Whole
Whole
Whole
By
Yu-Fei Wu
Doctor of Philosophy
June 2002
PUBLICATIONS
The following publications were written based on the work presented in this thesis.
Book chapter
Wu, Y.F., Oehlers, D.J., and Griffith, M.C. (2002). "Partial interaction analysis of
Wu, Y.F., Griffith, M.C. and Oehlers, D.J. (2002). "lmproving the strength and ductility
Oehlers, D.J., Wu, Y.F. and Griffith, M.C. (2002) "Rectangular RC columns retrofitted
Wu, Y.F., Oehlers, D.J., and Griffith, M.C. (2002) "Numerical simulations of composite
Struclures.
ll
vy'u, Y.F., oehlers, D.J., and Griffith, M.c. (2002) "Ductility of RC columns". In
preparation.
Conference paper
wu, Y.F., oehlers, D.J. and Griffith, M.c. (2001). "composite plated columns".
Departmental reports
wu, Y.F., Griffith, M.c. and oehlers, D.J. (2001). Behavior of plated RC columns.
University of Adelaide.
Wu, Y.F., Oehlers, D.J. and Griffith, M.C. (2001) "Numerical simulation of composite
plated columns", Research Report No. R172, Dept. of civil & Environmental
ill
ACKNOWLEDGEMENTS
The author wishes to express his heartfelt appreciation and gratitude to Associate
Professors M.C. Griffith and D.J. Oehlers for being his supervisors for the research that is
described herein. Their genuine interest in both the research project and my future career,
their motivation, encouragement and intellectual supervision are sincerely appreciated.
Their full support has made the past three years most productive and rewarding as well as
an enjoyable stay. The new retrofit concept studied in this work was initially proposed by
Associate Professor Oehlers which is specifically acknowledged.
The author would also like to thank the laboratory staff, Mr. Bruce Lucas, Mr.
Gregory Atkins, Mr. David Hale, Mr Jeffrey Hiorns and Mr Steven Huskinson, in the
Department of Civil & Environmental Engineering at The University of Adelaide for all
their assistance with the experimental work. The laboratory technicians played an
important role in the design and manufacture of the test rigs and test specimens. Special
acknowledgement and thanks are due to Mr. David Hale for his full support and co-
operation, excellent workmanship and intellectual input into the experimental work.
The author is also thankful to Dr. Stephen Carr and Mr Paul Carter for providing
computing assistance, the departmental administrative staff, Ms. Diane Keable, Mrs.
Bernice Golledge and Ms Josie Peluso, for their help throughout the course of his study.
This research work was made possible only under the sponsorship of a University
of Adelaide Scholarship (UAS). The experimental work was funded by a Small
Australian Research Council grant as well as the departmental financial support. All these
supports are gratefully acknowledged.
The author appreciates the assistance and friendship of his fellow research
scholars, academics, staff and all his friends who have made his stay during the entire
period of the research a most rewarding and enjoyable one.
Special thanks are extended to his wife, daughter and son. This work is made
possible only under their understanding, patience and support.
Finally, the author dedicates this thesis to his parents as a mark of respect and
love.
lv
ABSTRACT
Both steel jacketing and fibre reinforced polymer (FRP) wrapping have been shown to
effectively enhance the seismic resistance of circular reinforced concrete (RC) columns
by confining the concrete. However, jacketing and wrapping are much less effective in
efforts have been made in the literature to solve this problem, a satisfactory solution is yet
uses a completely different concept has been proposed and investigated in this work. This
new retrofit scheme makes use of composite action by bolting plates to the column
surfaces for improving the flexural strength and/or ductility of rectangular RC columns.
This new composite partial-interaction plating approach does not rely on confinement to
improve the strength/ductility but instead relies on the composite action between the plate
investigate the behaviour of the plated RC columns, from which the proposed retrofit
scheme is found to be effective in
design such a plating system based on the required inter-story drift ratio. Numerous
II
ABSTRACT v
JACKETS ..21
vl
3.3 LIMITATIONS AND ASSUMPTIONS OF THE STUDY 39
4.1 CROSS-SECTIONALANALYSIS 44
4.1.2.1 Triangularmodel 48
4.2.1 SmallDeflection........
4.2.2 LargeDeflection.......
4.4.1 Procedure 77
4.5 COLUMNS VYITH PLATES BOLTED TO TENSION AND COMPRESSION FACES .....,,,,....96
vll
1.5.1 Existence ofSolution....... .........98
11n
6.2.i.i Equiiibrium anii compati'oiiiiy lJt
vill
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS 162
162
7.1.2
7.1.3
7.1.4
7.2.2 For Axial Load Level Lower Than Critical Load t76
7.2.3 For Axial Load Level Higher Than Critical Load...... ....................... I 7 7
IX
9.2.3 Gaps..................... ..22s
I1.6
X
1 2.1.1 Monotonic Tests.......... 317
REFERENCES.............. 349
XI
LIST OF FIGURES
Fig. 1 .I Typical shear failure of columns in an earthquake (Seible et al. 1997) ...... ...........3
Fig. I .2 Typical flexural plastic hinge failure of columns (Green, 1987) ...........3
Fig. 1.3 Stress-strain relation for confined and unconfined concrete (Mander et al .1988a)
...........5
xll
Fig.4.12 Menegotto-Pinto model of steel 6l
xlll
Fig. 5.4 Response of columns with different axial loads.................. 112
Fig. 5.i3 Effect of gap for the 6mm plated column, K¡:23kN/mm................................126
Fig. 5.18 Slip distributions when maximum slip occurs at the bottom l3l
1a a
Fig. 6.i Piastic-eiastic mociei. .lJ+
Fig. 6.6 Slip distributions compared with the numerical results ..155
XIV
Fig.7 .l Plain concrete column and section
Fig.7.7 Moment-curvature chart for section with 3.68% steel and N:So/oN".... 184
Fig. 7.8 Axial force-curvature chart for section with 3.68% steel and N:So/oN. 185
Fig.7.9 Concrete stress block at a curvature of 7.5x10-5 1l/mm) for section with 3.68%
XV
Fig. 8.5 Centroid coefficient ô. 198
Fig. 8.7 Profiles when tension side yielded in tension ... 203
Fig. 9.9 Drilling of holes in RC specimen through holes in steel plate.... 221
r: la I -1-ll-rl---
rtB. ^Y.lJ rnslarrauurr ul
-fl--lr-
uulrs, LL.)
. ^41
xvl
Fig.9.2l Right hand side pin support....... 229
Fig. 10.5 Stress-strain curve of Y16 bar sample Yl6f .... ...257
XVII
Fig. 10. l0 Bolt shear test... .264
Fig. t0.18 Failure mode of Hilti HSL Ql2 bolt test............ .........269
Fig. 10.19 Hilti HSL bolt and concrete block after test............ ...270
Fig. 10.20 Uniform distributed supports to prevent concrete block failure ....................270
Fig. 11.2 Measured strains in reinforcing bars for specimen 1AMR.., 279
Fig. I 1.3 Front view at displacement of l23mm for specimen IAMR 280
Fig. I 1.4 Rear view at displacement of 123mm for specimen IAMR. 280
xvlll
Fig. I1.6 Measured strains in reinforcing bars for specimen 24MF12..... .284
Fig. I1.7 Measured strains in steel plate for specimen ZAMF|2 .285
Fig. I 1.9 Rear view after test for specimen 2AMF 12 ................, .286
Fig. 1 l.l0 Front view after test showing major cracks for specimen 2AMF12 . .286
Fig. I I .13 Measured strains in reinforcing bars for specimen 18MP6........ .291
Fig. I Ll8 Enlarged rear view at displacement of l06mm for specimen 18MP6...........293
Fig. I1.19 Front view at displacement of l06mm for specimen lBMP6... ...294
Fig. I 1.21 Measured strains in upper reinforcing bar for specimen 34CR........ .............297
Fig. I I .22 Measured strains in lower reinforcing bar for specimen 34CR........ .............297
Fig. I I .24 Rear view at -56.3mm displacement of I't cycle for specimen 34CR..........298
Fig. I 1.25 Front view at -56.3mm displacement of l't cycle for specimen 34CR.........298
Fig. I 1.28 Measured strains in upper reinforcing bar for specimen 4ACP6 302
XIX
Fig. I L29 Measured strains in lower reinforcing bar for specimen 4ACP6 302
Fig. I I .31 Measured strain in steel plate on upper external side for specimen 4ACP6 303
Fig. I1.33 Measured strain in steel plate on upper internal side for specimen 4ACP6, 303
Fig. 11.35 Measured strain in steel plate on lower external side for specimen 44CP6..304
Fig. 1 1.36 Measured strain in steel plate on lower internal side for specimen 44CP6...304
Fig. 11.37 Open up of steel plate at -82mm of 2''d cycle for specimen 44CP6..............305
Fig. I I .39 Rear view of plastic hinge zone after testing for specimen 44CP6...............306
Fig. 1 1.40 Front view of plastic hinge zone after testing for specimen 44CP6..............306
Fig. 1 1.42 Measured strain in upper reinforcìng bar for specimen 28CP6G..................31 I
Fig. I 1.43 Measured strain in lower reinforcing bar for specimen 28CP6G..................31 I
D:^
r rÉ. Ir 1 A /l l\Ã^^-,,-^l
r.TT -+-^i- i-
lYtvoùutvu ùtlcrrr uPù f^-
-+i--,,^-
rrr Jtrrl rur .^^^i*^^ 'lÞl1DÃ.(:
ùPvvrrrrvrr ¿uvl uu 211
J IL
F-ig. 1 1.45 Measured strain of' steel plate on upper external side fbr specimen 2tsCP6G
312
Fig. I 1.46 Measured strain of steel plate on upper internal side for specimen 2BCP6G3I2
Fig. I 1.47 Measured strain of steel plate on lower external side for specimen 2BCP6G
313
Fig. I 1.48 Error in measure point C due to plate lifting up.............. .............3 l3
Fig. I I .49 Last cycle of test for specimen 2BCP6G after changing instrumentation .....314
XX
Fig. I 1.50 Front view after testing for specimen 2BCP6G . 314
Fig. I l.5l Rear view after testing for specimen 2BCP6G .. 315
Fig. 12.7 Comparison of test result with theory for specimen lAMR 330
Fig. 12.8 Comparison of test result with theory for specimen2AMF12......... 331
Fig. 12.9 Comparison of test result with theory for specimen IBMP6 )JZ
Fig. l2.l I Comparison of test result with theory for specimen 3ACR .. 334
Fig. 12.12 Comparison of test result with theory for specimen 4ACP6. 335
Fig. 12.14 Numerical simulation with un-symmetrical base gaps on two sides .............337
XXI
Fig. 4.4 Young's Modulus test of CYL04... ..376
Fig. 8.1 Front view at displacement of 7mm for specimen IAMR 381
Fig. 8.3 Top view at displacement of 28mm f'or specimen 14MR............ 382
Fig. 8.4 Rear view at displacement of 4lmm for specimen lAMR 382
Fig. 8.5 Front view at displacement of 57mm for specimen 1AMR......... 383
Fig. 8.6 Rear view at displacement of 89mm for specimen IAMR 383
Fig. 8.7 Rear view at displacement of lOmm for specimen 2AMF 12 ....... 384
Fig. 8.8 Rear view at displacement of l6mm for specimen 2AMF12....... 384
Fig. 8.9 Rear view at displacement of 42mm for specimen 2AMF 12 ....... 38s
XXII
Fig. B.l0 Rear view at displacement of 52mm for specimen2ÃMFl2 ........385
Fig. 8.1 I Rear view at displacement of 82mm for specimen 2AMF12 ........386
Fig. 8.13 Rear view at displacement of 6.5mm for specimen IBMP6 .........387
Fig. B.l4 Rear view at displacement of l9mm for specimen IBMP6 ..........387
Fig. B.l7 Rear view at displacement of 93mm for specimen IBMP6 ..........389
Fig. B.20 Rear view at+24.5mm displacement of 2nd cycle for specimen 34CR..........390
Fig. B.2l Front view at -25.8mm displacement of 2nd cycle for specimen 3ACR.........391
Fig.8.22 Rear view at -39mm displacement of I't cycle for specimen 3ACR ..............391
Fig.B.23 Rear view at *39mm displacement of 2"d cycle for specimen 34CR........ .....392
Fig. 8.24 Rear view at +26mm displacement of l't cycle for specimen 4ACP6 ............392
Fig. 8.25 Rear view at -26mm displacement of l't cycle for specimen 44CP6...... .......393
Fig.8.26 Rear view at r26mm displacement of 2"d cycle for specimen 4ACP6 ...........393
Fig.B.27 Rear view at -26mm displacement of 2nd cycle for specimen 4ACP6 ............3g4
Fig. B.28 Rear view at t39mm displacement of I't cycle for specimen 4ACP6 ............394
Fig.B.29 Rear view at -39mm displacement of 1't cycle for specimen 4ACP6.............395
Fig. 8.30 Rear view at +39mm displacement of 2nd cycle for specimen 4ACP6 ...........395
Fig. 8.31 Rear view at -39mm displacement of 2nd cycle for specimen 4ACP6 ............396
Fig.B.32 Rear view at-r52mm displacement of I't cycle for specimen 4ACP6 ............396
XXIII
Fig. 8.33 Rear view at -52mm displacement of l " cycle for specimen 4ACP6 .............397
Fig. 8.34 Rear view at+52mm displacement of 2nd cycle for specimen 4ACP6 ...........397
Fig. 8.35 Rear view at -S2mmdisplacement of 2nd cycle for specimen 4ACP6 ....,.......398
Fig. 8.36 Rear view at -65mm displacement of I't cycle for specimen 44CP6.............398
Fig. 8.37 Rear view at +65mm displacement of 2nd cycle for specimen 4ACP6 ...........399
Fig. 8.38 Rear view at -65mm displacement of 2nd cycle for specimen 4ACP6 ............399
Fig. 8.39 Rear view at *78mm displacement of I't cycle for specimen 4ACP6 ............400
Fig. 8.40 Rear view at -78mm displacement of I't cycle for specimen 4ACP6.............400
Fig. B.4l Rear view at +78mm displacement of 2nd cycle for specimen 4ACP6...........401
Fig.B.42 Rear view at -78mm displacement of 2nd cycle for specimen 44CP6....,.......401
Fig. 8.43 Front view after 2 cycles of t78mm displacement for specimen 4ACP6 .......402
Fig. 8.44 Rear view at *91mm displacement of I't cycle for specimen 4ACP6 ............402
Fig. 8.45 Rear view at -83mm displacement of l" cycle for specimen 4ACP6 .............403
Fig. 8.46 Rear view at +91mm displacement of 2nd cycle for specimen 4ACP6 ...........403
Fig. 8.47 Rear view at -82mm displacement of 2nd cycle for specimen 4ACP6 ............404
Fig. 8.48 Front view after 2 cycies oi-82mm to -|9imm for specimen 4ACPó... .........4A4
Fig. 8.49 Rear view at -26mm displacement of 1't cycle for specimen 28CP6G...........405
Fig. 8.50 Rear view at-r26mm displacement of 2nd cycle for specimen 28CP6G.........405
Fig. B.5l Rear view at -2îmmdisplacement of 2nd cycle for specimen 2BCP6G... .......406
Fig. 8.52 Rear view at l39mm displacement of l't cycle for specimen 28CP6G.... ......406
Fig. 8.53 Rear view at -39mm displacement of l " cycle for specimen 2BCP6G. ..........407
Fig. 8.54 Rear view at *39mm displacement of 2"d cycle for specimen 2BCP6G... ......407
Fig. 8.55 Rear view at -39mm displacementof 2"d cycle for specimen 28CP6G..........408
XXIV
Fig. 8.56 Rear view aI-r52mm displacement of l't cycle for specimen 28CP6G..........408
Fig. 8.57 Rear view af -52mm displacement of I't cycle for specimen 28CP6G.... .......409
Fig. 8.59 Rear view at -52mm displacement of 2nd cycle for specimen 28CP6G..........410
Fig. 8.60 Rear view at +65mm displacement of l't cycle for specimen 28CP6G..........410
Fig. 8.6l Rear view at -65mm displacement of 1't cycle for specimen 28CP6G.... .......411
Fig. 8.62 Rear view at *65mm displacement of 2nd cycle for specimen 28CP6G.... .....411
Fig. 8.63 Rear view at -65mm displacement of 2nd cycle for specimen 28CP6G.... ......412
Fig. 8.64 Rear view at l78mm displacement of l't cycle for specimen 2BCP6G. .........412
Fig. 8.65 Rear view at -78mm displacement of I't cycle for specimen 28CP6G.... .......413
Fig. 8.66 Rear view at t78mm displacement of 2nd cycle for specimen 28CP6G.... .....413
Fig. 8.67 Rear view at -78mm displacement of 2nd cycle for specimen 28CP6G.... ......414
Fig. 8.68 Rear view at +9lmm displacement for specimen 2BCP6G .........414
Fig. 8.69 Rear view at +l54mm displacement for specimen 2BCP6G ........415
Fig. 8.70 Front view at +l54mm displacement for specimen 2BCP6G ......415
Fig. B.7l Rear view at - I 00mm displacement for specimen 2BCP6G ........416
XXV
LIST OF TABLES
XXVI
NOTATION
EI
a1 , passive slip strain coefficient
(L tL
a2 : \
active sliP strain coefficient
ffi'
a-J , axial flexibility of element I
(EA),
XXVII
_ (h, 'r / EA
\"r+ h,). --- , unit slip curvature
a4
EI
os : a,
rJ-, slip resistance ability coefhcient
ao
C : constant of integration
centre ofstirrups
E : modulus of elasticity
.f,o
Ec ) slope of softening part of stress-strain curve
t", - t,
XXVIII
E, : modulus of elasticity of reloading branch of concrete
E
" : modulus of elasticity of steel
EA :definedu, !
EA (E.A), (E.A),
F' : lateral force at top of column considering P-A effect by simplified method
xxlx
4 : longitudinal shear force on shear connectors in shear span of length x from top
Fn : longitudinal shear force on shear connectors in shear span of length x from top
F,n, : Fl + F,,* F,n, total longitudinal shear force on shear connectors in shear span of
F,n : longitudinal shear force on shear connectors in shear span of length x from top
f ," : concrete stress at common return point from reloading branch to envelope curve
in Mander's model
.f ,,,
: stress at onset of reloading branch in Mander's model
XXX
-f, : tensile strength of concrete under monotonic loading in Mander's model
M : moment of section
XXXI
M 2 : moment due to element 2
Mc(, : moment of cross-section when compressive strain at extreme fiber equals to e,o
Mclt : moment of cross-section when compressive strain at extreme fiber equals to t,,,
N
ot,
: axial f-orce on plate
n : number of bolts
XXXII
p : intensity of distributed load
Qt
: shear force on element I
Q,
: shear force on element 2
R, : slip radius
r : radius of curvature:' or / -
E"
E" - Er",
^ç
: longitudinal slip of bolt; or coordinate along length; or longitudinal centre to
,s
: clear longitudinal spacing between stirrups
xxxiii
v : lateral deflection of column
I
v Jitll : lateral deflection induced by lateral force ,Fwith full interaction
w, : longitudinal gap width between side of plate hole and side of bolt
y :ycoordinate;functions
d : L'Ja'
Q,1
: experimentally determined material constant for Menegotto-Pinto model
section
XXXIV
Lr, : axial shortening of element 2 at centroid
L" : lateral displacement at top of column due to elastic deformation above plastic
hinge
P
: lateral displacement at top of column due to plastic hinge rotation only
^
: lateral displacement at first yield of tensile reinforcement
^v
u
: ultimate lateral displacement when lateral force equals to 80% of lateral force at
^
yield point
c
= strain
o
: (when used in Menegotto-Pinto model) strain at point where two asymptotes of
co
: strain at maximum stress ;{., of unconfined concrete
XXXV
cc
: strain at maximum stress /1, of confined concrete
e
cll
: ultimate compressive strain of concrete
: (when used in Menegotto-Pinto model) strain at point where last strain reversal
tr" : concrete strain at common return point from reloading branch to envelope curve
in Mander's model
a,
.l
: steel strain
XXXVI
: concrete strain at start of unloading branch in Mander's model
K : curvature ofsection
K co : curvature of section when extreme compressive fiber reaches yield strain e",
6 = stress
oc : stress ofconcrete
(when used in Menegotto-Pinto model) stress at point where last strain reversal
JI
b , normalized coordinate x; variable of integration; plastic strain of last
L
XXXVIII
Symbols:
CHAPTER 1 INTRODUCTION
1.1 BACKGROUND
Many existing reinforced concrete structures that were designed and constructed prior to
the adoption of modern earthquake resistance design codes are vulnerable to the attack of
an earthquake above the moderate scale. It has been repeatedly highlighted overseas as
buildings and elevated freeway/bridge structures that collapsed were designed and
and Mahin l99l;Blong 1993; Pham and Griffith 1995; Bracci et al. 1995a & b; Lynn et
al. 1996; Seible et al. 1997; Park 2001). Therefore, "The improvement of the seismic
before qn event is one of the most essential strategies to mitigate disaster." (CEB report
tee7)
Particularly vulnerable are reinforced concrete columns resulting from the then
typically used nominal and inadequately detailed transverse reinforcement (Park 2001).
Bridge columns in the U.S. that were designed before the l97l San Fernando earthquake
typically involved the use of No.4 (l2.7mm diameter) transverse peripheral hoops placed
at 12 in. (305mm) centers for both circular and rectangular columns, regardless of the
column section dimension (Chai et al. 1991). These hoops were often closed by lap
CHAPTER I INTRODUCTION
splices in the cover concrete instead of being lap-welded or anchored by bending back
into the core concrete. As a result, the ultimate curvature that can be developed within the
potential plastic hinge region is limited by the strain at which the cover concrete begins to
spall, which is typically around 5olo strain. The hoop steel unravels when the longitudinal
strain is higher than 5o/o, resulting in the loss of the small confinement and shear strength.
Although the diameter and spacing of the hoops varies somewhat for different structures
typical of pre-1971 building and other structures all over the world. Structures with such
columns cannot satisfy the high ductility demand and will fail prematurely during a
(1996) on eight full-scale RC columns with typical pre-1971 detailing, which was based
orì a survcy of builclings built from l919-1970, confirmed these observations. Research
outside the US also indicated that most of the existing ground floor columns were
expected to fail under a strong earthquake due to inadequate flexural ductility and brittle
transverse reinforcement under seismic load (Seible et al. 1997).The first failure mode is
the shear failure as shown in Fig.l.l, where inclined cracking, cover concrete spalling,
and disintegration of the concrete core develop sequentially in an earthquake. The second
failure mode is the flexural plastic hinge failure due to large displacements incurred
2
CHAPTER I INTRODUCTION
where, subsequent to flexural cracking, cover concrete crushing and spalling, buckling of
plastic hinge deterioration. Fig.l.2 shows a typical failure case of the flexural failure
mode.
Fig. 1 .I Typical shear failure of columns in an earthquake (Seible eI" al. 1997)
3-
CHAPTER I INTRODUCTION
Thirdly, many existing columns have lap splices of the longitudinal reinforcement
at the bottom where both the maximum flexural moment and potential plastic hinge
occur. The lap length of the longitudinal bars in pre-1971 structures was typically
designed as a compressive lap of usually 20 times the longitudinal bar diameter, which
was insufficient for development of the tensile yield strength under load reversal due to
seismic load. Lap splice debonding occurs once vertical cracks develop in the cover
concrete and progresses with increased transverse dilation of the column and cover
concrete spalling, associated with rapid degradation of strength and deformation capacity.
Even lap-splices in a plastic hinge region that satisfl modern design code requirements
invariably break down under cyclic inelastic action in a severe seismic event (Priestley &
Seible, 1995).
irrvulvcs tlrc usc of sLccl ur FRP (Fiber Reinfolced Polynrer') jackets, which wraps a layer'
of steel or FRP plate on the external face of the column to provide additional external
hoop action in the transverse direction. This technique is effective in enhancing the shear
with the jacket. The deformation capacity of plastic hinges can also be increased
conhnement to a column's external face can prevent cover concrete spalling and buckling
of longitudinal reinforcement, and most importantly, enhance the concrete strength and
deformation capacity, as shown in Fig.l.3, hence improve the ductility performance. The
enable displacement ductility factors greater than 2.5 to 3.5 to develop (Priestley and
-4-
CHAPTER 1 INTRODUCTION
Seible, 1995), and typical ultimate compressive strain in the extreme fìber as high as 0.01
jacketing. For the third failure mode, confinement in the form of jacketing can again be
Confined Fìrs f
\U concrete hoop
f
'cc
"')
u)
N
Lo
s f.o
an
conf
(4
s,
a-
E
o
(J
et cc
I
Compressive Slrotn , €c
Fig. I .3 Stress-strain relation for confined and unconfined concrete (Mander et al. 1988a)
Theoretical and experimental studies have already proved that external jacketing
is highly effective in preventing existing columns from failure for all the above three
failure modes in the,case of circular columns. Therefore, this kind of retrofitting work has
already been widely used in engineering (Chai et al. 199l&.1994; Priestley et al.
1994b&c).
Transportation (Caltrans) in the late 1980's, bridge columns with pre-1971 details for
1100 state-owned bridges and 1500 city- or county- owned bridges, to which a prime
concern was the lack of adequate confinement, were scheduled for retrofit for the
5
CHAPTER I INTRODUCTION
estimated cost between US$420 million and $550 million (Chai et al. 1994). Steel
jacketing was used extensively for the retrofitting of these columns. The effectiveness of
steel jacketingwas evident in the field in fhe 1994 Northridge Earthquake. An estimated
sixty retrofitted bridges, many of which had been steel jacketed, were located in the
region of intense ground shaking with peak ground acceleration éxceeding 0.259. None
of the steel-jacketed columns were reported to have sustained any significant damage. In
contrast, two adjacent un-retrofitted bridge structures collapsed due to column failures
building was also reported in Los Angeles (Elhassan and Hart, 1995). The building was a
seven-story hotel with 2-story basement located in the City of Los Angeles. During the
Magnitude 7.5 Landers Earthquake of 28 June 1992, centered 175 km away from the
builtling, many of Lhe ground floor columns suflerecl significant diagonal cra<;ks. In the
rehabilitation work, the ground floor story columns were retrofitted with fiberglass/epoxy
composite jackets. This seismic retrofit scheme was completed a few weeks before the
occurrence of the Northridge Earthquake of 1994, and the building suffered no damage
However, jacketing with steel or FRP plate is not always as effective for
square/rectangular columns as it is for circular columns. While it is still very effective in
enhancing the shear capacity (Priestley et al. 1994b&.c), the square/rectangular jackets
provide less confinement to the concrete core inside compared to that for circular
columns and so are less effective for the second and third failure modes. The reason is
simple: the lateral dilation of concrete under compression must increase the perimeter
6
CHAPTER I INTRODUCTION
length of a circular jacket/wrap, as the circular shape encloses the maximum cross-
sectional area for a given perimeter length. The change of perimeter length mobilizes the
shaped jacket/wrap not to increase in spite of lateral dilation of the concrete, as a small
change in the shape of the cross-section from the original rectangular shape towards a
circular shape will provide sufficient room for the dilation to occur without hoop action
being mobilized. In practice, even if the perimeter length of the rectangular section does
increase due to the dilation effects, the confinement stresses are concentrated at the
corners of the cross-section rather than being uniformly distributed around the cross-
rectangular thin steel jacket would not be so elfective, due to the sides bowing out when
Rcctanguler J¡ckct
[n¡dequa¡c
Confirxmrnt
The poor performance of rectangular jackets in confining the concrete core has
already been experimentally verified (Chai et al. 1990; Mirmiran et al. 1998; Pessiki et al.
2001). However, there were also test results showing that effective confinement could
7
CHAPTER I INTRODUCTION
still be achievable with rectangular jacketing (Wang and Restrepo 2001; Parvin and
Wang 2001; Chaallal and Shahawy 2000). Nevertheless, the general understanding is that
jacketing for rectangular columns is more effective with smaller columns and thicker
jackets and less effective for larger columns. The corner radius in a cross-section plays an
important role in confining the concrete core (Mirmiran et al. 1998). It was also
concluded that significantly more research work is necessary to investigate the effect of
cross-section shape and size (Rochette and Labossière 2000; Pessiki et al. 2001).
reported in the literature. One technique was to enhance the out-of-plan flexural stiffness
of the jacket by using additional stiffeners in the cross-section. However, test results
showed that the improvement was not satisfactory (Chai et al. 1990). Corrugated steel
jackets were also investigated to improve the out of plan flexural strength of the jacket by
undulating the steel plate (Tomii 1993). The test results by Chobarah et al. (1996;1997)
shear strength of joints, whereas improved confinement to the concrete was also
techniques, it seems that no further research effort in this direction was made and no
shaped jacket. Research works has shown that this technique is highly effective, and is
8
CHAPTER I INTRODUCTION
always desirable or practical in engineering. The major problem with this technique is
that it changes the size and shape of the original columns significantly. A
square/rectangular shape may still be the most suitable with respect to having the
minimum disturbance to the original design both architecturally and structurally. A good
example was the Bixby Creek Bridge in California, where due to its historic nature, the
overall structure and member geometry had to be maintained for aesthetic purposes,
thereby disallowing the use of circular or elliptical steel shelljacketing (Karbhari 2000).
square/rectangular steel jacket after the 1995 Kobe earthquakes in Japan (Uehan and
Meguro, 2000), regardless of the lack of (or immaturity of) design guidelines. Therefore,
order to reduce/prevent loss of human life and properties from future earthquake damage.
detailed description of this new retrofitting system and the research framework is given in
Chapter3.
