accepted 10 August 2012 Present work empirically studies buckling/failure behavior of uniformly loaded, circumferentially reinforced
Available online 7 November 2012
circular plates with clamped brims. To this end, a number of locally thickened/stiffened circular steel plates
were tested under uniform pressure. Obtained findings were processed to examine the structural behavior of
Keywords:
Circular stiffened/thickened plates
such structures. In addition, results were verified by FE analysis of corresponding tested models. Acceptable agree-
Uniform pressure ment was ultimately detected between both experimental and numerical results which were basically aimed at
Experiments generalizing the behavior of circular plates in full scale within the range of the current work.
Buckling and failure © 2012 Elsevier Ltd. All rights reserved.
Yield lines
0143-974X/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jcsr.2012.08.001
K. Imani et al. / Journal of Constructional Steel Research 80 (2013) 308–316 309
Upper ring
Plate
Base brink
2. Experimental program with the accuracy of 0.001 mm were attached over the desired spots
in order to measure vertical deflections. During the experiments, a digi-
2.1. Testing apparatus tal pressure manometer was used to gauge the amount of the pressure.
All results were saved in a computer file using a data logger as well as a
A schematic illustration and an overall view of the experimental digital scanner.
set-up are presented in Figs. 1 and 2. This machine was designed by In all, seven tests were performed. The specimens were labeled in a
Golzan and Showkati [11,12] to conduct some empirical studies re- way that ECC specimen was the plain specimen without reinforcement;
garding circular plates and conical shells with different geometries ECP specimens were thickened with annular plates in the edge zone and
and edge conditions. ECS specimens were stiffened circumferentially. Thin steel plates were
The strain values were recorded by strain gauges of YEFLA-5 (series used to manufacture the specimens, stiffeners and local thickening ele-
YF/Y), Tokyo Sokki Kenkyujo Co. Ltd. Also, KYOWA type digital LVDTs ments as well. The material properties and stress-strain curve (Fig. 3)
Data logger
Test rig
LVDTs
Circular plate
Strain gauges
Pressure gauge
Control valve
3. Experimental observations
150000
3.1. General observations and milestones
100000
50000
It is interesting to note that the deformational behavior of the
present specimens subjected to uniform loading can be divided into
0 the following phases. In the first stage of loading, the bending defor-
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 mations were in a quite static manner and could not be detected by
Strain naked eyes. Flexural deflections were gradually formed in the central
part of the specimen with further increasing the pressure. As the deflec-
Fig. 3. Stress-strain curve of the material.
tions were increased, the specimen shrank in both radial and circumfer-
ential directions which led to the compressive stresses to be developed
were obtained from standard tension coupon test conducted on the in the circumferential direction. Notice that nonlinear elastic character-
strips of the steel material used for the specimens. Geometric properties istic (is seen in the load deflection curves) appeared in the specimens in
of the models are shown in Table 1 and Fig. 4(b). the pre-buckling region. Under a certain amount of pressure, onset of
Clamped edge condition was provided for all of the specimens in a buckling occurred and subsequently the first distinct wave/waves of
way that in-plane forces during testing were free to be released. All spec- buckling could be observed. Through further increasing the pressure,
imens were meshed to the parts according to the layout of Fig. 4(a). Geo- nearly non-periodical buckling waves grew near the edge zone of the
metric imperfections were also measured through a method described specimens gradually and in a mild manner to cover the whole circum-
thoroughly in Ref. [12]. ference of the plates so that overall buckling was completed (Fig. 5).