In general, mathematical, numerical and experimental tools are used to study the
behavior of the new retrofitting system in this work. As no existing software is available
for this specific complicated structural system, a numerical model and the corresponding
computer program has been developed in this study to calculate the response of the
9
CHAPTER I INTRODUCTION
proposed system, which are reported in Chapter 4 and 5. The mathematical study in
structural system. Experimental works have also been conducted to veriff the results and
conclusions drawn from the theoretical studies that are reported in Chapter 9-ll.
Furthermore, a methodology for the design of the composite plating system is developed
-10-
CHAPTER 2 LITERATURE REVIEV/
A general review of the literature for the seismic retrofit of RC columns by jacketing is
undertaken in this chapter to provide an overview of the structural retrofit history and
technologies. More detailed reviews and discussions on related subjects are given
The l97l San Fernando earthquake caused the complete collapse of five, and
serious damage to another 37, modern and apparently well-designed concrete freeway
bridges (Priestley & Park, 1987). The lesson of this earthquake raised serious concerns
over the seismic design philosophy embodied in the then existing design codes. An
upsurge in research interest into the seismic behavior of concrete bridge structures has
Zealand National Roads Board. As a pioneer work, fundamental problems such as the
ductility of concrete columns were investigated (Priestley & Park, 19S7). Particular
emphasis and significant achievement was made on quantifying the influence and
columns in increasing ductility (Mander et al. 1988a & 1988b; Watson and Park 1994;
Watson et al. 1994). ln the mean time, research work on retrofitting of RC structures
- ll -
CHAPTER 2 LITERATURE REVIEV/
were also conducted (Rodriguez and Park 1994; Hakuto et al. 1995; Liu and Park 2001;
In the U.S., research emphasis was primarily directed towards the development of
retrofit program, which was based on ongoing experimental and analy'tical research being
conducted at several institutes. A major research program was started in 1987 at the
University of California at San Diego to study the various problems related to seismic
Substantial experimental and theoretical research on column jacketing with steel plates
and other advanced composite materials has shown that circular or elliptical shaped steel
jacketing was highly effective in enhancing the shear and flexural performance of
existing substandard RC columns.
In the University of Texas at Austin, extensive research work has been conducted
J.O. Jirsa. Research focused on the retrofit of frame joints (Alcocer and Jirsa 1993,
Alcocer 1993) and inadequate lap splice of reinforcement in RC columns (Valluvan et al.
since the 1970's on concrete filled tubular (CFT) columns which were closely related to
and effectiveness of confinement to the core concrete by the external steel tubes (ackets),
-12-
CHAPTER 2 LITERATURE REVIEW
especially for square tubes, by the research group in Kyushu University headed by
In Australia, little work has been reported in the literature on the subject of
have been made in a related subject - design of new concrete filled tubular columns
(Bridge 1976 Bradford and Gilbert 1990&.1992; Rangan 1991; Bradford l99l; Bridge
and Webb 1992; Rangan and Joyce 1992;Bradford and Nguyen 1994; Bridge etal.1997;
Kilpatrick and Rangan 1997&,1999; Bradford et al. 1999; O'Shea and Bridge
1999&2000; Uy 2000&200 I ).
column. These traditional methods usually involve the application of an additional layer
of concrete, steel plate or other material onto the external face of an existing RC column
(Frangou et al. 1995; Ramirez 1996; Dritsos 1997). The retrofitting methods can be
I . Concrete jacketing
This retrofithepair method applies an additional layer of concrete together with additional
longitudinal and transverse reinforcement onto the surface of the existing columns, as
shown in Fig.2.l. Concrete jacketing was widely used in Mexico City after the 1985
earthquake. It was the most popular jacketing method all over the world for strengthening
of existing RC building columns. However, this retrofit method was found to be very
- t3 -
CHAPTER 2 LITERATURE REVIEV/
labor intensive (Park 2000). In addition, it significantly increases the size as well as
stiffness of the existing columns.
{ jackrt
benddown
b¡¡r¡¡
new reinforcement
Ä A
existing colr¡.nrn
2. Steeljacketing
Steel jacketing is currently the most common retrofit technique (Tamai et al. 2000).
There are many kinds of steel jacketing. It usually involves the use of steel plates, steel
strips or steel bars in the transverse direction, which can be broken down further into two
plate around the external face of the column. In construction, two half-circle shells (for
circular columns) or two L-shaped plates (for square/rectangular columns) are placed
around the column and then site welded to provide a continuous jacket. The small gap
between the steel plate and column face is normally pressure grouted with non-shrink
grout subsequently to ensure a firm contact. Another popular method of steel jacketing
involves the use of steel angles placed at the corners of the column. Steel plates are then
-14-
CHAPTER 2 LITERATURE REVIEV/
welded to the angles to form a jacket. Gluing steel plates to the face of the column using
Steel caging consists of longitudinal angle sections at each corner of the column
and transverse steel strips or bars. Transverse steel strips are welded onto the angle to
create collars around the concrete column, as shown in Fig.2.2. The tie strips can be
laterally stressed to ensure a tight fit by special wrenches prior to welding or pre-heated
to about 200-400 degree before welding and left to shrink in cooling after welding. The
spaces between the cage and column can be filled with cement or epoxy mortar. A cover
of concrete or shotcrete reinforced with light welded fabric is usually also applied for
corrosion or fire protection. Alternatively, the longitudinal steel angle may be omitted
and thin steel strips or wires can be wrapped in a continuous spiral or in discontinuous
rings around the column. A new technique, which involves pre-tensioning of steel strips
around the column using standard strapping machines found in the packing industry and
subsequently securing them in place by metal clips, has recently been developed for
Detåil A
50 mm
Dêt8¡IA
existing column
dimensions, minimal skilled labor required for application and speed of installation. The
- 15 -
CHAPTER 2 LITERATURE REVIEV/
3. Compositejacket
Advanced composite materials such as fiberglass and carbon fiber jacket/wrapping have
been investigated to replace steel jackets in recent years (Fardis and Khalili l98l;
Katsumata et al. 1988; Saadatmanesh et al. 1994&.1996; Karbhari and Eckel 1994;
Priestley and Seible 1995; Nanni and Bradford 1995; Seible et al. 1997; Xiao and Ma
1995811997; Mirmiran and Shahawy 1997; Hanna and Jones 1997; Xiao et al. 19991'
Saafi et al. 1999; Teng et al.20001. Liu et aI.2000; Pantelides et a|.2000; Machida et al.
2000; Green 2000; Bonacci 2000; Banthia 2000; Karbhari 2000&2001:. Fam and Rizkalla
2000; Thériault and Neale 2000; Yao et al. 2001; Lau and Zhou200l; Pessiki et al. 2001;
Parvin and Wang 2001). As compared to concrete and steel jacketing, composite
jacketing has advantages in speed and ease of installation, reduced maintenance, high
strength and better fatigue performance, as well as light weight and superior durability.
The retrofit of existing structures using fiber reinforced polymer (FRP) has been
adopted for more than a decade in Japan (Arockiasamy 2000) with its first application to
column wrapping in the mid 1980s. There have been many reports on retrofitting of RC
columns using FRP jacketing since then. An increased interest in the research on
composite jacketing has been found in the literature in recent years. It was found that
widely accepted procedure of steel jacketing, particularly where access was difficult,
since in many cases, hand lay up procedures were used (Priestley and Seible 1995;
Hwang and Wehnes 1997). Successful applications in engineering were also reported in
-16-
CHAPTER 2 LITERATURE REVIEW
the retrofit of bridge piers (Gergely et al. 1998) and building columns (Elhassan and Hart
yet to come mainly due to the high cost of the material (Dritsos 1997) and unfamiliarity
of engineers with composites and the current lack of design guidelines (Karbhari and Gao
1997).
following sections.
columns and frames by concrete jacketing were conducted in the Middle East Technical
University, Turkey as early as in 1986 (Ersoy et al. 1993). Square RC columns with
concrete jacketing were tested under uniaxial load, axial load combined with monotonic
bending, and axial load combined with reversed cyclic bending. It was found that for
axially loaded columns the repair and strengthening of RC columns were effective when
the jacketing was introduced after unloading. Columns strengthened under load also
behaved well. However, repair under load was not very successful. For transverse load
tests, the repair and strengthening jackets behaved well both under monotonic and
the University of Canterbury (Rodriguez and Park 1994) showed that existing columns
-17-
CHAPTER 2 LITERATURE REVIEV/
(before damage) or repaired (after damage) with this technique, which converted the
original low ductility and fast degradation performance under cyclic load to a ductile
manner with higher strength and much reduced strength degradation. This technique was
structure while work was carried out and the column size was increased significantly.
structure at one-third scale was tested by Rodriguez and Santiago (1998). Test results
showed the high flexibility and low measured displacement ductility capacity of the
structural system.
JACKETS
The first phase of a Caltrans funded project, conducted at the University of California at
San Diego, investigated the enhancement of the inadequate and undependable flexural
strength and ductility capacity of tall flexure dominated bridge columns by encasing the
plastic hinge region of the columns with steel jackets. Chai et al. (1991) conducted
flexural tests on six 40Yo scale model (diameter D: 610mm, 3.657m high) bridge
columns. The test results indicated that retrofit of pre-1971 columns with cylindrical4.T6
mm thick ASTM A36 (fy:248MPa) hot-rolled steel jackets enabled the column to
5%. Displacement ductility in the range of 4-6 might be anticipated under extreme
seismic response (Priestley et al. 1995). The response of the retrofitted column thus
- t8 -
CHAPTER 2 LITERATURE REVIEV\/
indicated very satisfactory perforrnance. The jacket was only provided in the potential
plastic hinge region at the bottom of the column with a length of 12l9mm (2xD). An
axial load of 1779KN, corresponding to l7o/o of the design column crush load or
0.17 As .
f ,o, was applied to the test column. Bond failures that might otherwise develop
reinforcement were also prevented by steel jacketing. As compared to the complete loss
of cover concrete in the lap-splice region and bond failure of longitudinal reinforcement
of the un-retrofitted reference column, failure of the test column was eventually by low-
a strain lower than that obtained under monotonic load when subjected to large inelastic
displacement cycles.
Figure 2.3 shows the typical hysteretic response of the retrofitted and un-
+i" c.
retrofitted reference columns. It can be seen that bothductility and the energy dissipation
capacity were significantly increased in the retrofitted columns. Design guidelines on the
jacketing details to satisfy the strength and ductility requirements were also developed
Taylor and Stone (1994) theoretically studied the required thickness of circular
steel jackets based on time history inelastic analysis of bridge columns and a cumulative
damage index concept. Their study showed that even a very thin steel jacket (2mm jacket
maintain the column within acceptable damage limits for important structures.
-19-
CHAPTER 2 LITERATURE REVIEW
After the study of flexure dominated bridge columns, squat and shear dominated
spaced peripheral hoops. The experimental work showed that properly designed steel
jackets on squat columns subjected to double bending could conveft brittle shear failures
1997). Abundant research work has also been reported in the literature on FRP jacketing
of circular columns (Toutanji 1999; Demers and Neale 1999; Thériault and Neale 2000).
indisputable.
-20 -
CHAPTER 2 LITERATURE REVIEV/
COMPOSITE JACKETS
Rectangular shaped jackets are as effective as circular shaped jackets in enhancing the
shear capacity of RC columns (Priestley et al. 1994b&.c; Aboutaha et al. 1999). Although
rectangular jackets provide full effectiveness for shear retrofit, only limited enhancement
providing confinement, a conclusion that has been well established in the literature.
Tomii et al. (1987 as referred"by Priestley et al. 1994b) investigated the flexural
the jacket observed by Sakino and Ishibashi (1985), the jacket was deliberately debonded
from the existing column ensuring that the jacket acted only as hoop reinforcement rather
than also participating in flexure. Satisfactory results were obtained for circular columns.
However it was found that a degradation of the response was inevitable for rectangular
columns conhned by rectangular jackets, even when very thick jackets were used. This
was considered to be due to the inadequate confinement of the concrete and compression
reinforcement in the flexural plastic hinge region rather than due to inadequate shear
strengh.
Three kinds of jacket have been tested for rectangular columns by Chai et al.
(1990). The test columns were a cantilever type fixed at the bottom and free at top with a
cross section of 19.25" by 28.75" (489mm by 730mm). The reinforcement detail was
typical of the mid 1960s construction and lap spliced at the bottom. One jacket was an
elliptical steeljacket made of 3/16 inch (4.8mm) thick steel plates. In this case, the space
-21 -
CHAPTER 2 LITERATURE REVIEW
between the rectangular column and elliptical jacket was filled with normal weight
concrete. The second was 3/16 inch (4.8mm) thick rectangular steel jacket with
orthogonal stiffeners. Thethird was a grouted steel collar made from channel section. All
steel jackets were 48 inches (1219mm) in height and terminated slightly short of the
adjoining footings. Specimens were tested under reversed quasi-static cyclic load at the
top and a compressive axial load equivalent to 0.1As.f". The elliptical retrofitted
column behaved well up to a 3.75Yo drift ratio. The steel channel retrofitted column
showed an acceptable response up to2.75%o drift ratio. However the stiffened rectangular
steel jacket showed rather poor response compared to the other two jackets. This poor
response was attributed to the inability of the jacket to provide adequate confinement due
Mirmiran et al. (1998 and 2000) systematically studied the square shape effect of
shown to be generally less effective in confining concrete than circular ones. It was found
that the confinement effect by a square section can be quantified by a shape factor of
2r/D,whete r is the conrer radius of the square jacket and D is the depth of the concretc
cross-section. Therefore, for a square jacket with a sharp corner (r:0), the confinement
effect is zero or the jacket does not provide any confinement to the concrete inside. When
the shape factor drops below 0.3 (in this case, the_curved corner surface area is less than
34Yo of the flat surface area), a sharp decrease of hoop stress occurs. The same
1a
CHAPTER 2 LITERATURE REVIEW
ç)-cu9 "
observation was also made by others (Frangounl995; Rochette and Labossière
1996e2000).
Thériault and Neale (2000) concluded that the confinement effectiveness by FRP
wrap is far less for rectangular columns than it is for circular columns. The gain in
strength provided by FRP rectangular jackets is therefore very small compared to that
Pessiki et al. (2001) conducted full scale concentrated load tests on FRP jacketed
columns including four 457mm square by l830mm tall RC columns. The corners of the
columns were rounded to a radius of 38mm. The test results showed that the confinement
effect from the FRP jacket was significantly lower for the square columns compared to
has been reported in the literature (Xiao and Ma 1995; Katsumataet al. 1988).
The experimental work by Seible et al. (1997) found that for columns with side
theoretical thickness derived with their design guidelines for an equivalent circular
column diameter performed well up to the design target ductìlity levels. However, it was
also suggested that extrapolations beyond the tested side aspect ratios and side
-z)-
CHAPTER 2 LITERATURE REVIEW
It was observed by Demers and Neale (1999) that FRP wrapped square speclmens
showed increases in ductility comparable with those obtained for circular specimens.
However, the maximum strength levels of these square specimens showed very little
jacketing was reported by Parvin and V/ang (2001); Wang and Restrepo (2001) and
Due to the large scatter in experimental results and very limited data available in
the literature (Thériault and Neale 2000), it was concluded that significantly more
research is necessary to investigate the effect of cross-sectional shape and size (Rochette
and Labossière 2000; Pessiki et al.200l). It was also suggested that at the present time a
San Diego established that elliptical jackets could provide excellent enhancement of
was provided by hoop membrane action in the jacket in the same way as in circular
jackets. It was therefore recommended that oval shaped jackets or bolsters be used for the
However, it is easy to imagine that the oval shaped jacket will lose it effectiveness
when the one side of the rectangular cross-section is much longer than the other side.
Experimental work by Tan and Yip (1999) showed that the effectiveness of confinement
-24-
CHAPTER 2 LITERATURE REVÌEV/
by elliptical hoops diminishes as the side aspect ratio, that is the ratio of the major axis to
the minor axis of the cross-section, increases and becomes insignificant when the aspect
Marsh (1992) investigated the use of pretensioned external steel hoops for
concluded that the steel hoop in the splice region dramatically improved the lateral
the 1986 Kalamata Earthquake. Most of the jacketing used steel caging. Experimental
studies of steel caging was conducted in the University of Patras, Greece on ten
l50mmx150mmx750mm tall columns under monotonic axial load (Dritsos, 1997). Test
results showed that the method was successful, provided that pretensioning of the
transverse ties in the external cage is applied. When the method was used with low level
Similar tests on steel caging were conducted in the Czech Republic (Cirtek 2001).
with concrete strength f":12-17MPa. The cage were heated to 150-200'C prior to
loaded test results showed a possible maximum increase of 55o/o in load carrying
capacity. This method of strengthening was reported to be successfully used on more than
-25 -
CHAPTER 2 LITERATURE REVIEW
Valluvan et al. (1993) used external reinforcement ties to confine the lap splice
region of longitudinal bars. Concrete cover was not removed in the work and the
additional ties were grouted to the columns. Test results for the 72"x12"
(305mmx305mm) columns showed a signifìcantly improved confinement and splice
strength. It was concluded that the external reinforcement must be grouted to the column
Anchor bolts were used to enhance the confinement from rectangular steeljacket
by Aboutaha et al. (1996). Focus was placed on RC columns with inadequate lap splice in
the longitudinal reinforcement. Steel plates augmented by adhesive anchor bolts were
used to increase the passive pressure between splice bars to increase the frictional bond.
Anchor bolts were used when the column size was large and the steel jacket itself failed
unretrotitted ref'erence columns, were tested with the test variables of concrete strength,
width of column, number and location of anchor bolts, spacing between anchor bolts, and
height of steel jackets. Results showed that up to 36 inch (914mm) wide (longer side of
the rectangular section) columns were successfully retrofitted with tÁ inch (6.4mm) thick
steeljackets combined with a small number of adhesive anchor bolts. Satisfactory ductile
response under cyclic load to drift ratios in excess of 4o/o and large energy dissipation
capacity were obtained, compared to the very limited energy dissipation and virtually no
ductility of the reference un-retrofitted column. No axial load was applied to the column
in any of the tests because that was considered to be a more critical case for lap splice
problem.
-26 -
CHAPTER 2 LITERATURE REVIEW
jackets were proposed by Japanese researchers based on their extensive research work
(Sakino and Sun 2000). The jacket thickness can be calculated with simple formulae
based on the flexural/shear strength and ductility requirements. However, good seismic
response will be predicted if these design guidelines are applied to the column tested by
Chai et al. (1990) which was observed to behave poorly in the experiment.
Concrete filled steel tubes (CFT) are usually used in new construction. Although the
construction process of CFT is different from that for steel jacketing of RC columns,
there are similarities between their structural behavior in terms of interaction between the
steel plate (tube) and the concrete inside. Therefore, results obtained from CFT studies
l-î
However, it has been suggested that little interaction exists between the concrete
and the steel tube when both the concrete and the steel tube are loaded longitudinally
(Saafi et al. 1999). The reason was given in two parts. First, prior to minimum volume, )
concrete has a smaller Poisson's ratio than steel, and separates from the latter when both
materials are under the same axial strain. Second, shrinkage of new concrete causes a gap
at the interface between the two materials, even prior to any loading. As a result, such
columns fail by longitudinal yielding of the steel if yield stress is less than 400 MPa,or u, 1,.,-i
=;tt
by concrete crushing if yield stress is higher than 530 MPa. Therefore, the effect of 'l .
composite action is limited to prevention of local steel buckling and, to some extent, an
increase in concrete ductility. For this reason, the steel tube was cut short of the support
-27 -
CHAPTER 2 LITERATURE REVIEV/
in many studies of CFT columns in order to avoid activating the longitudinal strength of
the steel tube (Tomii et al. 1987; Aboutaha and Machad'o 1999). In this case, the CFT
columns are similar to steel jacketed RC columns from a confinement point of view.
The use of CFT columns dates back to the early 1900s when a number of bridges
and buildings were built using CFT columns in some countries such as England, Belgium
and Switzerland. Research on CFT columns were repofted since early 1950's (Russell
1953; Kloppel and Goder 1957a&,1957b Furlong 1967&.1968; Gardner and Jacobson
1967; Knowels and Park 1969e.1970). In the early study by Furlong (1967), twenty-two
circular and seventeen square columns were tested under various levels of constant axial
load as moments were increased. Another eight circular and five square columns were
loaded only axially. The interaction of the steel tube and the concrete was studied by
measuring longitudinal and transverse strain at four faces of the specimens. In the test,
the confinement of the concrete core by the steel tube was noticed when a sudden
In Japan extensive studies have been carried out on CFT columns. These
investigations have been focused on the effects of cross-sectional shape, aspect ratio
(diameter to tube thickness), concrete strength and column length (Tomii et al. 1973;
Tomii and Yoshimaro 1977; Tomii and Sakino 1979; Okamato and Maeno 1988; Orito et
Tomii and Yoshimaro (1977) conducted 286 column tests on CFT columns under
concentric axial load. Six sizes of circular, square, and octagonal steel tubes with
different wall thickness were used in the tests. For short columns, it was found that the
load deformation behavior of CFT columns was remarkably affected by the thickness of
-28 -
CHAPTER 2 LITERATURE REVIEV/
steel tube, the cross sectional shape and concrete strength. The load-deformation relation
for circular and octagonal columns showed strain hardening or an elastic-perfectly plastic
behavior with considerable increase in ultimate strength due to confinement, while for all
square columns the load-deformation curves were of a degrading type without an increase
columns and found that plastic buckling of the steel tube in the hinge regions tended to
occur when the columns were subjected to large cyclic lateral displacement.
specimens were tested with the variables of axial force, aspect ratio (diameter to
hysteresis response with large ductility and stable loops. Local buckling of the steel tube
occurred when the aspect ratio was greater than 39. The maximum load was determined
by the local buckling in this case. Local buckling was formed in the square columns with
an aspect ratio greater than 33. Square columns with smaller aspect ratios indicated richer
hysteresis response. The circular columns behaved in ductile manner despite the level of
axial force and aspect ratio. It was observed that ductile behavior can be expected only
from those square columns with small aspect ratio and low level of axial load. A similar
Sakino et al. (1996) reported that the degree of confinement by the square steel
tubes decreased as the aspect ratio B/t (sectional breadth/thickness of tube) and/or the
-29 -
CHAPTER 2 LITERATURE REVIEW
concrete strength increased. lt was also found that the effects of steel tubes on the flexural
behavior of tubed columns became more significant as the applied axial load was higher.
Stress-strain model for the concrete confined by a steel tube was also studied and
Seismic performance of circular CFT columns was also studied by Boyd et al.
(1995). Tests conducted on five 8" (203mm) diameter circular columns with aspect ratios
(D/t) of 73 and 107 gave poor hysteresis responses similar to those for ordinary RC
columns with poor confinement. A constant axial load equal to 0.17 Arf'"" was applied in
the tests. It was concluded that the seismic performance of CFT columns with a large
aspect ratio was questionable. This conclusion contradicted with others such as Chai et al
(1991) where aspect ratios more than 128 were found to be effective for a similar axial
load level of 0. I 77 A
s.f ,,, .
To maximize the lateral strength of steel for confinement, the steel tube was cut 25mm
short of column-beam joint, so no direct axial load could be applied to the steel tube.
Three levels of constant axial load corresponding to 0.0, 0.l2Ar.'f',,, and 0.76As.f,.,,
were used when the cantilever column was tested under cyclic push and pull at the top. In
contrast to many other researches, significant improvement on ductility was reported for
the steel-tubed columns compared to the normal RC column without steel tube. Outward
bulging of steel tube due to crushing and dilation of concrete was observed in the
potential plastic hinge region, but it was reported to have no detrimental effect on the
-30
CHAPTER 2 LITERATURE REVIE''ü
performance of the column. The performance of these CFT columns appears similar to
was used to apply pretension in the steel tube. Similar test results were reported by the
5
same author in another paper (Aboutaha 2000) with axial load level of 0.25A-.f", and
Às
o.36As. f",,.
-31-
CHAPTER 3 STUDY OF A NEW RETROFIT SCHEME OF COMPOSITE PLATING
COMPOSITE PLATING
In the study of column retrofitting, most attention in the literature was focused on
providing confinement to the core concrete. However, due to the congenital property of
which is as effective as that from a circular jacket. Therefore, the possibility of using
different retrofitting measures that are not dependent on confinement is explored in this
work. As shear retrofitting is not a problem with rectangular shaped RC columns, the
research focus is placed on columns with defectìve flexural performances that is, plastic
hinge failure modes. The major problern with the plastic hinge failure mode is concrete
crushing under large flexural strain. Therefore, the enhancement of the compressive
resistance will be a prime Target of a retrofitting design. Because of this reason, and also
The new retrofit system is illustrated in Fig.3.l. With this new scheme, steel or FRP
plates are bolted to the opposite faces of a rectangular column as in Fig.3.l(a) and (d). It
is therefore suitable for rectangular columns because of their flat surfaces. The L-shaped
- -)z -
CHAPTER 3 STUDY OF A NEV/ RETROFIT SCHEME OF COMPOSITE PLATING
and tensile discontinuity of the plate across the joint as illustrated in Fig.3.l(b). When the
lateral force F is to the left as shown, the restraint at the bottom of the left hand plate acts
as fixed as in Fig.3.l(b). In contrast, the right hand plate can rise as shown in Fig.3.l(c)
Reversal of the force -F, such as occurs under seismic loading, simply reverses the
restraint conditions.
N N
,1,
<J RC column
d
I I
Steelplate
Column bolts
oundation bolts
fixed e
A Tension
(a) Actuat (b) Ioealisea stiffness
+0
Comer of plate
opening up under
!L
I
tension
L=
(c) o"tuil a. (d)c.oss-section I-I
This novel retrofitting system has the following characteristics andlor advantages:
1. It increases the capacity of the compressive resistance of the column. The bearing of
the steel plate at the corner can transmit additional compressive force from the
column to the joint, thereby increase the compressive resistance of the column which
-JJ.
CHAPTER 3 STUDY OF A NEV/ RETROFIT SCHEME OF COMPOSITE PLATING
is a major defect for RC columns with the potential plastic hinge failure mode. The
increase in the compressive resistance in the column delays the onset of concrete
2. It provides a minimal increase in the tension capacity of the column. The tensile
connection of the plate to the joint. The increase in tension capacity of a cross-section
is not desirable from a ductility point of view. The reason is that it will offset the
plate. In addition, the momenl capacity of the column will be significantly increased
resistance in a column is not always desirable as it may transmit additional load to the
is to increase the strength (moment resistance) of the column, it can also be achieved
3. It provides a flexible partial interaction between the plate and the RC column through
the bolt connections. The stiffness of the additional compressive resistant system, that
helps to stop or delay concrete crushing, depends on the stiffness of the steel plate as
well as the stiffness of the bolts. It will be shown later in this work that proper
handling ofthis system stiffness can be used to achieve different design purposes.
other direction - longitudinal direction of the column. The longitudinal plate strength was
-34-
CHAPTER 3 STUDY OF A NEV/ RETROFIT SCHEME OF COMPOSITE PLATING
might increase the flexural resistance of the retrofitted column that would in turn apply
additional load to the foundations (Priestley et al. 1994b). However, from this study it has
been observed (see Chapter 5 and 12) that by proper design of the composite plating
system, it is possible to manage the longitudinal strength of steel plates in order to serve
The mechanism by which the above plating system improves the deformation
N".
ec er
I
Coricrete stress distribution
Centroid of RC section
reinforcernent bars
Concrete
Plate
-35-
CHAPTER 3 STUDY OF A NEV/ RETROFIT SCHEME OF COMPOSITE PLATING
The tension plate is ignored in this analysis for the reason mentioned in Section
3.1. Bending in the plate is also ignored in Fig.3.2 due to its negligible contribution to the
total flexural moment of the column cross-section ((El)ptut << (EI)nc corumn). From
in which N,,,,, is the axial force applied to the concrete only; Nnt, is the axial force on
the plate; N,,. and N", are the axial forces from the compression and tension
reinforcement bars respectively; and l/ is the total axial force or applied external axial
load of the cross-section. For a special case in which both the compression and tension
reinforcement bars yield before column failure and in which the column is symmetrically
N:Npt,*Nror, (3.3)
The strain distributions at failure are shown in Fig.3.3(b) for the unplated and in
Fig.3.3(c) for the plated column. It is assumed that the column fails when the strain at the
extreme fiber of the compression zone reaches an ultimate value á",, (the strain at
complete crushing or zero strength is chosen, see further discussion in Chapter 7). From
Eq.3.3 and for the case of an RC column without a plate, the total axial load is resisted by
the concrete alone as in Fig.3.3(b). Whereas, and in contrast for the columns with a plate,
part of the axial load is transmitted to the plate as in Fig.3.3(c). Therefore, the axial force
-36-
CHAPTER 3 STUDY OF A NEVY RETROFIT SCHEME OF COMPOSITE PLATING
Nplt N"on"
N"nn" : N
a
Compressive rebar
strain tsc
Strain profile A
(wilhout plate)
Strain profile A
c(rC U (without plate)
f.o
Strain of plate
-37 -
CHAPTER 3 STUDY OF A NEV/ RETROFIT SCHEME OF COMPOSITE PLATING
Comparing Figs.3.3(b) and 3.3(c), it can be seen that when the axial force in the
concrete or the area of the concrete stress block reduces, the compression zone depth x
reduces, as the height of the concrete stress block l" is the same for both cases. For a
certain ultimate strain ár,, at the compressive face, reduction of the compressive zone
depth x means an increase in the ultimate curvature of the cross section, from ç to rc6
From the above analysis, it can be seen that the reason for steel plating to increase
the deformation capacity of an RC column is because the plate attracts part of the axial
Although the above analysis is based on the assumptions associated with 8q.3.2
and the criteria that the column fails when the ultimate compressive strain is achieved in
the compressive face, the conclusion that plating the compression face increases the
deformation capacity of RC columns is general. This is because the steel plate in the
compression face generally reduces the axial load in the RC column, and reducing the
axial load on an RC column will increase the deformation capacity of the column. The
observation that the deformation capacity/ductility decreases rapidly when axial load
increases is well documented in the literature from both experimental and numerical
studies (Berwanger 1975; Sheikh and Khoury 1993; Watson and Park 1994; Watson et al.
The above analysis provides a fundamental mechanism for the plating system to
-38-
CHAPTER 3 STUDY OF A NEV/ RETROFIT SCHEME OF COMPOSITE PLATING
chapters to further investigate this system. However, limitations and assumptions apply in
As the initial step toward retrofitting RC columns by composite plating, the work
reported in this thesìs began from first principles. Considering the given time frame, it is
therefore necessary to apply the following limitations and restrictions to both the
theoretical and experimental work:
1. The study focuses on the plastic hinge failure mode. The shear failure mode has
effectively been resolved previously by jacketing, and the lap splice failure of
2. The study is limited to the flexural behavior of the cantilever column shown in
Fig.3.l where the deformation occurs inside the loading plane. Therefore, only single-
axis (one-way) bending about the principal axis is considered. This means that the
conclusions drawn from this work are immediately applicable only to uni-directional
columns and frames that are restricted to move predominately in one direction. This
often occurs in bridge columns or in buildings that are stiffened in the other direction
by strong shear walls. The cantilever model can represent either a cantilever bridge
column or a typical portion of a column in sway frdmes from the point of contra-
flexure near the column mid-height to the point of maximum bending moment. From
this point of view, the cantilever model can represent a typical part of a general
column in a frame.
-39-
CHAPTER 3 STUDY OF A NEV/ RETROFIT SCHEME OF COMPOSITE PLATING
3. Steel plates are only placed on the compression and tension faces. Side plated
columns where the lateral load F is parallel to the plane of the steel plates and
columns that are plated on all four faces are beyond the scope of this study.
4. All the results from this study are based on the following loading sequence. Firstly,
the plates are installed to the unloaded column. Secondly, the axial load is applied to
the centroid of the RC cross-section at the top and held constant thereafter. Finally,
the lateral load that varies subsequently in magnitude and direction throughout the
loading process is applied to the same point on top of the column as the axial load. In
practice, the axial load (even some lateral load) is applied to the column before the
the difference in response between a column that is retrofitted before axial loading
and a column that is axially loaded before retrofitting is minimal. Therefore, no other
5. The width of the plates is the same as that of the RC column, for convenience of
study.
l) The RC column is prismatic, having the same rectangular concrete cross section,
longitudinal reinforcement layout and material properties along its entire length. The
lateral confinement to the concrete due to variation of stirrup spacing along its length
is allowed.
-40-
CHAPTER 3 STUDY OF A NEW RETROFIT SCHEME OF COMPOSITE PLATING
2) Bernoulli's principle that plain sections remain plain applies to the RC section and the
plate sections, separately. That is, the relative movement, or slip, between the plates
3) Perfect bond is assumed to exist between the longitudinal reinforcement and concrete.
5) No transverse separation occurs between the RC column and the plate on the contact
interface. Therefore, the curvatures for the RC column and the plates are the same at
each cross-section. This assumption is based on the observation that little, if any,
separation occurs when bolts are placed sufficiently close. While in reality some
separation occurs between two adjacent bolts, the overall average deformed shapes of
Detailed studies on this new retrofìt scheme are described in the following
chapters, which include numerical, mathematical and experimental works as mentioned
in Section 1.2.
-41 -
CHAPTER 4 NUMERICAL MODELS
The numerical models developed to simulate the response of the composite plated RC
columns shown in Fig.3.l are presented in this chapter. This numerical model and the
analyze this type of composite structure as no existing software can be used for the
purpose of this study. The computer program is required to have the following functions:
Fig.4.1, for both ascending and descending (softening) response branches, as well as
o Material and geometric (Eqs. 4.41 and 4.42) non-linearities are allowed in the
program.
o It is able to model the formation of a plastic hinge in the column at the large
deformation stage.
o It is able to model the partial interaction, or slip, at the interface between the plates
numerical modeling. This method basically discretises the cantilever column into a series
-42-
CHAPTER 4. NUMERICAL MODELS
further divided into many sub-sections as shown in Figs.4.2 and 4.3. The instantaneous
integration of all the forces in the sub-sections that retain their individual loading history
throughout the process to allow for full non-linear modeling. The curvature of all sections
along the column length is integrated to obtain the deflection of the column. Details of
Áe
i
I 1
N
Section 0
rF
tt ÀLr
I Section I
I
t
I
I L¡
t
I A¡
¡
I Section i- l
¡
AL¡
0i ¡ Section i L
¡
t
I
¡
¡
I
n-l
Fixed end
^L
Section n
section is firstly presented in Section 4.1.k is followed bythe derivation in Section 4.2of
the mathematical formulas for integrating the curvatures to get the deflection. Simple
numerical tests are performed in Sections 4.3 to establish the minimum refinement of
-43-
CHAPTER 4 NUMERICAL MODELS
meshes for an allowable error margin due to discretisations. The procedures to calculate
the response of an RC cantilever column with full interaction plating, where the slip at
the interface is considered zero, is presented in Section 4.4. Finally, in Section 4.5 the
methodology and procedure to calculate the response of the partial interaction (where
The typical cross-sectional layout considered in this study is shown inFig.4.2, where all
a. t
Reinforcement
a
layer I
Top plate
Concrete
O O
-44-
CHAPTER 4. NUMERICAL MODELS
The reinforcement bars are modeled by "n" layers of steel with the positions of
the first and last layer allocatedat'a'distance away from the face of concrete, as shown
in Fi9.4.2. The internal layers of the reinforcement are assumed to be equally spaced in
between the first and last layers. Each reinforcement layer can have a different area of
steel. Plates can be placed on any of the 4 column faces. The top and bottom plates can be
bolted (allowing slip) or glued (preventing slip) to the concrete faces. The side plates can
horizontal layers and vertical slices, as shown in Fig.4.3(a). The strain distribution of the
1 NR slice I slice
ì.
" NLp layers
X
d
Neutral
AXIS
-..-t
NLc layers K (qrvúrc)
-45-
CHAPTER 4 NUMERICAL MODELS
The strain at each layer is calculated at the center of that layer and assumed to be
a constant for the entire layer. The concrete section is equally divided into NLc numbers
of layers as shown in Fig.4.3(a). In order to account for the bending moment in the plate,
the top and bottom plates are also equally divided into NLp numbers of layers with
smaller thickness than that for the concrete layers. To allow for variations of concrete
properties at the same layer due to variations in lateral confinement (confrnement level is
different at different points inside the cross-section), the concrete cross-section is further
divided into NR vertical slices. The vertical slicing is not applicable to the steel plate.
Ê,=(x+a,-ll.o (4.1)
in which ¡cis the curvature of the cross-section;x is the neutral axis depth from the top of
the concrete section; d, is the distance from the center of the layer to the centroid of the
concrete section, as shown in Fig.4.3. The stresses for the concrete, reinforcement and
plate are calculated corresponding to the respective strains and using their respective non-
The axial force of the section, which is assumed to remain constant, is given by
the summation of forces in all the discretised areas (sub-sections) of the concrete,
N = Nr,,rr* Nourr N
r,,
NLc NR ,, (NI4),,,,,+(Nl.p)¡,, y!," f¿,.2a)
= Il=l fIi-l A,u .o,¡¡) *2A,,.d,¡ * LAr, .o +lAr.,, .o
p¡ r.,,
¡=t ¡=1 ¡=l
-46-
CHAPTER 4. NUMERICAL MODELS
B H
where A,, = ¡/À.NLc is the area of one discretised sub-section of concrete and o c¡ lS
the concrete stress at the corresponding sub-section; A-, and o*¡ are the area and stress of
layer 'i' reinforcement, respectively. The 3'd term in F.q.4.2a is for the top and bottom
plates where An, is the sub-area given by Ar,: B'tlNLp and or, is the stress of the
corresponding sub-section. The subscripts 'top' and 'bot.' denote the top and bottom
plate, respectively. The fourth term is for the side plates, where An,, is the sub-area given
by A0,,, =2t' D I NLc (Ap,, = t'D I NLc if one side plated only), and o u,, is the stress at
The moment at the cross-section is given by taking moment of the forces in each
Two stress-strain models are used for the concrete in this work. The first model is a
simple triangular model that is mainly used to corelate the numerical results to the
analytical results where the simplif,red stress-strain model is adopted. Another more
complicated model is used for the numerical simulation in order to model the response of
-47 -
CHAPTER 4 NUMERICAL MODELS
extensively studied in the literature (Sinha etal. 1964; Karsan and Jirsa 1969; Popovics
1970&.1973; Kent and Park 1971; Park et al. 1972;Desayi et al. 1979; Muguruma et al.
1980; Sheikh and Uzumeri 1980&1982; Ahmad and Shah 1982; Maher and Darwin
1982; Park et al. 1982; Shah et al. 1983; Yankelevsky and Reinhardt 1987; Mander et al.
1988a&b; Fujii et al. 1988; Razvi and Saatcioglu 1989; Saatcioglu. and Razvi 1992;
Madas and Elnashai 1992; Pekau et al. 1992; El-Dash and Ahmad 1994&.1995;
Almusallam and Alsayed 1995; Attard and Setunge 19961. Martinez-Rueda and Elnashi
1997; Hoshikuma ef al. 1997; Bahn and Hsu 1998; Spoelstra and Monti 1999; Sakai and
Kawashima 2000). The non-linear stress-strain model proposed by Mander et al. (1988a)
accounts for lateral confinement effects due to the transverse reinforcement. This model
has been extensively tested against experimental data and appears to be the most widely
accepted and used in the literature for the steel confined concrete (confinement is
provided from steel elements such as stirrups). The advantage of this model is that its
simple and explicit form minimizes the computational effort. As the confinement
considered in this work is only from the lateral reinforcement or stirrups, Mander's model
in which a linear relation is used for both ascending and softening branches of the stress-
strain curve. The tensile strength is ignored. The vertical axis .{, and horizontal axis e.
specify the stress and strain of concrete, respectively. The strength of the concrete is
-48-
CHAPTER 4. NUMERICAL MODELS
denoted by f",, with the corresponding strain at the maximum strength given by e...,. The
ultimate strain of the concrete is given by t,,, at which the concrete is completely crushed
with zero strength. The elastic modulus of concrete is given by E" used for the slope of
the loading/unloading branches, and E', gives the slope of the post-peak softening branch
as shown in Fig.4.4.
(e*,f")
f" envelope curve
unloading/reloading
branch
(t.,, tc
4.1.2.2.11o Section 4.1.2.2.5. Minor modifications have been made to the original model
due to some subtle problems in using the original model. These modifications will be
-49-
CHAPTER 4 NUMERICAL MODELS
Confined Ftrsl
\U hoop
14
frc f roct
{4
q,
(/)
q,
fro
l/l
ed
t4
\a-
G,
E r
I
Er cc
-f"
f ,, .x., (4.3)
r-l+x'
where
o^
| = ---)- (4.4)
€""
E"
(4.7)
E, - Er""
-50-
CHAPTER 4. NUMERICAL MODELS
and
-f""
E r.. (4.e)
cc
ln Equations 4.3 to 4.9, f"o and eco represent the unconfined concrete strength and
corresponding strain, respectively, and f", and €," denote the confined concrete strength
and corresponding strain, respectively fì is the effective lateral confining pressure due
to the lateral reinforcement. Mander et al. (1988a) and Watson,Zahn and Park (1994)
proposed a series of formulae to calculate the effective confining pressure for circular and
f": f,,'x'r
f,, r-7+x' (4.r 0)
where
E,
r (4.1l)
E, - Er""
E -fu,
(4.12)
-51-
CHAPTER 4 NUMERICAL MODELS
c
uc _c urn
-- - (4.13)
€
r¡ - €r,n
b=
f"' ,-7 (4.1s)
.f
",,
05
I cc
C= <l (4.16)
a
c-c
upl -uun (4.17)
to=o e .c (4.18)
and
F"l
lEØ,fun)
E€
l+r
-52-
CHAPTER 4. NUMERICAL MODELS
in which .fu, and t,,,, are the stress and strain at the onset of unloading branch,
respectively.
A linear stress-strain relation, shown inFig.4.7, is adopted to model the reloading branch
where
and
lfcl
kL!
-7'E
* l""l !-
(€un,lun) A---. C €r" ,1")
leun,f *n)
B (s,n,,fi"*)
Ept It.l
(a) (b)
-53-
CHAPTER 4 NUMERICAL MODELS
The parabolic transition curve between the linear reloading branch and the
given by
where
x= Ec - €re Ø.24)
8 r" = tr,,, +
.fu, - f u",, (4.2s)
E,Q+*)
and
E, - 8,"
A (4.26)
- 4l(f ,,",,, - .f ,") - E, (€,,,, - t,)]
in which 8,, is the return point modulus of elasticity on the monotonic stress-strain
curve
transition curve with two apparent problems. Firstly, the common retum point C
calculated wirh F,q.4.25 is irrational for the following reason. If we extend the linear
reloading path with the same slope E, from point B to point E which is at the same level
as the unloading point D, as shown inFig.4.7(b), it can be seen that the strain at point E is
given by
-54-
CHAPTER 4. NUMERICAL MODELS
{{
rvhere ¡r-Jm Jneu'
Substituting Ae into 8q.4.25 gives
E,
tr"=t,nlLt=Êr,,
From the above relation and Fig.4.7(b) it can be seen that F.q.4.25 gives a return strain
Ê,", i.e. the x co-ordinate of the common return point C, within I 13 of DE from point D.
This return point is therefore usually on line DA to the left of point A (A is the
intersection of DC and BE). The purpose of this parabolic transition curve is to account
for the cyclic degradation. When the common return point is to the left of point A, the
stress-strain curve is not degraded, instead it is 'upgraded'. This return point is therefore
This same difficulty has been noted by other researchers in applying this model.
Elnashai andElghazouli (1993) found that the transition curve of Mander's model "gives
occasionally stresses that exceed the envelope curve, a straight line possing through
(e,n,-fn"*), and having a reduced slope was thereþre used (from point B to C in
Fig.4.7(b))". Actually, based on the above analysis, the transition curve of the original
Mander's model gives erroneous result more than "occasionqlly". However, using a
straight line to connect point B to C is acceptable because the transition curve is only a
very small part of the total curve and will only have minor effect y/neoverall behavior.
'ç1
-55-
CHAPTER 4 NUMERICAL MODELS
moving á'" from left side of point A to the right side of point A with the following
6rc=ttn+
.fun - .f n"n, (4.28)
E,
.f"o
This revised reloading branch produces a nice and smooth transition, as will be seen later
from Fig.4.l0.
The second problem with Mander's original model is that the value of 8,." is not
given and the coefficient A given by 8q.4.26 is incomect. The transition curve must
satisfy three conditions based on Mander's original model: (l) it must pass through point
B; (2) it must pass through point C; and (3) it is tangent to the monotonic curve at point
C. These three conditions can used to determine the three coefficients .f,", 8," andAin
f", r I cc
-f,o (4.2e)
r-l+
a )'
E r 8."" (4.30)
õ
r-l+ r -1+
c
)' € cc )J,
and
-56-
CHAPTER 4. NUMERICAL MODELS
(4.3 r )
The tensile stress is modeled by Fig.4.8 in Mander's model. For monotonic loading
where I is the tensile strength of concrete under monotonic loading. To account for the
lfcl
fcc
--
ft t.l
Er
-f,=8,.(c,-trr) (433)
E "f,, (4.34)
8t
-57 -
CHAPTER 4 NUMERICAL MODELS
t f, (4.3s)
E a
Ept
when €ot 3t,,
f,, =-f, c
cc (4.36)
0 when tpt >t",
When the tensile strength is exceeded, cracks open and the tensile strength for all
when searching for the neutral axis of a cross section. The sudden drop of the tensile
stress when the tensile strength is reached can cause non-convergence of the searching
modifications to the above model are made with reference to the tension model by Rots et
al. (CEB report 1996). Rots' model is shown in Fig.4.9(a). The modified model used in
fc
f"
2tt
%
;_--.-, tc
tc
ft
(a) Tension modelby Rots et al. (b) Modified tensile stress model
-58-
CHAPTER 4. NUMERICAL MODELS
In the modified model, the values of t,, er, and .f,, in the modified model are the same
as that in Mander's model. Each time the peak of the tensile curve is passed, the
remaining tensile strength drops, e.g. if the previous tension loading goes along the path
A typical stress-strain curve from Mander's model with the modifications to the transition
curve and the tension model is shown in Fig.4.l0. The following parameters were used to
produce this curve: f,,, = 44MPa, t,n = 0.002, .f,, : 48MPa, f, :8.0MPa .
60
50
40
30
Stress
(N)
20
10
-10
Strain
For a similar reason to that for concrete, one simple and one sophisticated non-linear
model is used as the constitutive relation for steel elements which include reinforcing
-59-
CHAPTER 4 NUMERICAL MODELS
The idealized bi-linear model is shown in Fig.4.l1. This model is mainly used for the
steel plates. It is also used for reinforcing bars when verifications between numerical and
mathematical results are made, where the simple stress-strain relation is used for both
(ev
Es
The Menegotto-Pinto model, with modifications by Filippou et al. (Gomes and Appleton
(4.37)
where
o-o
o (4.38)
õu-4,
+ oc_ c
or
(4.3e)
c
(rr-c
-60-
CHAPTER 4. NUMERICAL MODELS
and
R=R - *t 5
n -F
(re (4.40)
dz*Ç
in which oand áare stress and strain of the steel respectively; oo and to ate the stress
and strain at the point where the two asymptotes of the branch under consideration meet
(e.g. point A in Fig.4.l2); o, and e, are stress and strain at the point where the last
strain reversal with stress of equal sign took place (e.g. point B in Fig.4.l2); b is the
strain hardening ratio or the ratio between Eo and Ë,,; and rR, given by Eq.4.40, is a
constant taking into account the Baushinger effect. R depends on the absolute value of the
plastic strain of last excursion ( (".g. €, and (, in Fig.4.l2) as well as the material
t¡
A b
n(gr)
b ar6
û
ø
F(È) = Âo - az+Ê
o
.tl
.)
{¡
D
E
o
.!
t
É
zo
R(ì')
-1 t
ll
l.-
-10 0 t0
Nømalized steel strain €'
-61 -
CHAPTER 4 NUMERICAL MODELS
To quantify the interaction between the plates and the RC column, the load deformation
relationship of the connecting agent, i.e. the anchor bolts or shear connectors, has to be
defined. The relation between the lateral load Fo on a bolt and the lateral deformation of
the bolt, or the slip s as shown in Fig.4.l3(a), is actually very complicated. This relation
is not only affected by the geometric and material properties of the bolt, but also related
to the geometric an{ material properties of the concrete and the plate. Because of the
Similar load-slip tests are conducted in this work (reported in Chapter l0). Based
on the test results, a bi-linear model, similar in shape to that shown in Fig.4.l I , is adopted
in this work. The softening part from point C to D shown in Fig.4.l3(b) is not likely to
occur in a practical column, as the concrete will have crushed long before such a large
slip occurs.
Fu
plate
:-,
RC column
'-
.+ o
)o
30
Slr¡r (mnr)
Fig. 4.13 Typical load-slip relation of an anchor bolt (Teh et al. 1999)
-62-
CHAPTER 4. NUMERICAL MODELS
The displacement at the top of the cantilever column can be calculated from the
integration of curvature along its length. In discrete form, the column is divided along its
length into'n' segments as shown in Fig.4.l. The curvatures at section '1' to section 'n'
are then used to calculate the displacement of the cantilever. Different discrete forms of
integration exist in the literature. From numerical tests, some were found to converge
faster (to get certain precision with fewer number of element), others slower. In this
section, equations relating sectional curvatures and deflection are derived which are
dv.
0:arctan('dx. ) (4.41)
and
t2
av
1d0 dxt (4.42)
rds
l'..,fr,'l%
in which v is the lateral deflection of the column; dis the rotation of a cross-section;x is
the vertical coordinate; r is the curvature; r is the radius of curvature; and s is the
coordinate along the length. Eqs.4.41 and 4.42 are general equations without
approximation, taking into account large deflections.
-63-
CHAPTER 4 NUMERICAL MODELS
For small deflection, ds*dx, and 0xtan9=dvldx. Eqs.4.4l and 4.42 can then be
simplified to
0--dv (4.43)
dx
de d2v
(4.44)
dx dxt
or
-l
0=0,+[rc dx (4.4s)
v=v,+ l0 dx (4.46)
0
where 0,, and vn aÍe the rotation and deflection at the boundary. For a cantilever without
rotation and deflection at the bottom (fixed end), the top deflection is given by
ao 0 .dx (4.47)
J
0
o= I* -dx
0
(4.48)
In discrete form
I
I n
-64-
CHAPTER 4. NTJMERICAL MODELS
-"-
=0.25 Il It",
,.f I ,
, +r,)'LL,l'a.1,+0.25'I ¡-r-Hrc t)' AI AT,
l¡=i i=t j ,=r
(4.51)
Equation 4.51 has double summations that are not convenient to use. It can be
=0.25-(rco+Kr)-&-LI.+0.25.(a+Kr).N".(M,+N")+...+0.25'(rc,-,+K).&.( h+N"+...+44)
+...+0.25. (rc*, +rc). M,. (N. + N"+...+ +...+ M,)
^4
=0.25.(rco+6).^4.1.+0.25.(rc,+rcr)-N"-1"+...+0.25'(r,r+rc).LL,-Ir+...+0.25-(rc*,+rc)-LL,-L,
=o.zs.f@,-,+K).M,.4
in which L,=ZM¡. Similarly the second part of Eq.4.5l can also be transformed to
j=t
n-'l
:0.25.2@, +rc,*r). M,*r. L,
:4, *A¿
"'Ao
n-l
: 0.2 5.2@, _, +rc,). LL, . L, +0.25.1(rc, +rc, u). LL,u L,
.
-65-
CHAPTER 4 NUMERICAL MODELS
let Lo =Q
I Mi
L,+L,_,
2
(4.s2)
This relation can be illustrated by the geometrical relation as shown in Fig.4.l4, where
6,:fõ0,.L,=f, K¡;+Ki
^o=É 2
Sec. 0
õ02
Sec. I
ô9,
Sec. 2
Sec. 3
Equation 4.52 relates the deflection of the column directly to the curvature of
-66-
CHAPTER 4. NUMERICAL MODELS
by others.
n
a^o =10,.LL, (4.s3)
Large deflections refer to the largely deformed shapes where it is not accurate to use the
simplifications of ds = dx, and 0 x tan0 : dv / dx. In this case, the exact expressions
,{
0- Irc-ds (4.54a)
0
and
sino = L (4.s4b)
ds
From Eq.4.54b
II
4 = ldv: [sino.ds
r=0 0
In discrete form, the following relations can be derived with reference to Fig.4.l5
which represents a typical segment from section i-1 (point A) to section i (point B).
e .-Ê
AB,=2.(R. si¡:-i:1----!-) = 1.sin '
) (+) .sin1LrL-e,) (4.55b)
ltc 2 Ki_t+K, 2
-67 -
CHAPTER 4 NUMERICAL MODELS
Based onEq.4.54a, the discrete form to calculate d, is still given by 8q.4.49. Therefore,
deflections shall be calculated by 8qs.4.49 and 4.55a&b when large deformations are
involved.
ê'
When the column member is in an elastic deformation stage, the deflection of the
cantilever column can be calculated by the formulae given in the previous 2 sections. In
this case, the calculated deflection result will not be sensitive to the choice of the
segmental length A,L, (i:I-n) as long as enough segments, or a fine enough mesh, is
used. However, this is not the case when a plastic hinge forms. Once the inelastic
deformation stage is reached, only the bottom cross-section can pass the peak (point A of
the moment-curvature curve shown in Fig.4.l6) and proceed onto the descending part of
the curve. All other cross-sections above the bottom section will go along their unloading
-68-
CHAPTER 4. NUMERICAL MODELS
paths from some point to the left of A as shown in Fig.4.l6 after the bottom section
passes its peak. This is because the reduction of moment in the bottom cross-section
reduces the moment of all the other sections above the base which are linearly related for
the cantilever column. Therefore, further increases in deformation of the cantilever can
only come from the deformation of the column at the bottom part, namely the plastic
hinge region.
Moment
Mt
--ascè-nTing
descending
branch bra¡lch
path
path
Curvature
This plastic hinge region must have a certain length, because deformation of a
cross-section with zero longitudinal length will not produce an overall deflection and
hence no further deflection after the peak point can be predicted which is certainly not
true in practice. This analysis is consistent with engineering experience that when large
However, it can be simply modeled by applying the curvature of the bottom cross-section
-69-
CHAPTER 4 NUMERICAL MODELS
to a certain length or the plastic hinge length 2,, (Priestley and Park 1987) which might
Zn =0.08'L+6'do (4.s6a)
where Z is the cantilever length shown in Fig.4.l and dh is the diameter of the
longitudinal bars. Amendment to Eq.4.56a was made by Paulay and Priestley (1992) that
AL,
length of the last segment by F.q.4.52. This half-length of the last segment where the
2
bottom curvature r,, is applied to is essentially the plastic hinge length Ln if 8q.4.52 is
used. In this case, the choice of length for the last segment will directly affect the result
hinge length that is calculated using either 8q.4.56a or 4.56b. The curvature of the
adjacent section or section r-1 is then applied to an adjusted length of M:'-' +A1.,-L,
"2
M"!M'.
instead to the original length
"2 of For convenience, the bottom curvature is
applied to the plastic hinge length regardless of whether the plastic hinge has formed or
-70-
CHAPTER 4. NUMERICAL MODELS
It is assumed in the previous sections that the cantilever column is perfectly fixed at the
bottom, or 0,,:0 in Eq.4.45. However, this is seldom the case. The cantilever column
may have a base rotation fl which is not zero, and which is caused by pulling out of the
tension bars and pushing in of the compression bars relative to the fixed base under
flexural loading, as shown in Fig.4.l7a. This base rotation may have a significant effect
on the calculated displacement of the column. Of course, this base rotation depends on
In fact, the base rotation was considered in the calculation of the plastic hinge
length in F.q.4.56a by Priestley and Park (1987). The additional rotation in the plastic
hinge due to the base deformation was allowed for in the second term of Eq.4.56aby 6dn
which was attributed to "yield penetration" into the base. The first term of 0.08.2 in
8q.4.56a was obviously due to the plastic deformation of the column itself. In their
original model, all plastic rotation from both the column and base was considered to be
accounted for by applying the plastic curvature at the bottom section to the plastic hinge
length calculated with Eq.4.56a. This plastic rotation was then concentrated into the
center of the plastic hinge that was above the fixed base. Based on this concept, the
However, it is found from the numerical simulations in Section 12.2.2 that the
theoretical response curves based on the model (Fig.a.l7b) do not agree well with the
experimental curves. This discrepancy is found to be caused by the shifting up of the base
-71 -
CHAPTER 4 NUMERICAL MODELS
rotation, which actually takes place at the bottom of the column, to the center of the
reinforcing
bars
column column
before after
loading loading
0
a
c
Lo¡
Fixed end base line E
without yield with yield
penetratton penetration
If the same plastic hinge length Ln inEq.4.56a and Fig.4.17b is now split into
two parts of L,,,,:0.08¿ which is above the base and L,,r,:6do which is belowthe base,
72-
CHAPTER 4. NUMERICAL MODELS
the shaded areas, will be different. Obviously, the model of Fig.4.l7c gives an extra
rotation at cross-section T that is equal to the area of ABCG. This means these two
different models give different response curves with the same plastic hinge length.
give an accurate response curve even if the plastic hinge length is adjusted arbitrarily
(changing plastic hinge length to check it's effect on the response curve). This will be
iv.