Thus, one can observe that post-buckling potency can be found in a sta-
2.2. Testing process ble manner in all the present specimens.
It should be mentioned that, the transition from the first wave to
After a careful fabrication procedure, developed to achieve high the fully overall buckling was quite gradual as the lobes were devel-
quality specimens, in order to install the specimens over the test rig, oped one after the other or in pairs near the edge zone. In most of
the base brink of the model was placed in the groove of a rigid circular well-fabricated models with small out-of-plane initial imperfections,
rim, which was in turn situated on vertical supports. To keep the the initial buckling was accompanied by the occurrence of all the
brink properly in the groove, it was covered by a grooved rubber waves and led to therefore complete buckling. After the buckling of
and both the groove of rubber and the rim were filled with silicon the plates was fully developed, the loading was kept increasing up
sealant. Following that, by connecting the vacuum pump to the rig to the plastic regime till plastic collapse appeared by folding of the
and installing the measuring probes, test was ready to be performed. yield lines on some buckled areas, ultimately causing local plastic col-
However, it is important to consider that prior to each test and in lapse (Fig. 6).
order for system conditioning, the load was increased up to a certain Take notice that despite extreme care during specimen fabrication,
value (less than 10% of the numerically predicted pressure) and then, due to unavoidable geometric imperfections of the plates, the buckling
the plate was unloaded. This was done so as to confirm the reliability waves appeared with rather different wave lengths and amplitudes
of the whole system. which caused a kind of asymmetric buckled shape. It was observed
Loading was performed in a way in which the air inside the pres- that, with more increasing of the load, the waves were considerably
sure chamber was incrementally expelled. In all stages, all aspects of deepened before the failure stage.
the test were under control. A release valve was employed in order It should be highlighted that yield lines were developed in the
to control the speed of loading and unloading, so that the results plates. In addition, due to the geometric shrinkage of the specimens
could be closely and exhaustively detected. It should be mentioned in radial direction and out-of-plane deformational behavior of the
that loading was increased uniformly and incrementally up to the buckled waves, the edge of the specimen was released from the
first wave formation, then loading was continued to reach the failure rings holding it and accordingly collapse happened with brim local
mode and final plastic collapse. Note that the speed of loading during luxating of the plate specimens (see Fig. 7). It can be stated that for
the tests was always below 3 kPa/minutes in order to keep the static all the specimens buckling occurred circumferentially in a wavy
mode near the boundary zone, while the central region remained al-
most flat and undistorted.
Table 1 In terms of buckling mode, it was seen that for ECC specimen, buck-
Geometric specifications of the test models.
ling waves were developed almost in a non-harmonic form and the
a) Stiffened specimens number of buckling lobes in such specimen was less than that in ECP
Specimen code r (mm) t (mm) ts (mm) rs (mm) hs (mm) and ECS specimens. On the other hand, for ECP specimens which were
thickened with annular plate one can observe nearly symmetric buck-
ECS1 300.0 0.5 0.5 295.0 10.0
ECS2 300.0 0.5 0.5 290.0 10.0 ling waves before the failure rather than other specimens. It was also
ECS3 300.0 0.5 0.5 275.5 10.0 clear that there was a short interval between the initial buckling and
ECS4 300.0 0.5 0.5 260.0 10.0 the failure for ECC specimen and the failure occurred quickly after buck-
ling, whereas for reinforced specimens this interval was quite distin-
b) Plain specimen and thickened specimens
guishable. Thus, one can conclude that strengthening the circular
Specimen code r (mm) t (mm) tth (mm) rth (mm)
plates may lead to stable development of the post-buckling stage in
ECC 300.0 0.5 – – comparison to the plates without reinforcement.
ECP1 300.0 0.5 0.5 40.0 Experimental load-displacement and load-strain curves are illustrat-
ECP2 300.0 0.5 0.5 90.0
ed in Figs. 8 and 9. Since the tested specimens were under “load control”
K. Imani et al. / Journal of Constructional Steel Research 80 (2013) 308–316 311
hs
ts
t t+t th t
r r
rth rs
Table 2 shows the buckling pressure of the tested specimens and the
Fig. 5. Wavy mode of the specimen before failure. ratio of the buckling load of the reinforced specimens over the buckling
312 K. Imani et al. / Journal of Constructional Steel Research 80 (2013) 308–316
a b
Fig. 6. Final deformed pattern of (some visible buckled waves): (a) ECS1 and (b) ECS3.
load of the plane specimen (ECC). Table 2 also gives a ratio for weight in- stresses developed by the shrinkage of the stiffened area, cause the ring
crease caused by reinforcement for each specimen. These results compar- stiffener to be tilted, as it is manufactured from thin-walled elements.