,."n tJq¡h" numerical simulations in Section 12.2.2.In contrast, the fixed end rotation is
adequately accounted for by the model of Fig.4.l7c, where the plastic hinge is split into
L
r, = 0.08' L (4.56c)
where the fixed end rotation is now given by 0,=K,.Ln6 which is the rotation at the
base. The fixed end rotation or yield penetration into the support has also been
The following numerical tests are conducted to check the convergence and accuracy of
-73 -
CHAPTER 4 NUMERICAL MODELS
l. Test No.l - Identifying the number of layers required for discretisation of a cross-
section.
For the cantilever column with a bottom curvature of ¡r: I .5 x l0-s, the maximum
The numerical results from the program "PLTCOL", using different numbers of
layers to discretise the cross-section, are given in Table 4.1. From these results, it can be
seen that 10-15 layers are usually adequate to give a result with an error less than 1%o.
6 1.23047285 1.56
9 |.24132358 0.69
15 1.246879 0.2s
30 1.2492229r 0.06
75 1.249879 0.01
-74 -
CHAPTER 4. NUMERICAL MODELS
deflection are
An=
F.I: =5(mm)
" ^-
3EI
In the numerical calculations, 150 layers were used to discretise the cross-section in
column is discretised with different numbers of segments and the numerical results of
the top displacement calculated using both Eq.4.52 and Eq.4.53 are listed in Table
4.2.
Clearly, the results using Eq.4.52 has a much smaller error for the same number of
length (mm)
-75 -
CHAPTER 4 NUMERICAL MODELS
It can be seen from these results that 4-8 number of segments are usually adequate to
The same test column is discretised into 150 layers in cross-section and longitudinally
with 8 segments of l25mm long each. lt can be seen from Table 4.3 that the
deflection calculated by F,q.4.52 with small deflection assumption is good enough for
most cases unless the inter-storey drift ratio is very large in which case the rigorous
-76-
CHAPTER 4. NUMERICAL MODELS
column with steel plates glued to its sides is described in this section. When the plates are
glued to the sides of the column, the strain distribution is assumed to be linear and
continuous across the entire cross section including the plates. In this case, no slip occurs
between the concrete and the plates. The numerical analysis is performed using a
deformation control procedure. The curvature at the bottom of the cantilever column,
where the bending moment is a maximum, is used as the control parameter. The
advantage of using this curvature as the control variable is that it can follow both the
ascending and descending parts of the moment versus curvature curve. It can also cater
for cyclic loading by just reversing the increment of the control parameter and can allow
4.4.1 Procedure
The cantilever column is discretised longitudinally as shown in Fig.4.l and all cross-
sections are discretised with an identical pattern as shown in Fig.4.3. However, the plate
affangements, such as termination of plate, at different sections are made. The following
L Starting with an initial curvature of the bottom cross-sectioî K,= 0;set curvature
increment Ar,
2. For the new curvature of the bottom cross-section rc,(new) = rc,(old) I A,rc,, assume
a neutral axis position, then calculate the corresponding strain and stress distributions
-77 -
CHAPTER 4 NUMERICAL MODELS
in the section. The axial force of the cross-section is then calculated based on the
stress distribution as described in Section 4.1.1. If the axial force of the cross-section
is not equal to the constant axial load applied, the neutral axis position is adjusted
until the difference between the calculated axial force and the constant axial load is
less than the allowable tolerance. This iterative process is called the search for the
neutral axis. Then the bending moment M,, of the cross-section is calculated based
3. The momenT M, for the rest of cross-sections along its length is calculated by
where Z, is the length from the top of the column to section i as shown in Fig.4.l.
to locate the neutral axis position based on K, and the constant axial load. Then
calculate the bending moment of the cross-section. If this trial moment is not equal to
the moment M, calculated in step 3, adjust the trial curvature r, until the difference
between the trial moment and M, is within the allowable tolerance. This process is
The curvature search may have multiple values of r, for the same M, value,
as shown in Fig.4.l6. For a non-linear system, the stress of an element is not only
determined by the strain but also depends on the loading history. With different
loading histories, the stress-strain curve goes along different paths. Therefore, the
loading history information must be recorded for all sub-section elements in all cross-
sections separately. For a specific loading history, there is only one identical
-78-
CHAPTER 4. NUMERICAL MODELS
one identical branch, normally two different values of K, can be obtained for the
same M,value, e.g. Kr and rc, as shown in Fig.4.16. Fortunately, only the bottom
cross section can pass the peak (point A) of the corresponding moment-curvature
curve. All the other cross-sections will follow the unloading path when the bottom
cross-section passes the peak point A going along the descending branch. In other
words, only one value on the left hand side of the peak on the corresponding curve is
To keep correct track of loading history, the step length or Ar, must be
sufficiently small. Otherwise incorrect results may be obtained. For example,if Lrn
is so big that the bottom cross-section moves from the current point B to next point F
as shown in Fig.4.l6, a nearby cross-section, say section n-1, should move from point
S to G following the path of S-E-G. However, the program will give an erroneous
result that is from point S to point C which has the same moment value as point C.
This is because the cunent loading history for the element only has the information
that occurs before point S for section n-1. When section /r goes to point F which has
the same moment as point D, the calculation of moment for section n-l in step 3, i.e.
M,_t = M,. L,,_t I L, gives the same moment as point C. Therefore the program will
take point C as the new solution for section n-l.lf the curvature increment of cross-
section n or Ar, is small enough, the computer will be able to trace the loading
-79-
CHAPTER 4 NUMERICAL MODELS
5. After obtaining the curyatures for all the cross-sections along its length from step 4,
calculate the displacement A¡ using formulae given in Section 4.2.The lateral force at
the top is given by r = M,,l L. This gives one new point for the force-displacement
response curve.
6. Record this step of the loading history. The information to be recorded includes the
maximum excursion along the envelope curve, unloading and reloading points
model, current and previous points information, etc. All the discretised sub-sections
of concrete and plates, as well as all reinforcement layers for cross-section 1to cross-
section z must have their separate records. Therefore, when a fine discretisation mesh
is used for the cross-section and along its length, the information to be recorded is
substantial.
7. To calculate the next point of the force-displacement curve, go back to step 2 for
another new K¿. Repeat this process until the required force-displacement response
curve has been generated. When cyclic response is required, simply reverse An,
In the calculation of displacement by 8q.4.57, the P-A effect is ignored. For slender
columns, the P-A effect can be substantial. To consider the P-A effect, 8q.4.57 must be
-80-
CHAPTER 4. NUMERICAL MODELS
The above equations cannot be solved directly since the moment distribution which is
required to calculated the displacement along the length is not known before the
displacement is given.
1. For a new curvature lc,, at the bottom section, calculate the column displacement
A¡(l = 0- n-l)with the same procedure as in Section 4.4.1 without considering P-Â
effect.
2. Revise bending moments for section I to n-l using Eqs.4.58 and 4.59 with the
displacement distribution obtained from the previous step. Search section curvatures
for section I to n-l based on the revised moment. Then calculate the revised column
displacement A, (i = 0 - n - 1) based on the new set of curvatures.
3. Repeat step 2 with the revised set of A, until the difference between the two
n-1
l=0 ",,,,",, =
moment along the column and, therefore, is called the rigorous method. However, it is
numerically intensive and very time consuming. To get each point in a force-
try&-
displacement response curve, about four iterationsþeeded. That is to say four times the
original computation time is required compared to the case where no P-A effect is
- 8l -
CHAPTER 4 NUMERICAL MODELS
To reduce the computation time without significant loss of accuracy, a simplified method
has been developed in this work for the P-À effect. From Eq.4.59, it can be seen that
where F' is the lateral force required to achieve a moment M,, at the base of the
cantilever column considering the P-A effect, and F = M,lL is the lateral force required
to achieve the base moment M,, wiThout considering P-A effect. Based on the above
equation, the lateral force considering the P-A effect, F' , can be obtained from the lateral
force without considering the P-A effect, F, if the deflection A0 at the top which
response curve without considering the P-A effect is produced, as shown in Fig.4.l8a by
the curve marked with 'without P-A effect'. Each point in the response curve 'without P-
A effect', e.g. point A, is then transformed to a new point, e.g. point B, in a new curve as
where F' and F are the y coordinates of the new point B and the old point A,
respectively; A'o is the x coordinate of point A, i.e. the calculated top displacement under
the lateral force I' without considering the P-A effect. The ¡ coordinate of a point in the
-82-
CHAPTER 4. NUMERICAL MODELS
new curve, e.g. A'o of point B, is the same as that of the old point A. The transformed
new curve will be proved to be very close to the actual response curve with the P-A effect
without A (À',. F)
P-A effect
actual response
P-À effect
Lateral
force at B (A'0, F C
top of
column
simplified
method
À'o
t _t
N
1
{------tü F
ìl
\r
\. 1
I
t
t L
I,
Section i
F'*L¡ N*
I M; F*L¡
¡
\
\
\
Section n
-83-
CHAPTER 4 NUMERICAL MODELS
F'= F-{'o'o - F-0.L'o= F-LF. This transformation is similar (but not exactly
LU
equivalent) to rotating the original force-displacement curve without considering the P-A
To estimate the error in the response curve produced from the simplified method,
consider the moment distribution of M, - F .L, without the P-A effect, as shown in
Fig.4.l8c, and the corresponding deflected shape of the cantilever column, as shown by
Fig.4. l8b. Introducing the revised lateral force F' which satisfies Eq.4.60 and
equivalent to a moment F''L, caused by a revised lateral force ,F' plus a moment of
u.li.t
L
0 caused by the axial load with the eccentricity of A, =1.O'o If the
T.
eccentricity I
A'o is the same as that under the deflected shape of Fig.4.l8b which
L
lateral force ,F' together with the lateral deflection A'o is a precise point in the force-
displacement response curve with the P-A effect strictly considered. In other words, the
-84-
CHAPTER 4. NUMERICAL MODELS
force F' with the exact P-A effect considered if the eccentricitv'Lof lt A'^ is correct
However, the deflected shape under the moment distribution of M, - F 'L, is not a
Li
straight line, hence the eccentricity given by A, = À'o is not accurate except at the top
L
and bottom of the cantilever. Nevertheless, the error due to this linear assumption of
The actual eccentricity under the deflected shape that is calculated from the
in which Err(i)- õi-!.O', is the error due to the linear assumption of eccentricity as
LU
shown in Fig.4.l8b. This error in moment is insignificant compared to Eq.4.62 for the
following reasons.
o When the deflection of the column is small, P-A effect is small. The error in moment
. When the deflection is large and the plastic hinge forms, the deflection of the column
is mostly due to the plastic hinge rotation at the bottom, and the deflected shape
above the hinge base of the column is relatively straight. ln this case, Err(i) is
-85-
CHAPTER 4 NUMERICAL MODELS
a For deflections in between the above two cases (between the small deflection and the
which gives
cannot be very small. Hence F'l N is usually not very small. Therefore, the term
+¡/ :=
Err(i)
usually gives a very large value. From Fig.4.l8b, ¡t can be seen that the
second part of Flq.4.64, ¡.. --4 , is also a very large value near the bottom of the
Í,ru1t)
column where the moment contributes most to the deflection of the column. At the
middle and top part of the column this second part of is relatively smaller than
Eru(i)
that in the bottom of the column, however, it is still greater than 1.0. In the mean
time, the contribution of a moment in this middle and top part of the column to the
deflection of the whole column is also smallerthan that from the bottom part (as can
be seen from Eq.4.52), which means that an error in moment at this region causes less
-86-
CHAPTER 4. NUMERICAL MODELS
error in deflection than that caused by the same error in moment at the bottom of the
column. As both the first and the second parts of 8q.4.64 are positive, the summation
M',-M, 11 M',.
Eq.4.61will only cause a relative small error in moment as compared to the total moment
of a cross-section. For all the cross-sections except the bottom section, the moment-
curvature relation is in the ascending part of the curve to the left of the peak (point A) as
shown in Fig.4.16. A small error in moment will only cause a small error in curvature
value hence a small error to the deflection. Although, a small error in moment may cause
a large error in curvature forthe bottom cross-section in the descending part of the curve
shown in Fig.4.16, the error introduced by Eq.4.60 for the bottom cross-section is zero. In
other words, the error in curvature distribution hence the deflection caused by the
Therefore, the moment at all cross-sections, except the top and the bottom, is
underestimated slightly. That is to say the actual deflection should be slightly larger than
A'0, as shown by point C in Fig.4.l8a. In other word, the actual deflection under the
lateral force f'' should be Ao at point C instead of A'o at point B. This small increment
-87 -
CHAPTER 4 NUMERICAL MODELS
in displacement from the simplified response curve to the actual response curve where
the P-A effect is strietly considered will be seen from the numerical example in Fig.4.l9.
The conclusion that the error due to the simpliflred method is small has been
relation for a200(B)x200(D) RC cantilever column with an axial load of 360kN is shown
are used for the stress-strain relation of the concrete and the longitudinal reinforcement,
Longitudinal reinforcing bars - 4 Y16 with .f,y : 547 MPa, E" : 200 GPa,
-88-
CHAPTER 4. NUMERICAL MODELS
(
4
4
a
z
o
o
o 2
õ
o 2 no P-D effect I
o method,
J I
" -rigorous method
I 000
-simplified
10 10 30 50 70 90 110 13 0 l5 0
45000
40000
35000
z 30000
o
o 25000
o
G 20000 no P-D effect
o 15000 plified m ethod
g
10000 i- -' .-- =rigorous method
5000
-sim
0
0 Á 10 15 20 25 30
d isplacem ent at top (m m )
-89-
CHAPTER 4 NUMERICAL MODELS
Watson and Park (1994) is analyzed in this section. The test column, as shown in
and 25 slices as shown in Fig.4.2l. Four layers of reinforcement are located exactly at
their design positions in the cross-section with the areas of 804 mm2 (4Y 16),
402mm2 (2Y16), 402mm2 and 804 mm' for the l't to fourth layer, respectively. The
actual double cantilever columns with a stub are modeled with a l.7m long cantilever
column fixed at the bottom. It is longitudinally discretised into 5 segments with lengths
of 400mm, 400mm, 400mm, 27}mm and 230mm from top to bottom, respectively. The
230mm long bottom segment is used to model the plastic hinge length of 230mm as
-90-
CHAPTER 4. NUMERICAL MODELS
P
bter rBmm
tu
E
Potentiol Ê
E
a
plostic hinge
reg,onS
E
a
o
\
o
(ô
E
6lnmn
H
I\t E
o
o
qt
Units z
E tb)
E
I
=h
In
P
40Omm Elevotion
lol
25 slices -t
1\
25
layers
Mander's model and Menegotto-Pinto's model are used for the stress-strain
relation of the concrete and the longitudinal reinforcement, respectively. The P-À effect is
considered and the rigorous method (Section 4.4.2.1) is used in this example. The plastic
-91 -
CHAPTER 4 NUMERICAL MODELS
where Z is the length of the cantilever column and ú is the diameter of the longitudinal
bars. The calculation of lateral confinement is performed manually as follows. The lateral
3.41'Au'fr,
""'r/ 3.41x 50.27x360
Í-" =2.16(IlPa\
s.b, 78x (400- l3x 2 -8)
where lo is the area of one leg of transverse reinforcement; s is the spacing of stirrups;
and b, is the height or depth of concrete core of column section enclosed by stirrups
measured from center of stirrups. The effective confinement pressure and the confined
o-Zffitrt- n"-^t
l- p,,
.ft = 0.713x2.16 =1.54(MPa)
where rø, is the ith clear transverse spacing between adjacent longitudinal bars; d"=b,
in this case pcc is the ratio of area of longitudinal reinforcement to the core of section;
The confinement effect of the concrete is allowed for by assigning the concrete
stren$h às f'" or 54MPa in the confined zone as shaded inFig.4.21. The confined zone
is assumed to be the area within the center-line of the perimeter stirrups in Mander's
model. The unconfined concrete strength .f,.,, oÍ 44 MPa is used for the unconfined
-92-
CHAPTER 4. NUMERICAL MODELS
The cycling of the response is controlled by driving the control parameter, i.e. the
6 r,
increment Lon=4x10 isthenappliedtothepreviouscurvature tocalculateanew
point in the force-displacement curve. The calculation of successive new points continues
until the displacement at the top of the cantilever reaches the specific "target" value,
when An, changes sign so that the column is moved in the opposite direction. The
"target" displacements are chosen to match the cyclic amplitudes of the experimental
results.
The numerical results for the test column are given inFigs.4.22 and 4.23. Figure
4.22 shows the moment-curvature relation of the bottom cross-section. The displacement-
force relation of the cantilever column is given in Fig.4.23. The calculation terminates
and cannot finish all the cycles as the experiment did when the strain in the concrete
reaches a value of more than 0.0614 at which time an error occurs in Mander's model.
This error seems always to occur in Mander's model when the strain is extremely high.
However, such high strains are not of practical significance, as can be seen from Fig.4.23
The test results by Watson and Park (1994) are given inFig.4.24. For comparison,
the force H or y coordinates in Fig.4.24 has to be divided by two, because two columns
are loaded simultaneously in the test and the force applied at one cantilever column is
0.5H as shown in Fi9.4.20. The comparison between Fig.4.23 and the test result shows
that the test column is stronger and degrades slower than the numerical one. It suggests
the possibility that the actual materials of the test column were stronger than the material
properties given, and the actual confinement was greater than the calculated one.
-93 -
CHAPTER 4 NUMERICAL MODELS
As mentioned in Section 4.1.2.2.1, the computer program pan also account for
general configurations of confinement such as that shown inFig.4.25, where the confrned
concrete strength is 57.4MPa in the shaded area and the unconfined strength is 44MPa in
the non-shaded area. Basically, the computer program allows for independent confined or
As the response curve is much 'weaker' than Fig.4.23 and the test results, it suggests that
2.5E+08
Ê
E
z
0.0E{ü)
tr
o
E
o
= -25E+æ
-5.0E+08
qnatre(1/lrrl
-94-
CHAPTER 4. NUMERICAL MODELS
2 E+05
1.E+05
Force
at top 0.E+00
(N)
-1 E+05
-2 E+05
-3.E+05
Displacement at top (mm)
-w tæ
ASN.ACENENT.
Lløml
ol lirst tensrcn
longtludinol bor
,.0
llt\w
-ô-6-.J
J
---
Firsl cycte
Second cycle
lJOflAL OtiPLaiftlfíl (Ncllallf fACltR lltñl
llll
-2-6 -{¿l -t.6 .r 9
fiEAt ottPtÆtvfxl oucliltlr fÆton tw)
-9s -
CHAPTER 4 NUMERICAL MODELS
L 25 columns -l
-
l I
FACES
When plates are bolted, instead of glued, to the faces of a column, methods developed in
the previous sections cannot be used, because of the relative movement or slip between
-96 -
CHAPTER 4. NUMERICAL MODELS
the plates and the RC column. For columns with bolted plates, the slip on the interface
causes discontinuity in the strain profiles in a cross-section. In this case, the strain of the
plate cannot be determined directly from the curvature and neutral axis position of an RC
cross-section.
To simulate the response of the composite plated column shown in Fig.3.1, new
mathematical models that consider slip or partial interaction between the RC column and
the plates that are bolted on the tension and compression faces are developed in this
section. Columns with plates bolted on the other two side faces are not covered in this
study.
To calculate the slip on the two interfaces, the longitudinal segments are divided
in such a way that one segment is taken between two adjacent longitudinal bolts, as
shown in Fig.4.27. The first and last segments are taken from top and bottom of the
t
RC column
Plates
Bolts
Beam or footing
-97 -
CHAPTER 4 NUMERICAL MODELS
Fig.4.28, where the superscripts 'Z' and 'R' represent the left-hand-side (LHS) and the
g""¡
stpt
X
¿å.
Fig. 4.28 Typical strain profrle in a cross section with partial interaction
There are l0 unknowns at a cross-section. Six of them are related to the cross-
a The slip strains at the LHS rti,o(i) and RHS t:ir\, respectively;
O The axial forces on the LHS and RHS plates: N';,(i) and N'),,(i), respectively
The other 4 unknowns are related to the movement of the bolts at each cross-section. It is
assumed that the longitudinal slip at the interface between the plate and the RC column is
-98-
CHAPTER 4. NUMERICAL MODELS
equal to the lateral movement of the bolt (when a gap exists between a bolt and the plate,
it is not equal. See Section 4.5.4) as shown inFig.4.29. The four unknowns are:
. The LHS & RHS slips, s/' and s/ì, respectively; and
o The lateral (shear) forces on the LHS & RHS bolts, Fj'(t) and pl'(¡) , respectively.
Fb p ate
RC column
o.
v) +
The 10 conditìons that can be used to solve for the above unknowns are:
a The bending moment M, must satisfy the overall equilibrium condition given by
8q.4.57 or 4.58.
a The relations between the slip strains and slips are given by
"/'
=
"1,
+ e';,n1i + l)'a¿¡*r Ø.65)
",n
=
"Í,
+ ellrçt+ l).^¿i*r G.66)
a The relation between the shear force applied to the bolt and the amount of slip, as
shown inFig.4.29, is given by a shear force versus deformation model (based on test
-99 -
CHAPTER 4 NUMERICAL MODELS
p| (¡): f ß) Ø.67)
o The relation between the shear force of bolt and the axial force on the plate is given
by
. The relation between the strain distribution of the plate and axial force in the plate
where e';,lt), t'),{i) ur" the strains on the interface for the LHS and RHS plates, as
shown inFig.4.28.
From the above analysis, it can be seen that there are'n' cross-sections, hence n x
l0 unknowns in total for the whole cantilever column. There are also lOn independent
these equations directly is not practical when non-linearity is involved. Trial and error
methods, i.e. iterative procedures, are used in this study to solve the problem.
If the slip strains or forces of the plate in a cross-section are known, the analysis of the
cross-section will be no different than that given in Section 4.1. By assuming the forces
-100-
CHAPTER 4. NUMERICAL MODELS
in both the tension and compression plates at the bottom cross-section (control cross-
section), the forces in the plates at other cross-sections can be calculated one by one using
equations 4.65-4.72, ending with the forces in the plates at the top of the column. The
forces in the top of the plates should be zero for both the LHS and RHS plates. If this
"boundary" condition is not satisfied, then the trial forces at the bottom cross-section
were incorrect and another pair of forces is tried until a solution is found. Based on this
I . Set the initial curvature K,, = L.K, for the bottom cross-section.
2. Guess the axial forces Wt;,,(r)&.N'J,,(n) in the LHS and RHS plates at the bottom
cross-section.
3. Find the strain distributions and et;,(n)Ue')(n) in the LHS and RHS plate based on
4. Calculate the slips s';Us') for both the LHS and RHS plates at the bottom of the
Fig.3.l(b). The slip or displacement of the spring can be calculated from the force on
the spring, i.e.NI;,,(n),N'ì,,@). Assuming the plate cannot penetrate through the
concrete, the slip is set to zero when a downward slip going into the concrete is
calculated.
5. For the given curvature r,, and plate forces N';,,(n),N'),(n), adjust the strain profile
in the RC section as shown by the dotted lines in Fig.4.28 or search for the neutral
axis depth x to satisfy the condition of constant axial force N in the cross-section.
-l0l-
CHAPTER 4 NUMERICAL MODELS
6. Calculate the momenf M,, of the bottom cross-section based on the stress and force
ignoring the P-A effect. Set the cross-section number i--n for calculations in the
following steps.
8. For both the LHS and RHS plates, calculate slip strain t.,, of the cross-section i
based on E,u(i):e,,,(i)-s.,(i), as illustrated in Fig.4.28. Then calculate the slip at
9. Calculate the shear forces Fu(i - 1) on both the LHS and RHS bolts located at section
i-l based on Eqs.4.67 and 4.68, i.e. Fu(i -l):,f(s,_,). A bi-linear slip versus force
relation as discussed in Section 4.1 .4 is used at the moment in the computer model.
10. The axial forces of the LHS and RHS plates in cross-section i-l are calculated by
ll. Based on N';,,(i-l)&.Nfi,G-l), the moment M,_, and the total axial force N for
cross-section i-1, search for the curvature rc,_, and neutral axis x,_, .
t;,(i
12. Calculate strain distribution and e -1) & e'i'"{i - 1) in the plates for cross-section i- 1
-102-
CHAPTER 4. NUMERICAL MODELS
la. (a) If the axial forces of the plates N';,,(1)&.N';,,() at cross-section I are not zero, the
trial axial forces Nt;,,{n)&N';,@) are not correct. Adjust U';,,{n)&N';,(n). Go back
(b)lf N;;,(l)&,N';,(l) are both zero, the trial forces N';,,(n)&.N';,,(n) are correct.
Calculate the lateral displacement A of the column based on the curvatures along the
column. The force at the top is given by F=Mnl¿. This F&A pair gives a point in
15. Record the stress-strain history for all the elements of the concrete, plates, bolts and
springs for the calculation of the next point of the F-A chart.
16. rc, : K n t Lrc,,, go back to step 2 until the required F-A chart is produced.
While the P-A effect is not considered in the above calculations, both the rigorous
and simplified methods as given in Sections 4.4.2.1 and 4.4.2.2 can be incorporated into
the above process to include the P-A effect. However, the rigorous method adds another
loop to the already very intensive numerical calculations. In view of the reasonable
accuracy of the simplifìed method, it is adopted to calculate the P-A effect for a plated
column with partial interaction. This means that the final results from the above process,
i.e. the force-displacement relation at the top of the column, are simply transformed by
- 103 -
CHAPTER 4 NUMERICAL MODELS
discussed in Section 4.5.2. Mathematically speaking this process is a root finding process
N'),,{r)) = o
{I w'n,,{t) = .f,(N';,,(n),
,__, -j,, .. - (4.73)
f
rujl,rll = f,(N';,,,(n), Nff,(r)) = o
The target functions are N';,,(1):0, N';,,(1) = 0 . The two variables are N';,,(n) B. Nt;t,@) .
In this work, the Newton's method for non-linear system of equations with 2 variables
Longitudinal gaps between the bolts and plates, as shown in Fig.4.30, are unavoidable for
apractical column. This gap will affect the interaction between the plate and the bolt and,
RC column
, gap
width:w*
-104-
CHAPTER 4. NUMERICAL MODELS
When gaps exist, the slip s at a cross-section between the plate and the RC
column is no longer equal to the lateral deformation of the bolt. The lateral deformation
where s,. is the plastic or residual deformation of the bolt which depends on the loading
history; and w, is the gap width. This s¿ is the true lateral deformation of the bolt. The
bolt force can then be obtained with the same model as discussed in Section 4.1.4 using s6
ln F;q.4.74 when ls-",1< wr, that is before the gap is closed, the bolt force is
always zero as given by the above model. This means that the bolt force is a constant
over a range of slip values. This kind of constant value will usually cause problems in
various searching procedures of the program. In order to avoid this kind of difficulty, the
where Kuo is an arbitrary but very small value. By providing this very small variation of
force in the bolt before the gap is closed up, numerical searching procedures are well
behaved
- 105 -
CHAPTER 4 NUMERICAL MODELS
4.5.5 NumericalExamples
Numerical examples for a plated column are shown in Fig.4.3l. The column studied is a
l2l8mm long cantilever column with a cross-section of 200x200. Longitudinal bars are
4Yl6 with concrete cover 35.7mm from the center of the bar to the face of the column.
except l't bolt 200mm from bottom of the column with a total number of 2 x 8 on
confinement effect is calculated based on the method discussed in Section 4.4.3. The
realistic response.