atively demonstrate the improvement of the buckling load of different Fig. 10 presents a typical tilting behavior of the ring stiffener attached
reinforced specimens and subsequently, the efficiency of edge-zone rein- to the specimens. One may observe that for each case the buckled
forcement of the plates for different cases. For instance, for the specimen wave number of the ring (tilting) is consistent with the circumferential
ECP1 and ECP2 the buckling load increases 2.37 and 3.44 times by in- buckling mode of the plate body nearby.
creasing 24.9% and 51% of the material consumption. Above all, for the
ECS specimens, increasing the material of the specimens ECS1 to ECS4, 3.3. Failure mode
only by about 6.6%, 6.4%, 6.1% and 5.8% leads to the buckling capacity to
be increased 1.33, 1.72, 2.15 and 1.46 times. On the other hand, results As regards deformational mode of the circular plates under uni-
of Table 2 show the most optimum location for the circumferentially stiff- form pressure, the results of this paper are schematically compared
ener which belongs to ECS3 with the highest load capacity. One may re- with that pointed out in “Yield Line Theory” for plain circular plates.
call that radial location of the stiffener in this specimen was in 91% of Jawad investigated the theoretical phase of unstiffened circular
the radius measured from the center of the plate. plates [13]. In fact, it was obtained from this reference that the
Overall, since the edge area in clamped circular plates under uni- yield-lines mode of the circular plates under normal pressure is theo-
form loading is a structural weakness, it is necessary for such zone retically radial. However, Jawad discussed that this outcome is just
to be well-strengthened. The reinforcement may be in the form of theory of these structures and it is strongly possible to obtain differ-
an annular ring plate or stiffener depending on the constructional ent failure modes and subsequently critical loads from the experi-
limitations for different structures. Considering the buckling load of ments (as detected in the current tests); thus, the experimental
the present circular plates versus the material weight increase of rein- evidences will be dominant and reliable in all cases. In the “Yield
forcement (steel consumption), it should be pointed out that if a ring Line Theory” [13], the yield-lines are extended radially over the
stiffener is provided in the edge zone of the circular plate, the buckling whole plate. One may note in passing that this mode is in fact, similar
load increases far more than the cases thickened by annular plates. In- to the experimental buckling mode of shallow conical shells (sole-fish
deed, the best current advice is to select an optimum place for the buckling mode) discussed in Refs. [11,12], in which the yield-lines
ring stiffener to have a well-strengthened circular plate. Table 3 shows meet each other at the central point of the plate. Whereas in the pres-
the ultimate load of the ECS specimens, wave number of buckling for ent series of the tests, the yield zones were formed circumferentially
the stiffener and maximum deflection of the center of the specimens. in a sine-waved shape which the deformed zone encompassed almost
It is observed that for ECS3 specimen with highest ultimate load, the 1/3 of the radial zone (measured from the edge of the plates) with
final deflection in the center is minimum. This clearly shows the influ- generally U shaped and seldom V shaped yield lines. In our sense,
ence of the stiffening on the center-point deflection of the present this discrepancy may have been caused by the experimental parame-
plate specimens. ters which, in essence, can be different from the theory such as the
It is noteworthy that out-of-plane deformations of ring stiffeners specified boundary conditions.