The first case indicated with "no plate" in Fig.4.3l is an RC column without
plating. The second case is the plated column without gaps around the bolts. The third
case is the plated column with a l.5mm gap around the bolts. The last case is a full
- 106 -
CHAPTER 4. NUMERICAL MODELS
interaction plated column where the steel plate is glued to the face of the RC column or
the bolt stiffness is infinity. Detailed discussions on these response curves are given in
Chapter 5.
60
no slip
50
z no ga
5oo
o)
o
L
m
o30
(õ
L
O20 nop
o
-J
0
0
10 20 70 80 90
ois'ftaceriÎent aitop 1'åm¡
4.6 SUMMARY
The numerical models to simulate the response of the plated RC columns are developed
in this chapter. This numerical model can simulate the response of RC columns under
monotonic load both on the ascending part and softening branches, as well as under
quasi-static cyclic loading. Each part of the numerical modeling and computer
programming has been carefully tested and verified and compared to theoretical results
where available. Reasonable responses for RC columns have been obtained by the
computer program "PLTCOL" as compared to test results which will be discussed further
in Chapter 12. This numerical model will be used to analyze the response of plated RC
columns in the next chapter in order to study the behavior of these columns.
-107-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
numerical simulations of column responses. The responses of the plated columns are
In order to gauge the effects of composite partial-interaction plating, the RC column with
N:360kN
J' L: 200
200 rypi
rl¡
200
35.7
7ù
200
Le :200
!
6mm steel plate
Yl6 bar R6@r00
- 108 -
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
Main reinforcement bars:f,y: 547 MPa, -8, :200 GPa, E¡: 600MPa, ,Ro:20.0,
ar:19.0, ar:O.3;
The confinement to the concrete core due to the stirrups and the P-A effect are not
considered in the present analyses (results from Figs.5.2 to Fig.5.5) in order to isolate the
effect of the composite action. These effects, however, will be incorporated later in
Section 5.2 in the further study.
The responses of columns with different plating details are given in Fig.5.2. The
response curve marked 'original unplated column' is for the RC column without plating,
which is used as a benchmark for direct comparison. The curve indicated 'plate bolted I
side' is for the plated column shown in Fig.5.l. The case with plating on both the tension
and compression sides is also studied, which gives almost identical results to the 'plate
bolted I side'case when the tension spring stiffness shown in Fig.3.l(b) is small. This
both sides plated case verified the assumption that Fig.5.l is equivalent to that in
Fig.3.l(a) when the plate can be easily opened up at the corner. The result for'plates
glued 2 sides' is for the case where both sides are bonded with full interaction plates (no
slip between concrete and plate) and without the uplift facility illustrated in Fig.3.1(c),
which is equivalent to adding extra fully anchored reinforcing bars. The case 'plate bolted
- 109-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
I side plus wrapping' is for the 'plate bolted I side' column plus an assumed lateral
increase the concrete strength from 40MPato a confined concrete strength of f",:48
MPa which can be achieved theoretically with FRP wrapping (Mirmiran et al. 1998).
Legend
plate glued 2 sides
100 À - conc 1ìrst cracks
20
original unplated column
0
0 10 20 30 40
lateral displacement at top (mm)
^
Critical points are highlighted in Fig.5.2 by various marks that indicate different
deformation stages as shown in the legend. More detailed definitions of these marks are
given in the Notation. All these critical points or stages refer to the cross-section at the
It can be seen in Fig.5.2 that adding extra reinforcement, that is the case of 'plate
glued 2 sides', can substantially increase the strength but at a considerable loss of
ductility which may not be beneficial in seismic retrofitting. In contrast, the system
represented by the 'plate bolted 1 side' case in Fig.5.2 substantially increases the ductility
order not to increase loads on the foundations. It is also wofth noting that combining
- 1t0-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
wrapping with partial-interaction plating, as in the case of 'plate bolted I side plus
Figure 5.3 shows the variation of axial force on the concrete alone at the bottom
cross-section, which excludes the axial forces on the reinforcement bars and the plate. It
can be seen that the axial force on the concrete is reduced for the 'plate bolted I side'
case compared to the 'unplated column', resulting in the increased deformation capacity.
For the case of the'plate glued 2 sides', the plates reduce the axial force on the concrete
when the lateral displacement is small. However, the axial force on the concrete increases
rapidly with curvature because of the increased tensile force in the tension plate. This
additional axial force causes the concrete to crush earlier, as indicated by the mark '0' in
'i
(ú 0.0 5.0 10 0 '1
5.0 20 0 25 0 30.0 35
capacity, the column case with the plate bolted only to the compression face is studied in
Fig.5.4 with different axial loads N, where N, = f"oAr and A, is the gross cross-
sectional area of the RC section. The responses of the plated columns are shown with
dark lines. For comparison, the responses of the original column without plating under
- lll-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
the same axial loads are also shown with light lines. It can be seen that partial-interaction
plating increases the ductility in all cases. In addition, the detrimental effect of increasing
axial load on the ductility/deformation capacity of the columns, which may be reflected
by the point '0' when the concrete crushes on the compression face, is also clearly shown
in Fig.5.4.
70000 N=0.4Nc
60000 l N=0 3Nc
z N=0.225Nc
ll. 50000 N=0.1 SNc
,
o
IJ 40000
o c
(tr
30000 ' N-0.0625Nc
(¡) 20000
N=0.1SNc
-g I 0000
N
N=0.225Nc
N=0.3Nc
0
0 10 20 30 40 50
lateral displacement at top (mm)
^
Figure 5.5 gives the curvature distributions along the length of the column at the various
critical points or stages for the case of 'plate bolted I side'. It can be seen that the
curvature increase concentrates in the plastic hinge region from yielding of the tensile
reinforcement, indicated by'+', to crushing of the concrete, indicated by'0', whilst the
curvature above the plastic hinge remains relatively unchanged (in fact it reduces a little
bit). That is to say the plastic deformation of the column can be considered to occur
mostly in the plastic hinge region, which is generally accepted in the literature. This
Chapter 8.
-112-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
1.5E-04
conc. crushing
E
E
1.0E-04
o tension reinf. yielding
G
5 0E-05 conc. cracking
È
o=
0 0 E+00
200 400 600 800 1000 1200
Distance from top (mm)
The advantages of partial interaction plating are clearly seen in the previous section.
More numerical simulations for columns with different plating, such as different plate
thickness, bolt stiffness, yield strength, gap etc., are presented in this section to further
study the plating effects. The column details are almost identical to those of the previous
section with only slight changes to match the experimentally tested columns: the column
length is L:l2l8mm; the first bolt is 200mm away from the bottom of the column; the
spacing of the other bolts are l00mm c/c with a total number of 16 bolts at 8 cross-
sections (2 bolts at a cross-section, only plated on the compression face). This bolt
configuration is identical for all the case studies in this section. However, the stiffness of
bolts can be changed for different cases. Confinement due to the stirrups (R6@.100) is
Chapter 4, which gives the confined concrete strength of f,,:47}l{Pa that applies to the
concrete core enclosed by the centre line of stirrups. The P-A effect (by simplified
method) is also included in these calculations.
-l13-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
Figure 5.6 shows the responses of plated columns with different plate thicknesses
but with a constant bolt stiffness (K¡:23kN/mm). The top curve, indicated by '30mm
plated column with infinite stiffness', is the result for a column plated with a 30mm thick
plate having an elastic modulus and yield strength increased by 1000 times the actual
values. With these values, the plate is essentially a rigid plate that represents an upper
bound.
0
0 20 40 60 80 100
Four more development stages are shown in Fig.5.6 by the four new marks that
are defined in legend. The more detailed definitions of all the marks can be found in the
Notation. Not every mark appears on every curve in Fig.5.6, because they do not occur in
some cases. For example, the bolts in the 3mm and 6mm plated columns do not yield
-114-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
because the plate yields first. Similarly, the plate does not fully yield for the cases of
l2mm and thicker plated columns because yielding of all bolts occurs first. Fig.5.7 gives
the results for 6mm plated columns with a range of bolt stiffnesses. For simplicity, only
points '*','L' and 'a' are given in Fig.5.7. The plate does not yield for the case of
60 Ku: oo
fpv:K u: co
50
2
5¿o
o
o
€30 Ko:5
Ka:46
õ
b20
(! Ka:23 u:0.75
0
K¡:11.5
Kr,:O(kN/
0
0 20 40 60 80 100
Legend:
+ - tension reinf yields t - pìate tull yields a - comp. reinf yields in tension
bolt stiffness in the ascending branches before yielding of the tensile reinforcement.
The yield strength of the column (at point '*' where the tension reinforcement first
yields) also increases with increases in either plate thickness andlor bolt stiffness. The
yield points are very closeJhe peak lateral resistance of the columns.
.l o,
- 115 -
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
2. There is a slight reduction in the lateral stiffness for all the cases when the concrete
cracks in the tension zone, as indicated by the point 'A' in the curves.
indicated by the point 'x', is delayed relative to the yielding of the tension
reinforcement (point '+') as the plate thickness increases. For example, point 'x'
occurs before point '+' for curves with r0 and 3mm, whilst it occurs after point '*'
for cases with r6mm and above. This indicates that the compressive resistance of the
column increases when the plate thickness increases. Consistently, the increase in
plate thickness also delays the onset of concrete crushing at the compression face as
4. The plating reduces the steepness of the post-peak softening branch, with thicker
plates giving less steep slopes, as shown in Fig.5.6. Similarly, an increase in bolt
stiffness also reduces the slope of the softening branch up to the point'A'where the
whole plate section yields, as shown in Fig.5.7. However, Fig.5.7 also shows that the
bolt stiffness has no effect on the softening slope once the yielding of the whole plate
occurs. This is logical since the bolt stiffness cannot generate further increases in the
compressive resistance from the plating system if the plate has fully yielded.
Generally, in the post-yield range of response curve between point '*' and the yield
point of the plating system (full yielding point of either plate 'A' or bolts 'o'), the
5. The plating system improves the integrity (capacity to keep stability) of the column.
As seen from Fig.5.6, the yielding of the compression reinforcement, as marked with
-n6-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
'X', and the crushing of concrete in the vicinity of compression bar, as indicated by
the point '-', do not occur up to the end of the chart for the 6mm and thicker plated
columns. This signifies a better integrity in the compression zone compared to the
original (benchmark) un-plated column. The points of '-' and 'X' for the 3mm-plated
column also occur much later than that for the benchmark column.
6. The plating system improves the displacement ductility of the column. The
displacement ductility factor here is defined as
p = L,,lL, (5.1)
where A, is the yield displacement (the point where the tensile reinforcement first
yields), as indicated by the point '+', and 4,, is the ultimate lateral displacement at
the point where the lateral resistance force on the softening branch equals 80% of the
lateral force at point '+'. The ductility factors for the curves in Figs.5.6 and 5.7 are
calculated and shown in Table 5.1.
Fig.5.6 / (mm) 0 J 6 12 30 30
(e6mm)
for(MPa) 250 250 250 250 250 250 250000
From Table 5.1 it can be seen that the plating generally improves the ductility.
Holever, increasing plate thickness or bolt stiffness does not always increase the
-117-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
ductility of the column. Generally, a response curve has a larger ductility factor when it
has a smaller yield strength as well as a larger plateau, or more accurately, a less steep
The steepness of the softening branch is an important factor affecting ductility. The
softening slope is largely decided by the P-A effect. Without the P-A effect, the response
curve of a column has a less steep softening branch, as shown in Fig.5.8. As discussed in
Section 4.4.2.2, the P-Â effect causes a steepening of the softening slope by an additional
slope that satisfies O:N/L, as shown in Fig.5.8, where N is the axial load and L is the
cantilever length.
60
withoilt P-^ effect
50
strength
e
^40
z
I
stiffened
Therefore, columns with a larger axial load N have alarger 0, and hence, a steeper
softening slope. So do columns with a shorter length Z. This latter relationship with
- ll8-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
regard to the length of the column seems illogical according to the engineering common
sense that the P-A effect is more prominent for a longer member. The shorter columns do
have a smaller lateral displacement response than that for longer columns. However, the
value of a response is different from the slope of the response curve. For any given
additional lateral displacement beyond the peak of the curve, the shorter columns have a
larger decrease in lateral strength than the longer columns, leading to a steeper softening
90
BO
original column L=609
70
0
)o 30 40 60 100
As seen in Fig.5.8, the softening branch from the point '+' to the point 'a' is less steep
than the curve after the point'a'. This less steep part of the curve, which extends from
yielding of the tension reinforcement (point '+') to yielding of the plating system, as
defined by either full yielding of the plate (point'A') or full yielding of the bolts (point
- il9-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
'o'), is named "strength stiffened region" in this work. It is this part of curve that
t
For convenience of discussion, Eq.4.59 is reprod uce below
Jt
F=(M,-N.L)lL (s.2)
where,F is the lateral force at the top of the column; M,, is the resistant moment from the
bottom cross-sectioni Ao is the lateral displacement at the top. In the ascending part of
Fig.5.8 before the yielding point'+', the resistant moment M, keeps increasing, leading
to the monotonic increase of tr.. After yielding of the column, if the resistantmoment M,,
remains constant or reduces, F given byEq.5.2 will decrease as and because the column
decrease after yielding (point'+'), the resistant moment M,, musf continue to increase. If
the increase in the resistant moment M,, is sufficient to counter balance the increase in
For an unplated RC column, the increase in moment resistance due to the strain
hardening of the tension reinforcing bars is limited. Therefore, adequate increase in M,,
is not possible unless the axial load is very small, in which case the required increase in
M,, to balance N.Ao is also small. However, it is possible to gain an adequate increase
in M ,, for a plated column even with a large axial load, as illustrated by Fig.5.l0
-120-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
7.E 4O7
Ê 6,E 1O7
6mm plated col. lG=13¡¡7tt
zE 5.E 1O7
E
o 4.E tO7
originalcolumn
o 3.E tO7
.Ct
r!
2ErO7
tr
o
1 E1O7
E
o
OEÐO
= 0 20 40 60 B0 100 120
Based on the force diagram shown inFig.3.2, the resisting moment of the cross-
section is given by
M ="".N",+e,.N.,, (5.3)
where e" and et ate eccentricities of N.. and N,,, respectively, with respect to the
When the tensile reinforcement yields, N-, can be considered as constant, hence based on
Eq.5.4 N", ir also a constant. Therefore, the only variable that changes in Eq.5.3 is e.,
which means that any increase in the resisting moment M can only come from the
increase in e". This increase in the eccentricity e" of the compressive resultant is due to
the transfer of axial force from the RC column to the plate as shown in Fig.5.l l.
-121 -
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
4.E+05
ate
3.Ë+tls
z
q)
g
o 2.E+05
't
rú
1.E+05
RC
0 E+00
0 20 40 60 B0 100 120
which also has an effect in increasing the moment resistance. However, the contribution
from strain hardening of the tension bars is small compared to the effect of the lever arm
increase.
closely related to the stiffness of the plating system, i.e. the stiffness of the plate and bolt.
Increasing the stiffness of the plating system reduces the softening steepness of the
stren$h-stiffened part, which can be seen from Figs.5.6 and 5.7. This is because the
increase in the resisting moment M, of the cross-section due to the strength stiffening,
i.e. the transfer of axial load from the RC column to the plate, is faster for stiffer plating
systems. lf M, increases as fast as N.Ao does, then M,-N.A', and henceF, remains
constant, leading to a zero softening slope. Therefore, in order to get a small softening
-122-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
steepness, the plating system must be able to increase the moment resistance of the
bottom cross-section at a similar speed as the P-A effect increases the additional moment.
Once tull yielding of either the plate or all bolts occurs, defrned by the points 'a'
and 'o' respectively in Figs.5.6 and 5.7, no further extension of the strength stiffened
region can take place, because no more transfer of axial load happens. Therefore,
increasing the strength of the plate or bolts, by using thicker plates, stronger bolts or
larger numbers of bolts, can extend the strength stiffening part. For example, the plate
yielding point 'a' occurs much later for a 6mm plated column compared to the 3mm
plated column in Fig.5.6. Plate yielding does not occur at all for the l2mm or thicker
plated columns in Fig.5.6, resulting in the yielding of all bolts at point 'o'.
stronger plating system. The reason is given below. For a plating system that is
suffrciently strong so that the plate and bolts will not fully yield under any large
displacement, the strength stiffening starts, as before, with the yielding of the tensile
reinforcement. However, the continuous increase in curvature at the bottom section will
move the neutral axis very close towards the side of the plate if the plating system is
sufficiently strong. When the compression zone is so small that the neutral axis moves to
the region between the compression reinforcement and the plate, the longitudinal
reinforcement at the compression side will actually be loaded in tension. In this case, N-"
becomes a negative value and N.o," becomes very small. This further increases the load
-123-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
where e,,,,, and ep are the eccentricities of N.u,. and N¡,, respectively, with respect to
the centroid of the cross-section. If the compression bar yields in tension and the strain
hardening is ignored, the moment terms due to N," and N", cancel each other, as the
M =€rurr.Nror"*e,r.Nnt, (5.6)
If the axial force in the concrete is ignored as the compression zone is small, substituting
M * .(A,, + A)f
N e p. N pt,
+ [¡r*¿ (s.7)
At this time, the maximum moment resistance of the cross-section has been achieved. No
further increase in the resistant moment, or strength stiffening, can be made regardless of
how strong the plating system is. This case gives the upper limit for strength stiffening.
An example of this case is shown by the top curve in Fig.5.7, where the resistant moment
of the bottom cross-section from the computer simulation is 82.7 kNm at the point 't'
where the compression bar just yields in tension. This moment value is very close to the
When the neutral axis moves to the compression concrete cover arca, the entire
applied axial load N plus the tensile forces in all the reinforcement, which is sometimes
much bigger than the axial load l/ itself, acts on a very small compression zone. This
highly stressed narrow compression zone raises concern over the stability of this region.
However, experimental work (see Section 11.2) showed that no excessive distress
-124-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
occurred in the test columns in the compression zone and the integrity of the column
From Table 5.1, it can be seen that an increase in the plating stiffness, i.e. stiffness of the
bolt and/or the plate, does not always increase the ductility of the column. For example,
the displacement ductility reduces from 4.5 to 4.2 when the bolt stiffness K6 increases
from ll.5kN/mm to 23kN/mm. The reason is that a stiffer plating system will have a
stiffer ascending response branch and greater yield strength, i.e. a greater "y" co-ordinate
value at point '+'. The greater yield strength of the column causes a reduction of the
ductility factor as calculated at a point corresponding to 80% of the yield strength in the
softening branch. A stiffer plating system may also result in an earlier yielding point of
the plating system, i.e. a smallet "x" co-ordinate value at point'O' or'o', as shown in
Figs.5.6&5.7. From these points of view, a plating system, that mobilises the compressive
resistance only after onset of yielding of the tensile reinforcement, produces a larger
ductility factor.
To verify this, a plating system with gaps between the bolts and the plate, as
shown inFig.5.l2, is analysed. The numerical results for the 6mm plated columns with
gaps of varying sizes between the plate and the bolts (bottom gap is set to zero in these
cases) are given in Fig.5.l3. From a ductility point of view, the advantage of the plating
system with gaps is obvious from these numerical results. The ductility factors for the
cases with lmm and 2mm gap are 4.7 and 5.0 respectively, as compared to 4.2 for the
case without gap. It can be seen from Fig.5.l3 that the initial responses of the plated
-125-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
columns with gaps are "exactly" the same as that of the un-plated column before the gap
closes up at the bifurcation point where the response of the plated column splits from the
gap
around
bolt
RC column
bottorn gap
50
gap
45
zll 40
35
6)
o 30 2mm gap
o
no plate
õ 20
15
(¡,
IE 10
0
10 20 30 80
Fig. 5.13 Effect of gap for the 6mm plated column, K¿:23kN/mm
Theoretically, there are both an axial force and a flexural moment in a plate cross-
section. The flexural stiffness of the plate will also contribute to the moment resistance of
the whole plated cross-section even when the axial force in the plate is zero (before the
gap closes up). However, the flexural stiffness of the plate is negligible compared to the
-126-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
RC column. This explains why the plating system has virtually no effect on the RC
designing the width of the gap. Ideally, a gap would close up when the column just
yields, as shown by the case with the l.5mm gap in Fig,5.l3. The slope of the response
curve after the bifurcation point can also be designed by properly choosing the rigidity of
the plate and/or the stiffness of the bolts as discussed in the previous section, which is
45
z^35
.Y
1.smm gap,
^^
o
o".
o
_20
(! no gap
o15
(!
Jro
5
0
0 10 20 30 40 50 60 70 80 90 100
It has been shown in the previous section that "strength stiffening" terminates
when the plating system yields, i.e. yielding of either the whole plate section or all the
bolts. Increasing the yield strength of the plating system can increase the ductility of the
system. As mentioned earlier in this section, increasing the stiffness of the plating system
may not increase the ductility. However, it has also been shown in Section 5.2.2 and
Fig.5.l4 that a smaller softening steepness in the strength stiffened region can be
-127-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
achieved by a stiffer plating system. Therefore, a proper balance of the strength and
stiffness of the plating system is important. An idealised plating system shall be: (l) weak
or not effective before the onset of column yielding point '+'; (2) adequately stiff after
yielding point to provide a small softening steepness; and (3) sufficiently strong (tough)
to delay the yielding of the plating system in order to achieve the maximum extent of the
gap with infinite yield strength but normal stiffness for both the plate and bolts, as shown
in Fig.5.15. The example further demonstrates that a higher yield strength for the plate
and bolts can increase the ductility of plated columns. This result suggests that FRP
materials may have advantages over steel for both plate and bolt. The strength stiffening
in Fig.5.l5 stops at tensile yielding of the compression bars shown by the point'.)'.
40
K5-46kN/mm, fpy-æ, For- -
zta
v30
(¡)
o t.5 lnm gap, 3kN/mm, for:250MPa
o
!zo
G no
(¡)
.E
J10
0
0 20 40 60 80 '100 120 140
resistance in the tension face can be furlher increased so that the resistant moment of the
-128-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
bottom cross-section further increases. Tension plating on the tension face, as shown in
Fig.5.l6, with the tension gap closing just before point '..l' can be used to serve this
purpose. However, this kind of system may not be practical in the sense that the
compressive force may be too high in the compression plate. On the other hand, the
tensile strain of the reinforcement bar may also be too high in the tension side at that
stage, causing fracture of the reinforcement bar. The tensile strains at the end of the chart
for the top curve in Fig.5.l5 are 0.0816 and 0.00387 for the reinforcement bars at the
Tension
gap
Compression
gap
Longitudinal slip between the RC column and the plate is very important in composite
structures as it reflects the degree of composite interaction between elements that are
connected by bolts/shear connectors, which in turn affects the overall stress distributions
in the members. Furthermore, mechanical shear connectors have only a limited slip
capacity. Excessive slip will cause the fracture of the bolts/connectors. Therefore, slip
-t29-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
and its distribution are very important considerations in the study of composite structures
Figure 5.17 gives the slip distributions along the length for the'plate bolted I
side' column studied in Section 5.1 at the three different stages of: concrete cracking at
the stress of f"¡, as shown by the mark 'Â'; yielding of the tension reinforcement shown
by '-|'; and crushing of the concrete at the compressive face indicated by '0'. The slip
distributions at the first and second stages shown by 'A' and '+' are similar to the slip
distribution given by the classic linear theory (Newmark et al. 1951 and 1952) of
composite beams which gives a zero slip at the maximum moment position, a maximum
slip at the zero moment position and a convex shaped distribution. The third distribution
indicated by '0' looks quite different from the classic distribution. The figure shows that
the slip near the bottom part, which is around the plastic hinge region, increases faster
than that of the other parts of the column. It is even possible for the slip of the first bolt to
become larger than that of the remaining bolts. Fig.5.18 gives an example for the case
where the spacing of bolts is half that for the case represented in Fig.5.l7 and K¿ is
't4
1.2
1.0 1.042
ç
Ë oo
.g 06
Ø
o4
o2
00
0 200 400 600 800 1000 1200
-130-
CHAPTER 5 NUMERICAL STUDIES OF PLATED RC COLUMNS
08
u.t
0.6
05
E
E 04
:o o3
U)
o.2
01
00
0 200 400 600 800 1000 1200
Fig. 5.18 Slip distributions when maximum slip occurs at the bottom
It will be shown in Section 8.2.1 that the slip of the first bolt at the top of the
plastic hinge is only determined by the deformation of the plastic hinge (see Eq.8.5)
which is largely determined by the total rotation of the hinge or the overall displacement
of the column. The slips of the other bolts are largely affected by the stiffness and total
number of bolts, i.e. the total longitudinal shear resistance of the bolts, which will be seen
from the linear elastic study in Sections 6.3-6.4. Therefore, for columns with strong shear
connections on the interface, i.e. stiff and/or large number of bolts, and when large
deformations occur in the plastic hinge, it is possible that the largest slip occurs at the
bottom where the maximum moment occurs rather than at the top of the column where
the moment is zero. This conclusion has practical implications for the design of
composite members. In the literature, the classical slip distribution developed from linear
elastic theory, by which the maximum slip is considered to occur at the position where
the bending moment is zero, is generally adopted to guide the design of shear connectors
(Oehlers and Bradford 1995). From the above study, it can be seen that this conclusion
- 131-
CHAPTER 5 NUMERICAL STUDìES OF PLATED RC COLUMNS
may not be applicable for a composite beam/column loaded to the plastic deformation
stage. Further studies on the slip distribution will be conducted in Section 6.4.
From Figs.5.l7 and 5.18 it can be seen that the slip distributions may be
considered to be uniform along the length at the ultimate loading stage. This observation
will be used to simpliff the plating design procedure presented in Section 8.3.
-132-
CHAPTER 6 MATHEMATICAL STUDIES
The advantage of partial interaction plating for RC columns has been shown in the
are performed. These analytical studies help to better understand the fundamental
behaviour of plated columns. Some results from these analytical studies will also be
-),'
The mathematical studies of this section are based on linear elastic theory, which
is applicable to serviceability t,-,, ,ruffiis. In order to exrend the linear theory
from serviceability analysis to ultimate limit state analysis, where large in-elastic
deformation occurs, an elastic analysis plus plastic hinge model is introduced in Section
6.1. In Section 6.2,the linear elastic theory of composite members (including beams and
columns) is developed which extends the classic linear elastic theory of composite
beams. It is shown in Section 6.3 that the response of composite members is governed by
just a few composite parameters regardless of the large number of variables involved,
which is a new concept first introduced from this work. The slip distributions are also
or more correctly, within the linear elastic limit for both material and geometrical
-133-
CHAPTER 6 MATHEMATICAL STUDIES
conditions. Strictly speaking, it cannot be applied when any part of the structure yields.
However, when plastic deformation occurs, most of the plastic deformation of the
member is concentrated in a zone where maximum moment occurs, namely the plastic
hinge zone as discussed in Section 4.2.3. Therefore, the region of the column above the
A" N
\\, F
t.
before
loading
Plastic hinge
*-' -
A
Plast
For the elastic member above the plastic hinge, the slip at the imaginary support
or the interface of the elastic and the plastic part, as shown in Fig.6.l(b), will not be zero.
Therefore a non-zero slip boundary condition, as discussed in Section 6.2.2, must be used
in the linear theory. This boundary slip .çp can be calculated by the plastic hinge analysis
introduced in Section 8.2. Therefore, by using the model shown in Fig.6.l, the difficult
-134-
CHAPTER 6 MATHEMATICAL STUDIES
linear-elastic analysis of the member with a shofter length L" and a given boundary slip
Classic linear elastic partial-interaction theory for composite steel and concrete beams
was f,rrst developed by Newmark et al (1951, 1952) and more recently extended to allow
for non-linearity by Johnson and Molenstra (1991) and Burnet and Oehlers (2001). In this
include axial load so that it can be applied to plated columns as well as prestressed RC
composite beams. Furthermore, a non-zero slip boundary condition at the fixed end
extended to encompass beams and columns at the ultimate limit state where large
deformations occur at the plastic hinge, using the model of Section 6.1 .
The basic geometric model underconsideration is shown in Fig.6.2. The origin of the co-
the top of the member. The x-axis is in the longitudinal direction of the column. A typical
portion from the top of the column to a cross-section that is x distance away from the
L For generality, a constant moment Mo at the origin and a distributed load p(x) along
-135-
CHAPTER 6 MATHEMATICAL STUDIES
the length are applied. ln the free body diagram of Fig.6.2(a), the normal (y direction)
stress on the interface is not shown as it has no effect in the following derivations.
p(x)
centroid of element
Element 2 \
s'tp Je2-
F Element I \c'
interface
Mo
X
A
N 1-
centroid of element I
Section 3.3 plus linear assumption. For convenience of reference, they are listed as
follows:
(3) Bernoulli's principle that plane sections remain plane applies to individual elements,
(4) The shear connectors between the two elements are continuous and uniformly
(5) No transverse separation occurs on the contact interface, therefore the curvature is the
-136-
CHAPTER 6 MATHEMATICAL STUDIES
Nr-F,0,=0 (6.1)
Mt.ht Nl
c-
çl- (6.s)
(Er), (EA),
o)
M..h. N"
(6.6)
-
(Er), (EA),
M.t_z M"
(6.7)
(Er), (Er),
Fh q'L,
r)- (6.8)
Kb Kh
-137-
CHAPTER 6 MATHEMATICAL STUDIES
where Fo is the shear force applied to the shear connectors at a given cross-section that
has a stiffncss of Kr1' L, is thc longitudinal spacing of thc shcar conncctors; and 4 is thc
longitudinal shear force per unit length, or shear flow, which is given by
(6.e)
Hence
L- dF,,,
Jt=- (6.10)
Kb dx
ds
tun I t2 (6.r r)
dx
ds _ L, .d2F"o, (6.12)
dx Kb dx2
By refening to Eqs.6. l, 6.2, 6.4-6.7 , the slip strain in Eq.6.l I can be transformed to
(6. r 3)
in which
tlt
_=__L_ (6.14)
EA (EA), (EA),
Zu =(Er)t+(Er), (6.15)
and
-r38-
CHAPTER 6 MATHEMATICAL STUDIES
where EI is (can be proved to be ) the flexural rigidity of the composite member with
full interaction. However, EA is not the axial rigidity of the composite member.
L
ao (6. r 8)
,ç
EI (6.1e)
ar
EA.>EI
a2 _(h+hr) (6.20)
Znt
and
a- (6.21)
(EA)'
Eq.6.l7 simplifies to
d'F"r,
"r-t#-at.F,n,+ar.M(x)rar'N =0
(6.22)
This differential equation can be solved for a given applied moment distribution of M(x).
The slip then can be obtained from Eq.6.10. To obtain the deflection of the member, the
d2v M,
(6.23)
dx2
-,' -
-t-
(EI),
-139-
CHAPTER 6 MATHEMATICAL STUDIES
ootþ _ r!)
+
dx'
=oo.
dx- EI
* ao. a,. N (6.2s)
where
(6.26)
The solution for the cantilever column shown in Fig.6.3 is just a special case of Section
6.2.1.
L
X
- 140-
CHAPTER 6 MATHEMATICAL STUDIES
In this case, M (x) = F ' x and the general solution of 8q.6.22 is given by
where the constants ct and c2 can be determined from the following boundary
conditions
{'r,l.=o = o (6.28)
L dF,n,
=sn (6.2e)
"1,=,. ,ç dx x=t.
For a fixed end support as shown in Fig.6.3, the boundary slip s, :0 atx:L.For
useful as discussed in Section 6.1 where the linear theory is extended to a non-linear
ar.N.e-o +ar.F
sn ^F
lo,
c1 (6.30)
a ao.at.cosh(a) 2ar'cosh(a)
a^'N
L2 - -çl (6.31)
al
in which
Kb-EI
a=L -L (6.32)
L-.EA Znt
where
cll
a5= (6.33)
ao
-141-
CHAPTER 6 MATHEMATICAL STUDIES
sinh(ø'f) sinh(a.f)
F"h, =
s
+-
ao .a, .cosh(a) al [,..t'u
-e-d'Ë -*.Ð"
*, )+a,'F-E,.r.t- cosh(ø)
)
=FÍrF,+F,, (6.34)
where F,., Fn and F,, are the longitudinal shear forces induced by the lateral force F, the
axial load i/ and the boundary slip so , respectively. They are given by
F,
NEA EI , -a.t 2sinh(a.(\
l-e'---_=:- (6.36)
(EA), EI g'" +l
Kn.sinh(a'Ç) (6.37)
F,,, =sp L
L" a.cosh(ø)
where f is the normalized coordinate x, or
x
5 0<€<l (6.38)
L
,=5.0':r,=srrs,*s,o (6.3e)
Khdxr
where s¡t s, and s,o are the slip terms due to F, N and sp, respectively, and are given
by
EA
st:F'(h+4)
EI
=F.9
as
(6.40)
-142-
CHAPTER 6 MATHEMATICAL STUDIES
¡/ .4.,
K.EIU nt.lL" ", -
L,
,s
(84'
l
J
=-F: þ-"
r" - 2coshe - .,[ a, .
€) f t""''
tr' * r¡l (6.41)
.a.. N
v = v'1,,, + ao'ao'F,r, +a." ' .x"+c3.x+c4 (6.43)
2
where
: I* F.x
vl1arr =Jdx (
I -Y)*
EI) EI
F.E
6EI
ç63 -t6 +z¡ (6.44)
which is the deflection due to the lateral force F with full interaction, i.e. no slip between
elements I and2. The constants c, and c4 are determined from the boundary conditions
''rl - 0 (6.4s)
0 (6.46)
which gives
Ku
ct -- -a+. ar. L. N - ao. ao .t n' (6.47)
L"
-143-
CHAPTER 6 MATHEMATICAL STUDIES
where
(h+hr) EA Ê .N-(6
vJirll -1)'
2Er.(EA)I
=].,nr.e+.oz.t].G-t)t
2
(6.50)
which is the deflection caused by a constant moment due to the eccentricity of the axial
load about the centroid of the combined composite cross-section of elements I and 2 with
full interaction.
the partial interaction effects for the lateral force, axial force and boundary slip,
respectively, where
9r(6)=(h,+hr)
EA
EI
ZII
(EA),
L Kb
c-ot- 2sinh(a-O.
-
I
,n *1 *rh(o)
-l
,, _2sinh(a .6)
- % l-""
_a3.a4 .l _
- *-*rh(")l
' (6.s2)
,\t
h,) 4' Isinh(ø'6)-sinh(a) +t-
c,(6) \"t+ "2/
Òr\?,= (h U "l 6
".*rnø¡
-144-
CHAPTER 6 MATHEMATICAL STUDIES
:a4.L (6.s3)
The functions gr $) and gr(€) approach zero when the stiffness of shear connectors
approaches infrnity, which leads to the result given by full interaction theory.
For a composite member, there are many member properties that affect the response.
Furthermore, the effect of each variable is not obvious. Some of the variables are inter-
related, fufther complicating matters. For example, in Eq.6.40 the slip appears to be in
Furthermore, ø also depends on l\ and hr. However, careful study of the variables
reveals that the response of a composite member is determined by only a few composite
parameters that are combinations of basic material and geometric properties as described
ln order to establish which factors affect the slip distribution in a composite member with
Eqs.6.l9-6.21 as
For a linear system, the superposition law applies. Therefore, the total slip strain of
-145-
CHAPTER 6 MATHEMATICAL STUDIES
ds,,
(e,,0), ar.N (6.s7)
dx
where (Ê.,,0)r, (e,,0), and (e,,0), are the slip strain terms caused by (l) longitudinal shear
force F,t,,, (2) bending moment M(x) and (3) axial load N, respectively. The
be rewritten as
ds. I
dx (6.62)
dx'on J"
6.63. Apart from the loading conditions which determine Fr(x) and Fr(x) , there are only
three composite parameters that affect the slip: a, , a, and ar. These three composite
-146-
CHAPTER 6 MATHEMATICAL STUDIES
discuss the parameter ss) the parameters ao and a, are also analyzed
As given by Eq.6.18, it can be seen that lf ao is the shear connector lateral stiffness
from the shear connectors, tending to give alarger value of slip in the interface.
Letting F,t,=0, N:0 and M-l in Eq.6.54 gives e",o =ar where a, is given by
8q.6.20. This means that the coefficient a, is the slip strain caused by a unit external
bending moment at the cross-section when no interaction exists at the interface (no
shear connectors). A larger value of a2 means that, for a given external bending
moment, a larger slip strain will occur. In other words, a member with a larger a,
value is prone to have larger slips under flexural loading. From 8q.6.20, it can be
Let M: N--0 and F,n,:7 in Eq.6.54, then e,,o = -at, in which a, is also a function of
slip strain of the cross-section caused by a unit longitudinal shear force. For a cross-
section with a larger al, a given longitudinal shear force will cause larger
longitudinal slip strains and hence slip. As the longitudinal shear force always tries to
-147-
CHAPTER 6 MATHEMATICAL STUDIES
stop the slip caused by the external forces (hence the name possive slip strain
coe/.ficient as opposed to the aclive slip strain coefficient a"), ã larger a, indicates
that the longitudinal shear is more effective in stopping the slip or resisting the slip
strain caused by external forces. In other word, for composite members of the same
length but with a larger cross-sectional property a,,lhe same shear connection (same
bolt and same number) will result in a smaller slip at the interface under the same
loading conditions.
4. The axial.flexibility a,
Letting F,n,=0, N:l and M:0, then Eq.6.54 gives tin=ar, where a, is given by
F.q.6.21. That is to say, a, is the slip strarn a unit axial force applied at the
axial load will cause a larger axial shortening of element I which in turn causes a
It reflects the overall ability of the composite member in restraining the slip at the
interface, which can be observed by considering the physical meanings of ao and a,.
As mentioned in 3, composite members with a larger passive slip straín coefficienl a,,
and the same shear connection will have a smaller slip at the interface under the same
loading conditions. When the shear connector stiffness l/ao is larger, there is a
stronger shear connection in the interface, providing greater slip resistance, which
further reduces the slip. Therefore, for composite members with larger ar,The ability
- 148 -
CHAPTER 6 MATHEMATICAL STUDIES
of the member to restrain the slip is greater, and the slip caused by a given loading
The above relations can be verified by the slip results from the example of Section
6.2.2 where the slip terms are given by Eqs.6.40-6.42. In all three terms, which
respectively are caused by the external moment or lateral force at the top (Eq.6.40), the
axial load (Eq.6.al) and the boundary slip (8q.6.42),the slip resistance ability coefficient
a, plays a key role in resisting the amount of slip. In other word, all three slip terms
Similar to slip, the deflection of a composite member is also governed by just a few key
parameters regardless of the many variables involved. With reference to Eqs.6.1l and
where ao is given by 8q.6.26. The first term in the above equation is the curvature
caused by the external moment with full interaction (no slip). The second is the additional
curvature of the cross-section caused by a given slip strain e,,r. And the third term is the
curvature caused by the axial load due to the eccentricity between the centriod of element
-149-
CHAPTER 6 MATHEMATICAL STUDIES
From Eq.6.64, the factors affecting the curvature and hence deformation of the
member are clearly seen. Apart from the external member forces N and M(x) as well as
the full interaction flexural rigidity EI , the only other composite parameters that affect
the deformations are a2) 03, cts and ao, as the slip strain á",o is determined by 02, e3,
a, and the loadings as discussed in Section 6.3.1. This observation is verified by the
Similar to a2t a3, and a,there is also a clear physical meaning for coefficient
ao. To visualise this physical meaning,let M(x):0 and N:0 inEq.6.64 and consider
8q.6.54 to get
For convenience, the corresponding free body diagram and strain profile in this
Focal point 2
Element 2
F.hr Rs
hl
X
tslp
Element 1
-150-
CHAPTER 6 MATHEMATICAL STUDIES
It is noted that at the point where at'aq'!-at:0 in 8q.6.66, a:0 for any
cl1
Yt - (6.67)
at'a+
In other words, the additional strain caused by longitudinal shear (or slip) is always zero
at this point no matter how the longitudinal shear force or slip changes. This means that
all the strain profiles, for any combination or distribution of shear connections, pass
The strain profile 'focal point' is an impoftant new concept in the study of
composite structures that was first discovered and introduced by Seracino, Oehlers and
Yeo (2001). It was found initially from their numerical simulations. They later verified
this finding using linear elastic analysis to show that all strain profiles calculated from the
elastic theory, with different shear connections, intersected at a common point. However,
the physical reason as to why the focal point exists was still not clear from their study.
From the above discussion, it can now be seen clearly that the longitudinal shear
force Fn, causes an addìtional axial force Nt=-F,t, and an additional moment
- l5l -
CHAPTER 6 MATHEMATICAL STUDIES
the cross-section, which cancel each other at the focal point regardless of the value of
This investigation also reveals that the 'focal point' concept is, strictly speaking,
the characteristic of linear systems. The linear strain distribution given by F,q.6.66 is not
applicable to a general non-linear system which cannot have a single y value for s:0 as
For the same reason, element 2 also has a focal point. The y co-ordinate of focal
I
V. =------------=-- - hr-lt, (6.68)
ot.oq.(EA),
The distance between these two focal points, R, as shown in Fig.6.4, is given by
(6.6e)
o,4
From Eq.6.65, it can be seen that ao is the curvature when the slip strain caused by the
longitudinal shear t,,o: l. Because of this reason, ao is named as the unit slip curvature.
Therefore, according to 8q.6.69, R. is named the slip radius as oppose to the unit slip
curvature ao. It is noted that the slip radius (and unit slip curvature) is also function of
cross-sectional properties.
From Fig.6.4, it can also be seen that the slip strain can be expressed as the slip
-152-
CHAPTER 6 MATHEMATICAL STUDIES
Comparing 8q.6.70 with Eq.6.65, it can be seen that they are the same which further
supports the concept of 'focal points'. Interestingly, the first and third terms inEq.6.64
can also be obtained with the concept of slip radius or unit slip curvature. For a
composite member without interaction, the curvature due to an external moment is given
M(x)
á"ø =#.(hr+hr) (6.71)
L"'
For a composite member with full interaction, this slip strain must be resisted by the
longitudinal shear force so that e.,,, = 0 , or the slip strain due to the longitudinal shear
force is
(6.72)
parts, or
which is the same as the first term in 8q.6.64. The third term of 8q.6.64 can also be
- 153 -
CHAPTER 6 MATHEMATICAL STUDIES
Due to the irnportance of tlie longitudinal slip, ¡ts distl'ibutions ale furthel discussed in
this section based on the results from linear elastic theory. Typical slip distributions for
the cantilever column obtained from Section 6.2.2, as given by Eqs.6.40-6.42, are shown
in Fig.6.5.
Total (by F
È
J sp
o bys
Support of
elastic colurnn
by- E
hyN
Plastic
Top of column Elastic column
The slip term due to flexural deformation given by 8q.6.40 is shown by the curve
marked with "by F" in Fig.6.5. The slip term caused by the axial load N given by Eq.6.4l
li,. -,,'
is marked with "by N" in the figure. 'When the load and deflection is--small, no plastic
hinge forms and the total slip is given by the curve marked with "by F+N". The other
hinge interface when the plastic hinge forms at large deformations, as marked with "by
sn" in Fig.6.5. Summation of the above three terms depicts the total slip when large
plastic deformation occurs, as shown by the thick line indicated with "Total" in Fig.6.5.
-154-
CHAPTER 6 MATHEMATICAL STUDIES
After the formation of the plastic hinge, the slip on top of the plastic hinge, s, in
Fig.6.5, continues to increase when the deflection of the column is further increased. The
slip s, depends on the deformation of the plastic hinge that will be shown in Section
8.2.1 (see Eq.8.5). However, the slip term due to N does not change. The slip term due to
F is actually determined by the moment distribution along the column that will not
increase significantly, instead it may decrease significantly after formation of the plastic
hinge. Therefore, this F term may continue to increase slightly after yielding of the
column, but this increase will be limited as compared to the fast increase in slip s,
caused by the fast increase in plastic deformation in the hinge. It is, therefore, possible
that when the plastic hinge deformation of the column is sufficiently large, which is
directly related to the slip sn, the slip term 'by sn ' is more prominent than the other two
terms. In this case, the position of maximum slip will occur atthe top of the plastic hinge
instead at the top of the column. An example that compares the analytical results to the
0.8
o.7
06
E
0.5 -*-ultimate stage num. result
CL
-- -À- - -elastic stage num result
0.3
-ultimate c stage analy. result
0.2
0.1 r
0.0
0 200 400 600 800 1000 1200
longitudinal distance from end (mm)
Fig. 6.6 Slip distributions compared with the numerical results
- 155 -
CHAPTER 6 MATHEMATICAL STUDIES
The numerical slip distribution just before the concrete cracks at the tension face,
as indicated by 'A' in Fig.6.6, agrees very well with the analytical results. This means
that the deformation of the column can be reasonably modeled as linear elastic before the
concrete cracks on the tension face. At large deformations, however, the numerical
results are generally greater than the analltical results, as shown by the curve marked
with '0' which indicates the stage just before the concrete crushes (compressive
strength:O) on the compression face. This difference is mainly due to the assumption
made in Section 6.1 that the column is still linear elastic above the plastic hinge at large
deformations. The small non-linearity in the part of the column above the plastic hinge
causes the numerical, or "ac|ûa!", column to deform slightly more than predicted by
elastic theory. The slight increase in column deformation corresponds to an extra amount
of slip. This is the reason that the analytical slip distribution at the large deformation
stage underestimates the amount of slip. However, the numerical results may slightly
overestimate the true slip due to the neglect of tension stiffening. If tension stiffening
were considered, the column would be slightly stiffer hence experiencing less deflection,
which would lead to a smaller slip. That is to say there would be a closer agreement
between the analytical and the numerical results if tension stiffening were considered in
the numerical calculations. As the difference in Fig.6.6 is not substantial, this comparison
verifies the legitimacy of the linear assumption outside the plastic hinge region.
From the slip term caused by the flexural moment as given by Eq.6.40, the slip s,
- 156-
CHAPTER 6 MATHEMATICAL STUDIES
o decreases monotonìcally when the slip resìstance ability coefficient a, increases. This
which is part of the slip term. There are two asymptotes for the curve in Fig.6.7(a).
Mathematically it can be proved that lim v =-. '2 t].0-6\. In this case, Eq.6.40
glves s/. F.or.L'.(l-5'?;. ttris is the case of zero interaction where no shear
2
connection exists on the interface. In another extreme case of full interaction when a,
a5
(a). Function,
=* [r-.osh1z./ãl.f)/co*1r ./ãl¡]
1.0 gE
€:o.s
E:o.e
v
0.0
ct
-157-
CHAPTER 6 MATHEMATICAL STUDIES
approaches infinity;
o is a maximum at the top of the column (€ = 0 ), where the moment is zero. The
a2
S,n*:F (6.7s)
a5
and;
a decreases monotonically in a convex shape to zero at the support where the bending
moment is a maximum.
From Eq.6.4l it can be seen that the slip caused by the axial load s,,
= L'(1 -6) which gives s,, : N' crz' L' (l- O ; and jgi"v = 0 gives s, : 0.
)i\y
-158-
CHAPTER 6 MATHEMATICAL STUDIES
a5
0 L
-t59-
CHAPTER 6 MATHEMATICAL STUDIES
o increases monotonically with increase in L at a point that keeps a fixed distance away
from the top (Z increases but x =ö'L =constant), as shown in Fig.6.8(b), with an
distance from the top of the column to the point considered. However, at a point that
has a relative position fixed (i.e. ftconstant but(+0, for example the point at the
middle of the column where ç0.5), the slip initially increases with increase in Z,
when Z is less than a certain length L, . After this point L" , the slip will reduce when
N.o,
S,r* _ (6.76)
fi-zf çu"''tÈ'*t¡f
l%
and;
Equation 6.42 shows that the slip along the column induced by the boundary slip s,
decreases when the slip resistance ability coefficient a, increases. For the case of no
interaction i.e. or:0, the slip is a constant value s, along the length. For the case of
- 160-
CHAPTER 6 MATHEMATICAL STUDIES
approaching full interaction when a, approaches infinity, the slip approaches zero
except at the vicinity of the support where the slip is equal to so at the support;
from Eq.6.42 that the slip at a point that has a certain distance away to the bottom of
a monotonically increases from the top to the bottom of the column in a concave shape.
The maximum slip value at the support when (:1 is so, and the minimum slip at the
-16l-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
In Chapter 8, a methodology will be developed to design the plating system where the
firstly defined and studied. In order to study the ultimate displacement, the flexural
behavior of RC cross-sections is analysed with some very simple models in this chapter.
This analytical study also provides a clearer insight into the various factors that affect the
CONCRETE SECTIONS
The model to be studied first is a plain concrete column with a typical cross-section of
breadth B and a depth D as shown in Fig.7.l. For simplicity, the stress-strain relationship
of the concrete is idealised with the bì-linear model of Fig.4.4 in this section.
For convenience of study, the strain in every fibre of the cross-section is assumed
This assumption means that the stress in any fibre always goes along the envelope curve
without going into the unloading/reloading branch shown in Fig.4.4. Strictly speaking,
this assumption is not correct, as some areas in the cross-section do load and unload in
-162-
ì
the course of the monotonic curvature increase in the section. However, most of this
iw9 i'¡ L
unload/reload^takes place within the ascending branch of the envelope curve, or in the
Ai\
linear elastic range, where the unloading/reloading does not affect the loading history at
conclusions drawn from this part of the analytical study are verified by the numerical
X:""v/
study in Sections 7.1.4 and 7.3 wherelull non-linear loading/unloading history in every
Nt (corßhnt)
+ I
F
(Ydv)
D D
B
Ser[A
Sec,AA
The variation of the stress profiles of the typical cross-section with monotonic
increase in the curvature of the section is shown in Fig.7.2. As the column is initially
loaded with the axial load N only without lateral force or moment, the initial stress profile
distribution A', B, C, D and E successively, as shown in Fig.7.2. In all these cases, the
-163-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
F
area of the stress blocks keeps unchanged due,(he condition of constant axial load. These
different stages of stress distribution are studied in detail in the following sections.
M^= o
N=G
Stress distribution A
M^.
2o,. N:G
Stress distribution A'
MB
x N:C
Stress distribution B
Mc
N:C
Stress distribution C
MD
N:G
Stress distribution D
ME
fco
N:C
Stress distribution E
-164-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
'When the curvature increases monotonically, the stress on the left-hand side of the
section is increased while the stress on the right-hand side is decreased. When the stress
at the top extreme fibre on the left-hand side reaches the compressive strength f,o, ãs
shown by the stress profile B, a non-linear or plastic deformation starts to occur when the
moving from the center of the cross-section towards the left-hand side monotonically.
Therefore the lever arm and hence the moment of the section is also increasing
monotonically.
neutral axis depth x is less than the depth of the cross-section D which is an adequate
N : B.f,".x l2
= B . .f,o.!oo tz (7 .1)
Kro
B ' f,u'€,u
2rc
or
K co -f "o I co B t"o
(7.2)
2N 2D.(N lN,)
where B is the breadth of the section; €"o and f,,, are the strain and stress at the peak of
the stress-strain curve as shown in Fig.4.4; N" is the crush load given by
- 165 -
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
N,=.f,u.Ar= f,u'B.D; and Kc(, is the curvature of the section when the compress¡ve
This stage starts from distribution B and ends with distribution D inFig.7.2. The stress at
the face of the left-hand side starts to decrease with further increase in curvature, as
moves further left then starts to move back towards right-hand side. In other words, the
N = B. f"".x l2
B't",,' f"o (7.3)
- 2rcLU
or
K cu
fro't"r'B c
(7.4)
2N 2D.(N1N
")
where a,,, is the ultimate concrete strain at zero strength as shown inFig.4.4; K.,, denotes
the curvature of the section at which the strain of the extreme compressive fibre just
reaches e.,,
-166-
CHAPTER 7 ULT1MATE CURVATURE OF RC SECTIONS
It is noted that distributions B and D must have the same neutral axis depth x in
order to satisfy the constant N condition. In the centroid of the stress block .y
^ruryfti^",the
- ..-2
--
D is on the left to that of distribution B. In other words, there must be a drop in moment
resistance depends on the depth of the compression zone x or axial load level.
The fof lowing relation is derived using the stress distribution C as shown inFig.7.2
(7.s)
in which Misthe moment of the section underthe curvature 6 andx is the compression
zone depth given by
*=''n *å['
K ü.Kl -(r+a-2rc'a'N ,)r11 (7.6)
J,,.t,,,.8' l
and
d !.
E"
(7.7)
t"u
p
c
=l+la.
where E" and E" are the elastic modulus of the ascending branch and slope of the
-167-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
Any further increase in curvature from distribution D inFig.7.2 will move thc zcro strcss
However, the axial load level cannot be sustained in this case. The reason is given below.
x,=t,,fK" (7.8)
N"= B.fr,,.x"l2
B'Ê""' (7.e)
- 2rc"
'f"o
K K ct
in which r" is the curvature of the section under distribution E, and x" is the distance
from neutral axis to the point where the compressive strain equals to r",,, as shown in
The physical reason behind Eq.7.l0 is that the horizontal length of the triangle
stress block reduces when the curvature further increases after distribution D. While the
tt'
{ 1"1t "'
1 t''
'
maximum stress" maia6'ÌË-constant ãi .f",,, the area of the stress block must reduce. If the
axial load remains constant at this time, the whole concrete section will crush suddenly
-168-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
To further illustrate the results of Section 7.1.1 to Section 7.1.3, an example of the
curvature moment relation for a plain concrete section is shown in Fig.7.3. Details of the
V\
column section is give below:
+Analytical
3 008+08
M.o * Numerical
2 50E+08
E 2 00E+08 distribution B cu
Ê 'yield'point
z
1.50E+08 l
o distribution D
Ê
o 1.00E+08
=
5.00E+07
K"o Kcu
0.00E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05 2.50E-05 3.00E-05
Curvature (1/mm)
history. The same stress-strain relation as shown by Fig.4.4 is used in the numerical
calculation. The agreement of these two results verifies the monotonic strain assumption
-169-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
From thc abovc discussion it can bc sccn that thc concrctc scction complctcly crushcs
when the compressive strain at the extreme fibre reaches t,,,. Therefore, the ultimate
reaches a value that satisfies Eq.7.4. This curvature is defined as the ultimate curvature of
point where the resistant moment of the section drops to a certain percentage of the
maximum strength (Watson, Zahn and Park 1994). Different values for this drop, such as
lÙyo, 20yo or 30o/o, have been used. This defrnition of ultimate deformation is based on
In fact, the strength drop at the true ultimate curvature given by 8q.7.4 can be
where M,,, is the moment of a cross-section when the ultimate strain á.,, is reached at the
.-
M(o N.(2-?. * (7.12)
'2 3 f",,'B'
¡
-170-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
N
M"o- Mrr, 2(B -1) N,
- . (7.13)
M", 3P
where N I N gives the axial load level. It can be seen from Eq.7 .13 that the moment drop
"
at the true ultimate deformation as relative to the yield point is only determined by the
material property pand the axial load level. Typical curyes given by 8q.7.13 are shown
in Fig.7.4. However, the moment drop given by Eq.7.l3 is relative to the yield point not
the maximum strength point as can be seen in FigJ.3. This modification simplifies the
discussion.
80o/o
70o/o
o
o 600/o
=5 50To
o 40Yo
=o 30o/o
(,
20%o
=
10o/o
0o/o
In engineering, the maximum axial load level is usually limited by design codes.
N =0.75.ú.N,,,,=0.382f",,.4, (7.14)
- 171 -
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
Therefore, Fig.7 .4 covers the whole practical range of the axial load level.
As sccn from Fig.7.4, thc moment drop from the 'yield' point to thc true ultimate
displacement ranges from l0% to20%o of the'yield'moment for axial load levels of 0.1
to 0.2. Therefore, the traditional ultimate displacement defined at a llYo of moment drop
will be equivalent to the true ultimate displacement for a column with axial load level of
0.1(10% of the crush load). For other axial load levels, the traditional ultimate
displacement cannot reflect the true deformation capacity of a column.
From Fig.7.3, it can be seen that the curve from the yield point to the ultimate
point may be defined as a yield plateau. In this case, the maximum curvature ductility of
definition where a certain percentage drop of the peak moment value is used to locate the
point of the ultimate curvature, regardless of the axial load level. In the current literature,
it is well established that the behaviour of a concrete column is closely related to its axial
load level. From this point of view, the traditional definition of ultimate curvature cannot
reflect the true extent of physical deterioration in the column. The above definition given
by Eq.7.l5 adequately reflects the degree of damage to the concrete of the column based
on the specific axial load level. The yield plateau starts at the very first point at which the
-172-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
extreme compressive fibre just starts to yield and ends at the true ultimate curvature
where any further curvature increase will cause the complete crushing of the whole
concrete section. Therefore, distributions B and D in Fig.7.2 have very clear physical
meanings: the qualitative turning points of the concrete material in the cross-section from
Equation 7.15 tells us that the curvature ductility of a plain concrete section is
(a) in direct proportion to the material plastic deformation capacity t"u - t,,, ,
(b) in reverse proportion to the axial load level N/i/.. This relation was observed by
many researchers from experimental works (Berwanger 1975; Sheikh and Khoury
1993; Watson and Park 19941' Bayrak and Sheikh 1997); and
(c) in reverse proportion to the section depth. It is consistent with the common sense that
The ductility factor given by F.q.7.16 is only affected by strain properties t",,t,o of the
concrete material.