were detected in the ECS specimens. It is remembered that since the It is worthwhile to mention that plastic deformations lead to the ini-
ring stiffeners provide an effective radial restraint against plate buck- tial flat shape of the specimens to change into a curvy deformed shape
ling, the edge area can not be deformed rather than the central zone. (Fig. 11). Consequently, the ultimate post-buckled shape of the circular
As a consequence, rotation of the central area of the plate's wall around plates tends to appear as a dome-shaped deformational mode which is
the stiffened area in combination with the circumferential compressive mostly clear in ECP specimens. Furthermore, in addition to the distinct
a 14.0
b
35.0
12.0
30.0
10.0 25.0
Load (kPa)
Load (kPa)
8.0 20.0
Center
Center
6.0 15.0
(A+2) 13
A13
4.0 10.0 C13
(B+2) 13
(D+2)13
2.0 D13 5.0
0.0 0.0
0.0 5.0 10.0 15.0 20.0 0.0 5.0 10.0 15.0 20.0
Displacement (mm) Displacement (mm)
c 40.0
d
16.0
35.0 14.0
30.0 12.0
Load (kPa)
25.0 10.0
Load (kPa)
Center
20.0 Center 8.0
(A+2) 1
15.0 (A+2) 13 6.0
C1
C13
10.0 4.0 (D+2) 1
E13
5.0 2.0
0.0 0.0
0.0 5.0 10.0 15.0 20.0 25.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0
Displacement (mm) Displacement (mm)
e 35.0 f
25.00
30.0
20.00
25.0
Load (kPa)
Load (kPa)
20.0 15.00
15.0 Center
Center
10.00
(A+1) 1 (A+1) 1
10.0
(B+3) 1 (B+2) 1
5.00
5.0 D1 (C+4) 1
0.0 0.00
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0
Displacement (mm) Displacement (mm)
Fig. 8. Load-displacement curves obtained from the tests: a) ECC, b) ECP1, c) ECP2, d) ECS1, e) ECS3, f) ECS4. Note: e.g. (C + 3)1 means the measured spot was placed in radial dis-
tance of “C” + 3 cm (measured from the center) and circumferential line of “1”.
buckling waves, some small wrinkles are obviously seen in the defor- of Fig. 3) were obtained from the tension coupon test of the standard
mational zone of the specimens. All these circumferential waves cause steel strips. This curve was linearized, and then used as the stress–
the failure mode to approach a kind of skirt mode of deformation strain relationship for the material modeling of the FE analyses.
(Fig. 12). Yield and failure stresses, Young modulus and Poisson's ratio for
the current models were defined as 215 MPa, 310 MPa, 210 GPa
4. FE modeling and 0.3 respectively. All models were subjected to uniform pressure.
Measured geometrical imperfections resulted from the fabrication
In order to simulate the tested specimens, the well-known FE soft- process of the test specimens were implemented into the numerical
ware “ANSYS” was used to verify the experimental results. Non-linear models as each node of the FE model was simulated quite similar to
“Arc-Length-Type” method which is predominantly used in non-linear the equivalent zone of the test specimen considering depression/
structural analyses was taken into account in the present study. In this projection values of imperfection. Clamped boundary conditions
set of numerical models, material and geometrical nonlinearities were were modeled for the circular brim of the numerical models. The el-
both considered. The material properties (as seen in stress-strain curve ement type used in the current FE modeling was “Shell 181.” Time
314 K. Imani et al. / Journal of Constructional Steel Research 80 (2013) 308–316
a 14.0
Table 3
Test results of ECS specimens.
12.0 Specimen label Ultimate load (kPa) Stiffener buckling Max. deflection at the
wave number center (mm)
10.0 ECS1 15.2 12 31.92
ECS2 28.1 13 34.4
Load (kPa)
20.0
were fully perfect. For example, some non-detected geometric imper-
15.0 fections which can be developed in fabrication and installation pro-
Center cess, residual stresses, boundary nonlinearity (friction, sliding, etc.)
10.0 C1 (radial) and some other test and human related shortcomings were not possi-
(E+0.5)1 (radial) ble to model in the present numerical simulations. Thus, the discrep-
5.0 ancies between the two studies can be due to all these parameters.
(F+0.5)1 (radial)
However, acceptable consistence was obtained comparing the tests
0.0
-0.0005 0.0000 0.0005 0.0010 0.0015 0.0020
with the FE results.
Strain
3. The post-buckling potency was detected in all the present speci- reinforced specimens this interval was quite distinguishable. There-
mens and the post-buckling response of such structures was in a fore, strengthening can make the post-buckling response more stable.
stable manner. 5. Sine-wave shaped overall deformations were detected circumferentially
4. There was a brief interval between initial buckling and failure for ECC near the boundary zone of the plates.
specimen and the failure occurred quickly after buckling, whereas for 6. Yield-lines in the brim zone of the circular plates were developed
in generally U shaped and seldom V shaped mode. Plastic collapse
appeared by folding of the yield-lines at some buckled areas, and
ultimately by brim local luxating of the plate specimens.