The above conclusions are based on the simplified stress-strain model of Fig.4.4.
Another assumption made in this study is that the neutral axis is within the depth of the
cross-section from distribution B to D. 'When the axial load level is very high
(N > 0.5f,.Ar), the neutral axis will fall outside the right hand side of the section at
stress distribution B. However, it is outside the practical range of axial load level as
mentioned above.
-173-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
From Eqs.7.4 and 7.15,it can be seen that the deformation capacity of a plain concrete
of the concrete (e ,, and e.,,), the geometry of the cross-section (D), and the axial load
level (N/N.). For RC columns, there exist two constitutive materials of concrete and
reinforcing bars. ln this section, the transverse reinforcement or stirrups are ignored and
column and the longitudinal reinforcing bars as external members that apply additional
load to the concrete, then the effect of the longitudinal reinforcing bars is clear. Since the
presence of the longitudinal reinforcement will not change the material and geometric
property of a concrete section, the only way it affects the deformation capacity of the
concrete section is through changing the axial load level applied onto the concrete
section.
The axial load produced from the longitudinal reinforcement varies with the
only affected by the instantaneous axial load when the strain of the extreme compressive
fibre just reaches e.,,. Therefore, the effect of the longitudinal reinforcement on the
ultimate curvature of a concrete section can be studied with the ultimate strain profiles as
symmetrical in Fig.7.5.
- t74 -
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
tsc
Case A
t"u
Case C
C ase B
.16
ì
xa:0.5D
Xç
Tension bars
D
Firstly let us look at the strain profile case A shown in Fig.7.5, where the depth of the
compression zoîe is exactly half of the section depth. In this case, the strain of the
compression bars á". is exactly the same as that of the tension bars á",. If the same
stress-strain relationship is assumed for both compression and tension, the net additional
axial load produced by the longitudinal reinforcement is zero. In other words, the
existence of the longitudinal reinforcement will not affect the axial load level, hence it
will not affect the ultimate deformation of the RC section. In this case, the axial load
satisfies
N = 0.25D. B . .f (7.17)
",,
-175-
CHAPTER 7 ULTIMATE, CURVATURE OF RC SECTIONS
which indicates that the axial load level is 25Yoof crush load. Therefore, this is a critical
axial load level at which no matter how much longitudinal reinforcement is used the
deformation capacity of the section remains the same. This phenomenon will be seen in
Equation 7.17 is applicable only when the bi-linear triangular concrete model is
used. When a different stress-strain model is used, this critical load level will be different
due to the difference in the area of the stress-blocks under different models. When
Mander's model is used, the critical axial load is given by (see Section 8.2.2)
When the axial load level is lower than the critical load (approximately 25olo of crush
load as given by Eq.7.l7), the compression zone depth will be smaller than half the
section depth at the ultimate curvature K.,,, âS shown by the strain distribution B in
Fig.7.5. In this case, the tensile strain in the tensile reinforcement will be greater than that
in the compression reinforcement, and therefore there will be a compression force applied
to the concrete in addition to the external axial load N. In other words, the presence of the
longitudinal reinforcement will cause an increase in the axìal load level to the concrete
section. Based on Eqs.7.4 and7.l5, the deformation capacity of the concrete section will
greater the reduction of the deformation capacity of the concrete section. This
-176-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
observation is consistent with the conclusion that the ductility of an RC beam (where the
axial load is zero) reduces with the increase in the longitudinal reinforcement ratio
(V/arner et al. 1998). This phenomenon will also be seen in the full non-linear numerical
Unlike the plain concrete section, which crushes suddenly if the curvature is
further increased after the extreme compression fibre reaches r"u, the RC section will not
crush immediately. Further increase in curvature will cause some concrete to crush on the
top of the compression zone and the maximum axial load that can be carried by the
concrete will be reduced for the reason given in Section 7.1.3.\n this process, the neutral
axis moves toward the right hand (tensile) side so that the compression force in the
reinforcement is increasing and tension force is reducing. As a result, the additional axial
load due to the steel bars is reducing hence the total axial load applied to the concrete is
reducing. Therefore, the axial load balance of the section can still be sustained regardless
of the continuous crushing of concrete and diminishing size of the concrete stress block.
The change in axial load on the concrete and variation of the stress block will be
The compression zone depth wìll be more than half D at the ultimate curvature when the
axial load level is higher than the critical load. Therefore, the compression force in the
reinforcement will be larger than that in the tension reinforcement, as shown by the case
C in Fig.7.5. The resultant force in the longitudinal reinforcement will be such that it
reduces the axial load level applied to the concrete section. Based on the conclusion in
-t77-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
Section 7.1 (see Eqs.7.4 and7.l5), the ultimate curvature of the concrete section will be
increased due to the presence of the longitudinal reinforcement. However, the increase in
the ultimate curvature due to the longitudinal bars is limited due to the high axial load
level and this will be seen in the numerical study in Section 7.3.
It has been shown in Section 7.1 that the true ultimate curvature of a plain concrete
column is achieved at the onset of the ultimate concrete strain á.,,, at the complession
face of the column. This criterion may still be applicable to RC columns with little
confinement such as old columns with inadequately designed and detailed stirrups. The
reason is that (for old RC columns) the ultimate curvaÍure thqÍ can be developedwithin
the potential plastic hinge region is limited by the sîrain at u,hich the cover concrete
begins to spall, which is typically around 5'% strain. The hoop steel unravels when the
longiÍudinal strain is higher than 5'%, resulting in the loss of small confinement and
shear failure may occur at any time. Although the RC columns may not crush
immediately after the attainment of the ultimate concrete strain r.,, (spall strain) on the
compression face, the concrete resistance in the cross-section staús to drop significantly
after this point which will be seen in the numerical example of next section.
problems in the plastic hinge zone which were usually designed and constructed with
- t78 -
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTÌONS
work as the curvature at which the extreme compressive fibre of concrete reaches the
This definition is not applicable to columns with adequately designed and detailed
stirrups. For columns well confined by stirrups, the spalling of the concrete cover does
not impair the column significantly, as the well confined core concrete is strong and
ductile enough to further take additional load and displacement. In this case, the final
crushing of the core concrete. Therefore, the attainment of t,,, or spalling of cover
columns.
Numerical case studies are conducted in this section to visualise and further understand
the conclusions of Sections 7.1 and7.2. Analytical studies in the previous sections of this
I
to be made, the numerical simulations in this section will be based on full non-linear
models, such as Mander's model for concrete and Menegoto-Pinto model for steel bars as
described in Chapter 4.
reinforcing bars (4 bars each face), which is the same as the case of Fig.4.20 studied by
Watson and Park (1994). No stirrups and confinement is considered in this study in order
to investigate the net effect due to the longitudinal bars only. The following material
-179-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
To vary the reinforcement ratio, different bar diameters are used: l6mm,20mm,25mm
respectively. Concrete cover thickness (to the centre of reinforcing bar) is calculated to be
29mm from Fig.4.20 for the case of Yl6 bars (:13+8+8). To provide a uniform
benchmark for comparison, this cover thickness is adopted for all the other reinforcement
the lever arm of the longitudinal bars which is not desirable for comparison. A plain
Several axial loads are used in the study. They are 352KN,704KN, l408KN,
2l I 2KN, 281 6KN, and 3520KN corresponding to 5Yo, 10yo, 20yo, 30yo, 40Yo, and 50%o
The results for the case of N:5% of crush load are given in Figs.7.6-7.8 . Fig.7.6
shows the moment curvature relations for sections with various reinforcement ratios.
Fig.7.7 shows one case of Fig.7.6 with 3.68% steel and in which the moment
contributions from the concrete and the steel bars are split out. The splitting of the axial
force in the cross-section for the same case is given in Fig.7.8. The definitions of the
symbols "x","Q", "*" and "*" ate given in the Notation. The corresponding curvatures
- t80-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
forthe above four stages are named K",,, K,,,, K"ry and Krcr,, respectively. The locations
of points "x","0", "+" and "{<" on the curves may not be exact as the numerical outputs
are not continuous. They are plotted at the nearest numerical points. Fluctuations or
wobbling are seen in these figures atlarge deformation. It was later found to be caused by
the discretisation of the cross-section. When a higher layer number or finer layers are
From Fig.7 .6, it can be seen the difference on the horizontal coordinate between
point + (onset of yielding of tension bars) and point 0 (onset of complete crushing of
concrete on the compression face) gives a good indication of the ductility of the cross-
section. Although considerable strength and a large plateau remains after points 0 as seen
from the chart, this remaining strength is only theoretical and not reliable. The reason is
that the concrete deteriorates (strength drops) significantly after point 0 as seen from
Figs.7.7 and 7.8. To further illustrate the extent of damage to the concrete, a concrete
stress block (at rc = 7.5x10-5 limm) after point 0 is shown inFig.7.9, where it can be
seen that a significant area in the compression zone has completely crushed. After
crushing of the cover concrete, the longitudinal bars will lose support and hence may
buckle at any time. Therefore, the remaining strength in Fig.7.6 after point 0, which
comes mainly from the reinforcement bars as can be seen in Fig.7.7, is not reliable and
hence is only a theoretical strength. Figure 7.8 confirms the theoretical conclusion given
in Section 7.2.2 that the presence of longitudinal bars increases the axial load on the
- l8l -
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
The numerical results for other axial load levels are shown in Fig.7.l0 to Fig.7.l4.
From the appearance of these figures. it also looks reasonable to define points 0 as the
ultimate displacement of the RC columns. The key point curvatures K,,, (onset of peak
concrete strain e.o at compression face), r",, (onset of ultimate concrete strain t",, at
compression face), r,u, (yielding of tension bars) and r,.,, (yielding of compression bars)
versus axial load for different reinforcement ratios are depicted from Fig.7.l5 to Fig.7.l8.
More cases with higher axial load levels are calculated and included in Figs.7.15 to 7.18
for comparison, although these axial load levels are not practical.
From Figs.7.l5,7.16 and 7.18, it can be seen clearly that K',,> K"u, and K,"r,
reduce when the axial load increases. For the same axial load level and when the axial
load level is lower than 30%o, Fig. 7.15 shows that K,,, reduces when steel content
increases. It is the other way round for axial load levels higher than 309/o in Fig.7.l-5.
Similar observations can also be made for K.,, and K,,., in Figs.7.16 and 7.18. It was
concluded theoretically in Section 7.2that the ultimate curvature K..,, reduces when the
steel content increases for axial loads less than the critical axial load and vice versa for
axial loads higher than the critical load. This critical axial load is given by Eq.7.18 to be
Fig.8.4 when the concrete strength is 44MPa. These theoretical conclusions are
confirmed by the numerical results of Fig.7.15 where it can be seen that the
reinforcement content does not affect the ultimate curvature at the axial load level of
about 30% of the crush load. In other words, the numerical calculated critical axial load is
abouT 30o/o.
-182-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
Sections 7.2.1-7.2.3 are still applicable if e",, and K,,, in Fig.7.5 are changed to t,,, and
K",,, which explains the same trends in Fig.7.l6 as seen in Fig.7.15. For K,,4, nS shown in
Fig.7.l7, it increases when axial load increases and the steel content apparently has no
effect on its values. When axial load level is more than 40o/o, the tensile reinforcement
Another observation that can be made from the moment-curvature charts is that
the yield plateau can be well reflected between point + (first yield of tension bars) and
point 0 (spalling of concrete cover) for axial load levels lower than 30Yo of crush load, as
seen in Figs. 7.6, 7.10 and 7.1l. However, for higher axial load levels as shown in
Figs.7.12,7 .13 and 7 .14, it is apparently more reasonable to define the yield plateau from
point x (onset of maximum concrete strength on the column face) to point 0. In fact, point
+ or yielding of tension bars does not occur for any of the columns represented in
Fig.7 .14 where N:50% N.. No matter which definition is used for the yield plateau, it is
clear from these moment-curvature charts that the yield plateau decreases when the
reinforcement ratio increases for axial loads lower than the critical load, as shown by
Figs.7.6,7.10 and 7.11. For axial loads higher than the critical load as shown by
Figs.7.l2,7.l3 and 7.14, the yield plateau increases when the reinforcement ratio
increases. This observation further verifies the theoretical conclusions in Section 7.2.
- 183 -
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
8.00E+08
7 00E+08
6.03% of steel
6.00E+08
Ê 5.00E+08
E
1 00E+08
0 00E+00
0.00E+00 5.00E-05 1 00E-04 1.50E-04 2.00E-04 2.50E-04 3.00E-04
Gurvature (l/mm)
6 00E+08
5.00E+08
Curvature (1/mm)
Fig.7 .7 Moment-curvature chart for section with 3.68% steel and N:SYoN"
- 184-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
1.50E+06
1 00E+06
N on concrete
z
5 00E+05
E(E Total on RC section
o
J
G
'* 0 00E+00
5.00E-05 1 2.008-04 2508-04 3.00E-04
-5.00E+05 N on steel
-1.00E+06
Gurvature (1/mm)
Fig. 7.8 Axial force-curvature chart for section with 3.68% steel and N:So/oN"
4.5E+01
4.0E+01
3.5E+01
I
st I
É 3.0E+01
E
2.5E+01
I
z
o
an
2.0E+01
o
(t, 1 5E+01
1.0E+01 I I
5.0E+00
----s
0.0E+00
0 5 10 15 20 25
\ compression face Layer Number
steeland N:íYIN.
-185-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
8.00E+08
7 00E+08
6 03% of steel
6.00E+08
E
E 5.00E+08
z 3.68% of steel
4.00E+08
o
tr 3 00E+08
o 2.36% of steel
= 2.00E+08
plain concrete
1.SYo
1.00E+08
0.00E+00
0.00E+00 2.00E-05 4.00E-05 6.00E-05 8.00E-05 1.00E-04 1.208-04 1.408-04
Curvature (1/mm)
9 00E+08
8.00E+08
7 00E+08
of steel
E 6.00E+08
E
5.00E+08
=c 3.68% of steel
o 4.00E+08
E
o 3.00E+08
n concrete
= 2 00E+08
1.00E+08
0 00E+00
0.00E+0 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05 6.00E-05 7.00E-05 8.00E-05
0
Curvature (l/mm)
- 186-
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
I 0E+08
I 0E+08
7 0E+08
6.03% of steel
E 6 0E+08
E
5 0E+08
=c 3.68% of steel
o 4 0E+08
E
o 3 0E+08
= 2 0E+08 nf stccl
1.0E+08
steel
0.0E+00
0 0E+00 1 0E-05 2 0E-05 3 0E-05 4 0E-05 5.0E-05 6.0E-05 7.0E-05 8.0E-05
Curvature (1/mm)
I 0E+08
8 0E+08
7.0E+08
6 03% of steel
E 6.0E+08
zE 5.0E+08
3.68% of
o 4.08+08
E
o 3 0E+08
E plain concrete
2 0E+08
'1.0E+08
of steel
0.0E+00
0.0E+00 I 0E-05 2.0E-05 3.0E-05 40E-05 5.0E-05 6.0E-05 7 0E-05 8.0E-05
Curvature (1/mm)
-187-
CHAPTER 7 ULTìMATE CURVATURE OF RC SECTIONS
9.0E+08
8.0E+08
7 0E+08
3 68% of steel
E 6.0E+08
c
z . 2.36o/o of steel
5.0E+08
c
o 4.0E+08
E
o 3.0E+08
= 2.0E+08
1.0E+08
0.0E+00
0.0E+00 1.0E-05 2.0E-05 3.0E-05 4.0E-05 5.0E-05
Curvature (l/mm)
2.0F-04
--.-0% steel
1.5E-04 -r-1.50%
E -+2.36%
E
3.68%
J
'1.0E-04 +6 03%
o
Y
5.0E-05
0.0E+00
0 10 20 30 40 50 60 70 80 90
- 188 -
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
1.2E-04
*0% steel
1.0E-04 I ú- 1.50%
-+--2.36%
8.0E-05
r 368%
E -à+-6.03%
ts
6.0E-05
o
o
Y 4.0E-05
2.0E-05
0.0E+00
01020 30 40 50 60 70 80 90
'1.90E-05
1.80E-05
1 70E-05
I 60E-05
Ê
E 1.50E-05 6.03%
*-F3.680/o
1.40E-05 **--2.36%
th
Y
1.30E-05
-* 1.5% steel
1.208-05
1.10E-05
1.00E-05
0 10 20 30 40 50 60 70 80 90
- r89 -
CHAPTER 7 ULTIMATE CURVATURE OF RC SECTIONS
1.0E-04
9.0E-05
8.0E-05
7.0E-05
tr * 1 5% steel
6 0E-05
E *-]*2.360/o
5.0E-05 **-3.68%
(J
o 4.0E-05 6.03%
Y
3.0E-05
2 0E-05
1.0E-05
0.0E+00
01020 30 40 50 60 70 80 90
- 190 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
chapter. For a column, the ability to sustain a certain level of lateral resistance after
yielding until the required maximum lateral drift is important. Therefore, this concept is
adopted as a guideline to the retrofit design in this study. The design procedure developed
which the retrofitting system is required to work in its 'strength stiffened' region of the
response curve as discussed in Section 5.2.2. In this way, the lateral resistance capacity
and the integrity of the column can be assured at the target displacement. To achieve this,
the target displacement is designed to occur simultaneously with the onset of the ultimate
curvature, as defined in Chapter 7, of the cross-section in the plastic hinge zone of the
column. With this condition, the ultimate curvature of the RC column will not be
exceeded for drifts less than or equal to the target displacement, this ensures no excessive
deterioration of the concrete. This design procedure is fundamentally consistent with, and
can be easily adapted for use with, the modern displacement-based seismic design
philosophy (Calvi and Kingsley, 1995; Moehle, 1996; Priestley, 1997 and I 998).
To develop the design procedure, several fundamental relations are needed which
are firstly derived in Sections 8.1 and 8.2. The design procedures are then presented in
Section 8.3, followed by a design example in Section 8.4 to illustrate the procedure.
-191 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
Thc dcsign proccdurc in Scction 8.3 rcquircs thc slip on top of thc plastic hingc. To
calculate this slip, a general relation between slip and deformation of a column is derived
in this section. With reference to Fig.8.1, the strains in the plate and the RC column at the
tt=K.h,+t," (8.1)
where á,. and €r" are the strains at the cross-sectional centroids of element I (RC
t D
Nz+ Nr: N
._- Shair prulilc uf plate
Concrete stre SS block
Tensile strain
tz"
!
lve strain
tt.
Concrete strain profile
Centroid of RC column
Centroid of plate
Substituting Eqs.8.l and 8.2 into Eq.6.l 1 and integrating with respect to x gives
-192-
CHAPTER 8 DESIGN OF PLATING SYSTEMS
.' d0 : -K'dx
dL2 = -€2,'dx
d\t - -4".dx
where s is the slip; d is the rotation of the cross-section; and A, and A, are the
respectively.
.'.C=0
which gives
The above relation shows that the slip at any cross-section is given by the rotation
of that cross-section times the distance between centroids of element 7 and 2, plus the
difference of axial shortening between element I and 2. This relation is general, because
it is derived from the geometric relations with the only assumption being that plane
sections remain plane for element9 I and 2 separately. Therefore, it is applicable for
t\
general non-linear analyses.
The formulae related to the plastic hinge calculations that will be used in Section 8.3 are
derived in this section. It must be noted that the derivations in this section are only
applicable for monotonic loading because the stress of both the concrete and steel cannot
be determined by strain only under cyclic loading, as it is also dependent on the loading
- 193 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
history. Therefore, all the anal¡ical results that are related to stresses or forces are only
The slip at the top of the plastic hinge can be calculated using Eq.8.4
tr=(4+h)'0p+Lt"-L2" (8.5)
where do is the rotation of the cross-section on top of the plastic hinge, i.e. the total
pf astic hinge rotation; and 4," and L,r, are the respective axial shortening of the plastic
hinge at the centroids of element I and element 2.The three terms in Eq.8.5 are discussed
As discussed in Sections 4.2.3 and 6.1 (also see Fig.5.5), the plastic deformation
concentrates in the plastic hinge zone. Furthermore, the total elastic deformation is
relatively small compared to the plastic deformation in the ultimate deformation stage, as
displacement is achieved in the column, the displacement at the top of the column due to
Ln=0r'(¿-0.5'Lr)=K,u'Lr'(L-0.5'L) (8.6)
where the ultimate curvature rc.,, is assumed to be constant within the plastic hinge. The
plastic hinge length of the cantilever column Lo can be estimated by Eq.4.56. Eq.8.6 was
given by Priestley and Park (1987) assuming that the plastic rotation dn is concentrated
at the centre of the plastic hinge. In fact, this relation can be derived mathematically
-194-
CHAPTER 8 DESIGN OF PLATING SYSTEMS
without this assumption by double integration of the curvature along the length, or more
oe *ø-0.5*Lr)
oP*(L-Lp) 0.s0,
0.50,
Lp
K"u = (8.7)
(¿-0.5. Lp). Lp (¿-0.s.^L). LP
0o=K",'L, ^ (8.8)
(L-0.5.Lp)
- 195 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
in which t.,, is the ultimate concrete strain on the compression face at which the RC
the ultimate limit state occurs: when the column achieves the required maximum drift, the
concrete strain on the compression face reaches the ultimate strain 4,,. The legitimacy of
identifying the ultimate displacement by this concept was fully discussed in Chapter 7.
To calculate t2,, first the axial load N, on the plate section is calculated, which
is derived in Section 8.2.2. With the axial force N, known and the curvature of the plate
given by 8q.8.7, E2c can then be calculated. One of the ways to calculate er" is shown in
Section 8.2.3.
When the concrete strain at the ultimate stage is tc, on the compression face with zero
strength, the stress block is shown in Fig.8.3. If this stress profile is assumed to be
triangular as shown by the shaded area, the axial force on the concrete is given by
N =!'r .t"
rrttttt:-
^ Jttt 'B (8.1 2)
I K.u
where B is the width of the column and the remainder of the function of the right
hand side.f,ot"u/2K,, is the area of the triangular stress block. When the stress block is not
(8. r 3)
-196-
CHAPTER 8 DESIGN OF PLATING SYSTEMS
stress block
o d
q.
+
comp. face Centroid
ofRC t
section q
x:t"u/K",, tension stre SS
ignored
The above equation is similar to Eq.8.l2 with the constant coefficient lzreplaced by a
general coefficient B. For a given stress-strain model, the shape of the stress block is
(8.1 4)
where .Oq is the stress block area in which f is the distance of a point in the cross-
It
0
section from the compression face, ois the concrete stress at the corresponding point, and
x is the depth of compression zone shown in Fig.8.3. If Mander's stress-strain model (see
Section 4.1.2.2) for concrete is adopted, the value Bis found to be only a function of the
concrete strength. Thus, p can then be calculated through numerical integration of the
stress block. The results from the numerical integration of Eq.8.14 are given in Fig.8.4
The centroid of the stress block, which is needed to calculate the moment of the
cross-section, is also a function of the concrete strength with Mander's model. lts
-197-
CHAPTER 8 DESIGN OF PLATING SYSTEMS
^d
ò-- (8.1 s)
)c
where d is the position of centroid from the compression tàce shown in Fig.8.3 that is
given by
-f
J
o.€ .d6
0
d (8.16)
[o'ot
0
The numerical integration of Eq.8.l6 gives the value of é'as shown in Fig.8.5.
070
068
066
064
p 062
060
058
05ô
054
20 25 30 35 40 45 50
G o n c re-te-sTreñ gtllF"o{N[Paf -
0.455
0.450 'ê+-^ ^
o.445
ô o.440 -*r-
0.435
0.430
o.425
20 25 30 35 40 45 50
Goncrete strength f"" (MPa)
-198-
CHAPTER 8 DESIGN OF PLATìNG SYSTEMS
For longitudinal reinforcing bars, the strains at the compression and tension sides are
respectively
€rr=trr,-Kr, a (8. r 7)
Ert=trr,-Kr,,.(D-A) (8. 1 8)
where ø is the'concrete cover'distance from the face of the column to the centre of the
reinforcing bars. Forces on the tension and compression reinforcement are respectively
N, = N-Nt (8.22)
With the axial load on the steel plate given by F.q.8.22 and the curvature of the plate at
the column base given by Eq.8.7, the strain at the cross-sectional centroid of the plate
er., which is required in Eq.8.10, can be easily calculated. For convenience, the explicit
form of tr, is derived in this section, assuming an elastic-plastic stress-strain model for
-199-
CHAPTER 8 DESICN OF PLATING SYSTEMS
the plate material. Based on different yielding conditions on the faces, it is derived for the
following 3 cases.
When a compression axial load (N, >0) is applied and when the compression face is
linear elastic, the tension face must also be linear elastic. Therefore, the strain at the
N2
t2, (8.23)
B.t.E,
Figure 8.6 shows the stress and strain profiles of the plate under the elastic-plastic stress-
strain relation.
82"
€2c
tpv Strain profile
Stress profile
fot
lon
tl2
-200 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
Nz: B
f ,
-T)-f
E- -E I s-., *:'(er,
I t t u2c
L,;.
,,* ^'o'
2 tc -¡')
,,
2 -' - -' J
2
K )'En'(
2rc ) (8.24)
and
2err,+rc-t+ (2e
o, + K. t)' - - @. t - e,,). t,,
€2" =
^ l+
2
One of the two solutions above is false and can be found through the following
I
€,=t2"-r.o., (8.2e)
-201 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
Equation 8.30 means that with the solution of 8q.8.27 the whole cross-section yields for
any given axial load and curvature. Obviously it is not the corect solution and should be
(a) The first one is when the tension face just yields in compression or the whole cross-
section just yields in compression. The conditions for this to occur can be derived as
follows.
When Ae = 0, from 8q.8.30, the strain at the tensile face is t, = rn, or just yields.
balance of axial forces in the cross-section cannot be achieved even after the whole cross-
section yields. In this case, the plate thickness is too small to sustain the axial force and
must be increased.
-202 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
(b) The second special case occurs when the tension face just yields in tension. The
because substituting 8q.8.33 into Eq.8.29 gives €,:-Ery. This means the tension face
the tension face has already yielded in tension. In this case, the stress distribution shown
by Fig.8.6, which is used to drive F,q.8.26, is incorrect and therefore a third case must be
opy tPt
K K K
/i
Tension
/i
fot
v
Stress prolile
fot
Compression
., tlL tl2
-203 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
Nr-K
€2" = (8.35)
f
28. JPY
With all the necessary formulae derived in Sections 8.1 and 8.2, the design of the plating
L Calculate the ultimate curvature at the plastic hinge based on the target maximum
The design displacement or inter-storey drift A is given. Assume that the plastic
relatively small compared to the plastic deformation, hence Eq.8.7 is used to calculate
the ultimate curvature K,,, in the plastic hinge zone of the column.
2. For the cross-section at the plastic hinge location, calculate the following when the
-204-
CHAPTER 8 DESIGN OF PLATING SYSTEMS
The estimation can be based on the full yield thickness t,,,¡n=N2l(fr'B) taking into
account an allowance (e.g. a factor of 1.2) for non-uniform stress distribution in the
N'
t =1.2. (8.36)
B.fJPY
This is only a frrst approximation and can be adjusted following the calculations in
step 4.
4. Assuming the first bolt is immediately above the plastic hinge, calculate the slip sn
of this bolt by Eq.8.5. The three terms in 8q.8.5 are calculated using 8q.8.8, Eqs.8.9
and 8.11, and 8q.8.10, respectively, in which the plastic hinge length is calculated by
8q.4.5.6. Formulae derived in Section 8.2.3 are needed to calculate the strain term
t2, in Eq.8.10. The calculation of er, will indicate whether the plate thickness
assumed in step 3 is adequate or not. Adjust t if it is not adequate, and repeat this step,
otherwise go to step 5.
typical curve of Fig.4.13, the force at the first bolt can be calculated.
ìWhen sn is less than the elastic limit (from point A to B in Fig.a.l3(b)), a simple
linear relation between slip and shear load on the bolt can be assumed, i.e.
Ft, = s
o.K, . The bolt is usually working in the elastic range, because the limit in the
ultimate compressive concrete strain 4,, does not allow large slips to occur. If the
bolt is working outside its elastic range of deformation, another type of bolt should be
-20s -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
used since yielding of (all) bolts will result in a plated column working outside its
As a first approximation, assume the slip is uniformly distributed along its length.