7. For the practical cases without manufacturing limitations of using
the stiffeners, if a ring stiffener is provided in the edge zone of the
Fig. 12. Finial deformed shape of the specimen ECP2. Fig. 13. Failure pattern of the specimen ECS2.
316 K. Imani et al. / Journal of Constructional Steel Research 80 (2013) 308–316
50.00 Table 4
45.00 Experimental and numerical results.
30.00
ECC ECS2 16 20.5 28.1%
25.00 ECP2 ECS3 20 24 20%
ECP1 ECS4 13.6 16.8 23.5%
20.00 ECP1 22 26 18.2%
ECC (FE)
ECP2 (FE) ECP2 30.5 32 4.9%
15.00
ECP1 (FE)
10.00
References
5.00
[1] Timoshenko SP, Goodier JN. Theory of elasticity. 3rd ed. New York: McGraw Hill;
0.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00 1970.
[2] Timoshenko SP, Woinowsky-Krieger S. Theory of plates and shells. 2nd ed. New
Displacement of the center (mm) York: McGraw Hill; 1959.
[3] Lekhnitskii SG. Anisotropic plate. New York: Gordon and Breach; 1968.
Fig. 14. Comparative load-displacement curves of the specimens (experimental and FE). [4] Turvey GJ, Der Avanessian NGV. Full-range response of clamped ring-stiffened cir-
cular steel plates—comparison between experiment and theory. Comput Struct
1990;37(1):S-70.
[5] Lim GT, Turvey GJ. Full-range response of uniformly loaded clamped circular steel
plates: comparisons between theory and experiment. Proc Inst Civ Eng Part 2
Mar. 1985;79:211-3.
circular plate, the buckling load increases far more than the cases [6] Turvey GJ, Der Avanessian NGV. Axisymmetric elasto-plastic large deflection re-
thickened by annular plates. sponse of ring stiffened circular plates. Int J Mech Sci 1989;31(11/12):905-24.
[7] Del Coz Diaz JJ, Garcia Nieto PJ, Betegon Biempica C, Fernandez Rougeot G.
Non-linear analysis of unbolted base plates by the FEM and experimental valida-
tion. Thin Walled Struct May 2006;44(5):529-41.
[8] Del Coz Diaz JJ, Garcia Nieto PJ, Fernandez Rico M, Suarez Sierra JL. Non-linear
Acknowledgment analysis of the tubular ‘heart’ joint by FEM and experimental validation. J Constr
Steel Res August 2007;63(8):1077-90.
[9] Venstel E, Krauthammer T. Thin plates and shells. Theory, analysis and applica-
The work presented herein was carried out in collaboration with tion. CRC Press; 2001.
the Structural Research Center at Urmia University. We would like [10] Edmund S. Melerski, design analysis of beams, circular plates and cylindrical tanks
to put across our gratitude to professional technician Mr. Jafar on elastic foundations. London, UK: Taylor & Francis Group; 2006.
[11] Golzan S. Behnam, Investigation of the buckling and post buckling behavior of
Azimzadeh for his enthusiasm during the experimental program. conical and truncated conical shells under uniform external pressure. M.Sc. Eng
We also appreciate Swanson Analysis Inc. for the use of ANSYS Uni- thesis, Faculty of Civil Engineering, Urmia University, Urmia, Iran; 2006.
versity Intermediate program. Last but not the least, the authors are [12] Golzan BS, Showkati H. Buckling of thin-walled conical shells under uniform ex-
ternal pressure. Thin Walled Struct 2008;46:516-29.
indebted to B.S. Golzan from Sherbrooke University of Canada for
[13] Jawad Maan H. Theory and design of plate and shell structures. London: Chapman
doing his best regarding this work. & Hall; 1994.