This assumption is usually a good approximation at the ultimate limit stage, as was
seen from Figs.5.17,5.18 and 6.6. Therefore, the shear forces can be assumed to be
the same in all the bolts. Based on Eq.6.l, the number of bolts required is then given
by
n=-N., (8.37)
Fh
This estimation can be sufficient for design purpose. If higher precision is required or
the distribution of slip cannot be reasonably considered uniform at the ultimate limit
stage, the result from this step can be further modified by step 7.
First, calculate the bolt spacing based on the previously estimated number of bolts.
Second, calculate the moment of the bottom cross-section based on the forces
calculated in step 2, which determines the lateral force applied on top of the column.
Eq.8.l5 and Fig.8.5 are needed to calculate the moment due to the concrete. The slip
distribution is then calculated from linear elastic theory using Eqs.6.39-6.42 based on
the lateral force, axial load, bolt spacing and slip s, calculated previously. The slip at
each bolt position can then be calculated to get the shear force on each bolt. The
summation of all the bolt forces gives the axial force in the steel plate at the plastic
hinge. If this plate force is close enough to the value of N, that was calculated in step
-206-
CHAPTER 8 DESIGN OF PLATING SYSTEMS
2, The bolt design is adequate. Otherwise, adjust the number of bolts and spacing and
8.4 EXAMPLE
The column in Fig.5.l is to be designed for an axial load of 360 kN and a maximum drift
ratio of e : A/L :2.5Yo or /:30mm. The geometric and material properties are the same
t"u :
N"u,, = p. f,.. K,, . B 215 (kN).
thickness of 6 mm is chosen.
The first bolt is placed 200 mm (Lo:200mm) above the bottom of the column. To
calculate the slip of the first bolt, the following are calculated:
By Eq.8.8, 0, = K"u-Lr:0.0272;
-207 -
CHAPTER 8 DESIGN OF PLATING SYSTEMS
2tnr+K"u.t- (2e
n, + K,,.t)' - (rc"u. t - e r,) - e r,,
82"
^l+
2
:0.00124, checking with Eq.8.3l and Eq.8.33 indicates that the tension side of
the plate is neither yielded in tension nor yielded in compression. This confirms that the
condition to use 8q.8.26 is satisfied. It also suggests that the selected plate thickness is
adequate.
Assuming the same slip in the remaining bolts, the force at each bolt is
F¡ = s p. Kh: 23.7 (kN), and therefore the number of bolts required from Eq.8.37
IS
n=
N' : 12, o16 rows of two.
Fh
The above design results can be compared to the column in Fig.5.l analysed by
the non-linear numerical computer program. The numerical results in Fig.5.2 gives the
lateral displacement A: 33.9 mm when the strain at the extreme compressive fibre of the
concrete is equal to the ultimate strain of 0.006. This displacement of 33.9mm includes
the elastic deformation above the plastic hinge which is calculated to be 4.4 mm.
Therefore the lateral displacement due to the plastic hinge rotation only is 29.5 mm
which is very close to the specified design displacement of 30 mm. The slip at the first
-208-
CHAPTER 8 DESIGN OF PLATING SYSTEMS
bolt is calculated to be 1.034 mm in the above design that is also very close to the result
this Chapter. With this design procedure, engineers can calculate the required thickness
of the steel plate as well as number of bolts in order to achieve a given target
-209 -
CHAPTER 9 EXPERIMENTAL WORKS
Laboratory tests were conducted in this work for the following purposes:
l) To verify the effectiveness of the new retrofitting system in improving the ductility
..PLTCOL".
Many different methods are described in the literature for cantilever column tests. The
method used depends on the purpose of the tests and laboratory capabilities. Two of the
most commonly used test methods may be categorized as (l ) the single cantilever test and
(2) the double cantilever test. The single cantilever column test is schematically shown by
the test setup of Fig.9.l (Chai et al l99l; Priestley and Seible 1995; Xiao andMa 1997;
Bayrak and Sheikh 1998). In this scheme, one cantilever column, which is cast together
with a footing that is then anchored to a strong floor, is tested under a variable lateral
force and a constant axial load on top of the column. The double column test is illustrated
in Fig.9.2 (Park et al 1982 Priestley and Park 1987; Rodriguezand Park 1994; Watson
-210-
CHAPTER 9 EXPERIMENTAL WORKS
and Park 1994), where two cantilever columns on both sides of a beam stub are tested
simultaneously.
Axial load
Actuator
Column
speclmen
Footing
Strong floor
Axial load N
1l2F
Column speciman
Lateral load F
Beam stub
Column speciman
U2F
Both of the above schemes were studied and designed in detail for consideration
in this work. After comparing the two schemes, the double-column test method was
- 2',ll -
CHAPTER 9 EXPERIMENTAL WORKS
adopted for this project due to the laboratory conditions, convenience of testing and cost.
However, details of the test sehrp and the method of testing used here are quite different
from others. One major difference is that the specimens were tested horizontally instead
I 1218
roller support r
axial ilever column B stub cantilever column A
load N pin
support
ê
(a) Elevation
4
n Lateral support
4
(b) Section A-A
Another special feature of this testing system is that each specimen \ryas used for
two tests. In the first test, one half of the specimen was strengthened by steel members so
that the deformations (and member failure) occurred only on one side of the specimen, as
the strengthening was moved to the tested side and the undamaged side (that was
-212 -
CHAPTER 9 EXPERIMENTAL V/ORKS
protected by strengthening in the first test) was tested in the second test. In this way, the
cost of the tests was signif,rcantly reduced. To further minimize the stress on the
strengthened side, the vertical load was applied 100mm away from the center of the
l00mm
a
¡4
:r-. -
aa aa
aa a a a . . ol a
a
-213 -
CHAPTER 9 EXPERIMENTAL WORKS
The main test variables were chosen to be load type, stiffness of plating system
and gap (as discussed in Section 9.2.3). Two load types were used in these tests:
monotonic static loading and quasi-static cyclic loading with full load-displacement
reversal. The stiffness of the plating system depends on the stiffness of steel plate as well
as the bolt stiffness. The plate stiffness was varied by using different plate thicknesses as
well as different grades of steel. The same type of bolt was used for all the tests.
However, the overall shear connection stiffness was varied by employing full interaction
plating, which consists of plates both glued and bolted to the column, and partial
interaction plating which used only bolts to fix the plate to the column. One specimen
was tested with a bottom gap under the base plate to study the gap effect, as discussed in
Section 5.2.3.
Six tests were planned to achieve the stated aims for the tests. In the following
sections of this thesis. each test is referenced using a code reflectingthe characteristics of
2. 2AMFl2 - Monotonic loadéd ìest of the full interaction plated column with l2mm
3. IBMP6 - Monotonic loadg{-test of the partial interaction plated column with 6mm
5. 4ACP6 - Cyclic t"1ø{ test of the partial interaction plated column with 6mm thick
-214-
CHAPTER 9 EXPERIMENTAL V/ORKS
6. 2BCP6G - Cyclic loaded tes of the partial interaction plated column with 6mm thick
The first character of the code name identifies the specimen number, i.e. specimen l,
2,3 and 4, respectively. The second letter specifies the first (by letter'A') or second (by
letter 'B') test on a specific specimen, e.g.2A is the first test on specimen 2 and 2B is the
second test on specimen 2 but on the previously strengthened side of the first test. The
third letter shows the loading type with 'M' for monotonic loading and 'C' for cyclic
loading. The fourth letter gives the type of plating; the letter 'P' is used for partial
interaction plating and 'F' for full interaction plating. The number following 'P' or 'F'
specifies the thickness of the plate. The last letter'G'is used when gaps are set on the
specimen.
9.2.1 RC Specimens
The design details of the RC specimens are shown in Fig.9.5. Four numbers of
$14mm PVC sleeves were cast into the side of the beam stub that was used to anchor the
steel plates. The other four $2lmm PVC sleeves provided holes for the installation of the
strengthening steel frame which can be seen from Fig.9.4. The reinforcement details of
a) Yl6 straight bars - 2.6mlong main longitudinal reinforcement of the test column.
b) Y20U shaped bars - the main reinforcement of the beam stub with dimensions shown
in Fig.9.5(c).
-215 -
CHAPTER 9 EXPERIMENTAL V/ORKS
in Fig.9.5(d).
d) Q10 stirrups - the transverse reinforcement of the beam stub with dimensions shown
in Fig.9.5(e).
e) Top steel plates - the capping plate of the cantilever column welded to the ends of the
Yl6 bars. Dimension: l2mm (thick) x 200mm (Breadth) x 200 mm (Depth) as shown
in Fig.9.5a.
4Y 1 6
ö6@r00
stirrups 4Y20
:r 72mm steelplate
welded to the end
of Yl6 bars I
1
U bars cl
)i
C.l
\4
$21 sleeves ,
ôr 0@r 00 -
2 additiona stirrups stirrups aî
AI
+J
(a) Elevation
4Y16
$l4mm (internal) e..l
sleeves _ t50
-300-
t60 aa
ll0
O O
d
r 703
\o
2001
"
(b) Section A-A (c) (d) (e)
-216-
CHAPTER 9 EXPERIMENTAL WORKS
The design concrete cover from column surface to the outermost side of the
stirrups was 20mm. The design effective cover from column surface to the center of the
Yl6 bars was therefore a-20+6+ 16/2:34mm. The measured average value of a from the
completed reinforcement cages was about 35.7mm. This mean value will be used in the
numerical simulations in Chapter 12. Figures 9.6 and 9.7 show the completed
-217 -
CHAPTER 9 EXPERIMENTAL V/ORKS
(a) For the 12mm thick full-interaction plating on one side -the 12mm thick mild steel
plate was glued as well as bolted to the compression face of the column. Details of
this plating system are shown in Fig.9.8. All bolts were Ql2mmthreaded rod bolts
10.4. The holes in the steel plate for bolting were l2.2mm.
-218-
CHAPTER 9 EXPERIMENTAL V/ORKS
I 100 400
950
No
FPBW
A ñ l2mm steel plate
10
a a a a a o 200
a a a a a o
A.A
Note: FPBW - Full Penetration Butt V/eld
(b) For the 6mm thick partial-interaction plating on one side - the 6mm mild steel plate
was bolted to the compression face of the column. Bolting details are the same as that
(c) For the 6mm thick partial-interaction plating on two sides - the bolting details are the
same as for case (b) above, except that both compression and tension faces were
plated.
(d) For the 6mm thick high-strength plating - the bolting details are the same as for case
(c) except that high yield strength steel plate (see Section 10.3) was used and gaps
-219 -
CHAPTER 9 EXPERIMENTAL \ryORKS
When plating the concrete columns, the following procedures were used for partial-
interaction plating. V/hile this plating procedure was more complicated than what might
be done on a real construction site due to the very small clearance between the bolt and
the hole in the plate, the minimal gap between the bolt and the hole in the plate was
L Making of anchor bolts - the bolts were cut from Ql2mm high tensile strength
threaded steel rod with each bolt having a length of I l0mm.
2. Making of steel plates - Longitudinal and base plates were cut separately into the
þl22mm drill. Then the two perpendicular pieces of longitudinal and base plate steel
3. The plate was then fitted to the RC specimen and clamped in position, as shown in
Fig.9.9. The RC specimen rvas then drilled to a depth of 8Omm through holes in the
4. Remove clamps and plate. Re-drill the holes in the RC specimen with a $l4mm drill
to enlarge the holes in the concrete, as shown in Fig.9.10. This was necessary to leave
enough clearance between the bolts and concrete for the glue.
5. Clear concrete dust in the holes of the RC specimens with compressed air and inject
Hilti HIT-HY 150 adhesive into the holes, as shown in Fig.9.1 I .
6. Re-fit the plate to the specimen with 20mm thick packers to leave a space for excess
-220 -
CHAPTER 9 EXPERIMENTAL WORKS
8. Remove nuts, plate and packers. Remove hardened excess adhesive squeezed out
from holes with a sharp chisel. This finished the bolt installation as shown in
Fig.9.l4.
9. Refit the plate and fix it with washers and nuts. All nuts were tightened with a 25Nm
For the full-interaction plating of case (a), i.e. gluing plus bolting l2mm thick
plate, the procedures are similar to that of partial interaction plating with the addition of
gluing. Before installation of the steel plate, the interface of the concrete was roughened
with a needle gun. The interface of the steel plate was roughened by sand blasting. After
applying glue, the plate was installed and clamped to the RC column until the glue had
fully cured, as shown in Fig.9.l6. The brand of glue used was Ciba Performance
Polymers Araldite LC 340 with Hardener LC340.
-221 -
CHAPTER 9 EXPERIMENTAL V/ORKS
a1a
CHAPTER 9 EXPERIMENTAL WORKS
-223-
CHAPTER 9 EXPERIMENTAL V/ORKS
- 224
CHAPTER 9 EXPERIMENTAL V/ORKS
9.2.3 Gaps
For case (d) (bolted plates with gaps), the construction procedures were the same
as for partial-interaction plating. The bottom gap was set by installing a steel sheet
between the face of the stub and the base steel plate, as shown in Fig.9.17. The thickness
Steel sheet
Gap
-225 -
CHAPTER 9 EXPERIMENTAL V/ORKS
However, it was difficult to achieve a uniformly wide gap. The reason is that the
concrete face of the stub was not flat and smooth. In practice, it may even not be exactly
parallel to the face of the bottom steel plate. Therefore, the actual gap width is difficult to
quantifo by the construction method used in these tests. Attention must be paid in future
tests to this issue. One potentially better way to control the gap width more accurately,
for example, involves installing the steel plate with dental paste between the bottom steel
plate and the stub, and then lifting the steel plate up to leave the design gap width.
Furthermore, there are other gaps around the bolts, as the hole size in the steel
plate is 0.2mm larger than the diameter of the bolts. However, this average 0.lmm gap
around the bolts has a much smaller effect than the much larger gap at the bottom.
The uncertainty in the gap width caused a problem when comparing the test
results to the numerical simulations results. However, it should not affect the qualitative
observations.
A schematic of the test set up is shown in Fig.9.l8. The supporting steel frame is shown
behind the specimen. There is another identical steel frame in front of the specimen but is
not shown for clarity. A steel member spanning transversely between these two frames
supports the vertical load jack as shown. The completed test set up is shown in Fig.9.19.
There is another horizontal steel member, which is not shown in Fig.9.l8, in both the
front and rear steel frames at the same level as the test specimen. These two steel
0¡
members provided lateral restraint against out-oÊnlañ'¡novement of the test specimens,
-226 -
CHAPTER 9 EXPERIMENTAL WORKS
Steeì I 500
Verticaì
ìoad jack
conctete
reactron
Specimen
Axial load 400
actuator to
300
LC : Load Cell
S.iRÉSSII'IC
n'l l>RÕcRllSS
-227 -
CHAPTER 9 EXPERIMENTAL V\/ORKS
,1Ì
-¿
stra ining
k
\
The hinge support on the right hand side of the specimen that kept the test
specimen in position is shown in Fig.9.21 The hinge has two angle brackets that were
used to clamp the end of the test specimen. These angle brackets were bolted to the end
plate of the hinge that was mounted to the concrete reaction block as shown inFig.9.2l.
Fig.9.22 shows the left-hand side roller support. The supporting bracket was similar to
that of the right-hand side but narrower in order to avoid contact with the strengthening
frames installed on both sides of test specimen, as shown by Figs.9.22 and 9.23. The
-228 -
CHAPTER 9 EXPERIMENTAL WORKS
bracket was connected to a vertical load cell through a vertical frn. The load cell was then
mounted to another hinge that was fixed to the ground as shown inFig.9.22.
rl
t(
I
\
Fig. 9.22 Left hand side pin support
-229 -
CHAPTER 9 EXPERIMENTAL WORKS
However, the roller support was not mounted directlyto the top of the axial load
actuator. The actuator was not designed to take large transverse f'orces. To prevent
possible transverse loading to the actuator, a gap was maintained between the end plate of
the hinge and the end plate of the actuator as shown inFig.9.23. The gap was filled with
high strength low friction plastic membranes (TEFLON sheets) that were greased to
further reduce the friction. Therefore, the hinge support and the end of the actuator could
move relatively and freely in the transverse direction without affecting the application of
-230-
CHAPTER 9 EXPERIMENTAL V/ORKS
The vertical load was applied to a test specimen through the vertical load jack
shown in Fig.9.l8. The connection details are shown in Fig.9.24. The downward load
was applied through the compression roller, as shown inFig.9.24,which satontop of the
beam stub. The upward direction load was applied through two tension rods, one in front
and the other behind the test specimen, that were anchored to the strengthening frames on
This relatively simple test set up proved to work very well in the experiments.
TENSION
ROD
I
I
-231 -
CHAPTER 9 EXPERIMENTAL WORKS
9.4 INSTRUMENTATION
A combination of clcctronic and manual instrumcntation lvas uscd to mcasurc thc vcrtical
the column reinforcement bars and steel plates, the slip of bolts, as well as shear
The loads were measured using three load cells as shown in Fig.9.l8. Load cell I
(LCl) measured the horizontal load from the axial load actuator. Load cell 2(LC2) gave
the vertical load on the left-hand hinge support, and load cell 3 (LC3) provided the
The displacements were measured at the five points shown in Fig.9.25. Point A
was used to measure the horizontal movement of the left-hand support. This horizontal
movement was monitored during each test. As excessive movement of point A, which
rcflccts thc axial shortcning of a spccimcn, signalcd thc loss of intcgrity of thc tcst
column and alefted the technicians to stop the test to avoid a drastic failure mode. It was
also used to convert the LC2 reading, which did not remain in a vertical direction when
point A moved, into vertical and horizontal components. A Linear Voltage Digital
top of a cantilever column (i.e. Ao shown in Fig.4.1). The steel arm, as shown, was fixed
to the stub of each test specimen that acted as the base of the cantilever column.
Therefore, the stub and the steel arm together were considered as the reference ground of
the cantilever column and any relative movement to the steel arm at point E was the
lateral displacement of the cantilever column relative to the reference ground. A string
-232 -
CHAPTER 9 EXPERIMENTAL V/ORKS
pot was used at point E, as the maximum travel distance of the available LVDTs was not
00mm
C
A oao O¡
':-+ :r'-- o <>
<> -o4
r OO O¡
aa aa o
E
D
Steel
B l00mm
Points C and D together were used to calculate the rotation of the beam stub. The
beam stub rotation was needed in order to convert the vertical and horizontal forces into
the axial and lateral (shear) forces of the column. Details of these conversions are given
in Section 9.6. LVDTs were used at these two points (C & D), as shown in Fig.9.26.
Point B measured the movement of a point at the bottom of the stub with respect to the
ground. A string pot was again used at point B. The displacements measured at points B,
column, which is derived in Section 9.6. lt was found from tests that the calculated
the acquisition of data and as there was no direct measurement at point E in the first few
tests, it provided confirmation that the specimen deformations calculated using the
-ZJJ-
CHAPTER 9 EXPERIMENTAL WORKS
The strains in the reinforcing bars were measured at three key locations as shown
inFig.9.27(a). One strain gauge was installed on both upper and lower longitudinal bars,
approximatcly 50mm away from the face of the beam stub. The third strain gauge was
installed on the first stirrup that was located 100mm way from the face of the stub. The
accuracy of the strains measured by this method is limited as only one strain gauge was
provided on the face of the reinforcing bar. When the strain inside a reinforcing bar is not
uniform, as occurs under bending of the bar, significant differences in strain reading may
be expected between the strain gauge attached on the internal face of the bar and on the
external face of the bar (difference between compression face and tension face of the
bar). However, the strain measurement was not a prime target in this work and the strain
-234-
CHAPTER 9 EXPERIMENTAL V/ORKS
Two strain gauges were installed on each steel plate, one on the external face and
the other on the internal face, as shown inFig.9.26 andFig.9.27(b). Their position was
also about 50mm away from the face of the beam stub.
c ''l'l"L
pper gauge
Lower internal strain
The shear deformation of the column was measured in two of the tests to find out
employed three LVDTs installed in the bottom 300mm segment (expected plastic hinge
zone) of the cantilever column. This set up could only measure the shear deformation
within the 300mm long segment. As the shear deformation is mainly due to shear cracks
insignificant in the rest of the column. Observations from the tests confirmed that no
visible shear crack occurred outside the bottom 300mm segment. Detailed calculations of
-235 -
CHAPTER 9 EXPERIMENTAL WORKS
the shear deformation from these measurements are given in Section 9.6. As the shear
deformation was found to be insignificant, even in the highly stressed and deformed
plastic hinge region, no other specimen was measured for shear deformation, thereby
The slip of the steel plate (relative movement between the plate and the column)
was measured in one (specimen IBMP6) of the tests at three points along the length of
the cantilever column. One slip measurement point was at the top of the plate, one at the
middle of the plate and the other at the bottom of the plate. Three LVDTs, were mounted
on the face of the steel plate to measure the relative movements between these 3 points on
the steel plate and the corresponding side face of the RC column as shown in Fig.9.28.
These relative movements were considered to give a good indication of the amount of
slip between the steel plate and the face of the RC column. However, strictly speaking
slip is dehned as the relative movement at the same point on the interface.
-236-
CHAPTER 9 EXPERIMENTAL WORKS
I,VDT 2
I,VI}T I
-237 -
CHAPTER 9 EXPERIMENTAL V/ORKS
9.5 LOADING
Thc axial and latcral loads wcrc applied to each test specimen using a 50 tonne lnstron
hydraulic actuator (Model No Al01l5E) and a 50 tonne hand operated double acting
c
hydraulic ram, respectively. The centernof the two hinges at the ends of a test specimen,
section. This ensured that the axial Ioad was applied at the centroid of the RC section at
Strictly speaking, the method of loading, i.e. loading path and rate, will affect test
results for non-lìnear structures. However, the possible influence of the loading procedure
was not significant in this work for the purpose of assessing the effect of plating, as long
as the relevant specimens, that were used for comparison, were loaded in the same way.
A requirement for this project was that the axial load was to be maintained constant
throughout each test. However, in order to obtain the softening branch of a lateral force
(y-axis) versus displacement (x-axis) response curve for a column, tests must be
conducted under ¡-axis (displacement) control mode, i.e. to change the x co-ordinate
value and measure the corresponding y co-ordinate values. Therefore in these tests, the
constant axial load was first applied to each specimen under load control mode, i.e. the
horizontal movement of the axial load actuator shown in Fig.9.l8 was automatically
adjusted to maintain (within a tolerance of +0.2kN) a constant axial load of 360kN. Next,
the vertical loading jack shown in Fig.9.l8 was pumped manually to impose a specific
-238-
CHAPTER 9 EXPERIMENTAL V/ORKS
displacement, readings from all instruments were taken. This process produced one point
in each response curve. The 2nd and 3'd steps were repeated until the whole response
However, due to inexperience with such tests as well as safety concerns that the
very large axial load (360kN or 22.5Yo of actual concrete squash load) could cause
instability of the specimen andlor the test rig, the axial load actuator was actually
operated under a displacement control mode in the initial monotonic loading tests. In a
test where the displacement of the axial load actuator is fixed, drastic failure is not likely
to occur, as the axial force will be released immediately should instability occur. Thus,
for the initial monotonic test the loading procedure was, first, the axial load was applied
and the corresponding axial displacement was fixed. Second, the specified lateral
displacement was applied and then instrumentation readings were taken. With the change
in lateral displacement, small changes in axial load occurred. Hence in the third step, the
axial load was re-adjusted, through additional movement of the axial load actuator, to
restore it to the desired axial load value. The readings were then recorded which gave the
correct point in the response curve corresponding to the correct axial load. The process of
applying lateral deflection and adjusting axial load continued until the whole response
was considered to have a minor effect on the column response because small loading
steps were used for the lateral deflection and so the adjustment of axial load was small
-239 -
CHAPTER 9 EXPERIMENTAL WORKS
Furthermore, from the monotonic tests, it was found that the testing system was
actually quite robust. With increased confidence in the safety and the test procedure, it
was decided to use load control for the axial load in the remainder of the tests.
It is well known that loading rate can influence experimental results. Abrams (1996)
observed that damage was much more prominent at a slower loading rate because of
increased crack propagation. It is also established in the literature that while fast-rate
loaded specimens tend to have larger ultimate loads (strengths) than slow-rate loaded
specimens, fast-rate loaded specimens tend to lose strength at lower ductility ratios.
In the monotonic tests of this work, several seconds (5-30 seconds) were allowed
between each load step to wait for the specimen to set and cracks to develop. However, as
the load steps were controlled manually, the time intervals were not constant, especially
when the specimen was inspected and photos were taken. However, it is believed that a
For cyclic loading tests, which were conducted as quasi-static loading, the loading
rate was similar but slightly faster (about 5 seconds between loading increment). A
marginally faster loading rate was used to save time due to the large number of points to
be tested. In fact, one of the cyclic tests lasted for 3 days. When a test could not be
finished in one day, the test was stopped at a zero displacement position. The axial load
actuator and vertical jack were then un-loaded to zero force, and the specimen was left
overnight. The test resumed the next day by first applying the axial load and then
-240 -
CHAPTER 9 EXPERIMENTAL V/ORKS
The purpose of the quasi-static cyclic loading tests was to simulate the response of a
method of load cycling in order for the test results to be useful for inferring the likely
seismic behavior for the specimens. Many different cycling loading procedures have been
(Liddell, D.P. et al. 2001; Liddell 2000) to study the influence of applied loading history
Japan and New Zealand, and artificially generated histories derived from recorded
earthquake ground motions were considered. The following conclusions were made from
their research:
o There is little agreement between research institutions throughout the world regarding
. The New Zealand loading history replicates the earthquake demand of a reinforced
concrete member more closely than loading histories from other research institutions.
Because of the above reasons, the New Zealand loading history, which is shown
in Fig.9.29, was adopted in this project. The ductility factor in Fig.9.29 was defined as
the ratio of the displacement to the yield displacement. There were different definitions in
the literature for yield displacement. In this work, the yield displacement is defined as the
-241 -
CHAPTER 9 EXPERIMENTAL WORKS
Step number
From the monotonic test of the un-plated column, it was found that the yield
displacement was 26mm, which is exactly the peak point of the response curve as shown
in Fig.1l.l. The column axial instability (difficult to maintain axial load) occurred at a
*l04mm, -l04mm. This will give only three displacement steps of 26mm, 52mm and
l04mm, which raised the concern that not enough load steps would be tested. To give
more load steps, the New Zealand loading history was modified in this work to give a
loading history shown in Fig.9.30. This loading was used for all cyclic loading tests
benchmark for comparison. However, not all specimens were able to withstand all the
-242 -
CHAPTER 9 EXPERIMENTAL WORKS
91 35
7B 3.0
?65 2.5
2.0
->39 15 o
=26 10 o
b13 05
(ú
Eo
O
o::-1a.
(E -zÞ -1.0 =
-o. -39 -1 5
o
.2 -s2 -2.0 o=
o -65 -25
-78 -3.0
-91 -3.5
Step number
As the double-column test specimens were tested horizontally, the gravity load of the test
specimen including the strengthening frames was expected to cause a small initial
deflection in the column. For convenience of data acquisition, all instrumentation
readings were set to zero after the specimen had been placed in position but before the
application of any axial and lateral loads. In other words, the effect of initial gravitational
loading was recorded as zero in the test. The self-weight can be corrected in the response
The first data point reported in the test results is the point when the axial load was
increased to the specified load and before any movement of the vertical load jack. At this
point, the response of the cantilever column, in terms of both lateral force and
displacement on top of the column, may not be zero due to flexure, especially for the
plated columns. This is because the axial load, which was applied to the centriod of the
-243 -
CHAPTER 9 EXPERIMENTAL WORKS
As notcd carlicr in Scction 9.4, not all of thc kcy column rcsponscs could bc mcasurcd
directly. The conversions of the measurements are derived in this section for data that had
The column lateral force F is the force component at joint B that is parallel to the long
\-
side of the stu{, as shown in Fig.9.31. It can be calculated from the readings given by
load cell{, 2 and 3, as indicated by Ft, Fz and F3 in Fig.9.3l. The rotation angle g of the
¡'
beam stub is calculated from displacement measurements at points C and D as shown in
Fi9.9.25.
F¡
Lt Lz
l. N
Fr) a a B
-^-_ -44 aF---
J.
a
aa aa . . rl F horizontal
F a
a force F1,
force Fu
F,=F'Ltl(L,+Lr) (e.l)
Therefore,
-244-
CHAPTER 9 EXPERIMENTAL WORKS
N = Fn'cos(9)+{.sin(á) (e.4)
The deflection of a cantilever column can be measured directly through the steel arm
shown in Fig.9.25. However as noted earlier, this steel arm was not installed in the first
few tests. Therefore, it had to be calculated from other measurements. As the top of the
column is supported by a hinge, the deflection of the column, which is the relative
displacement of the hinge to the base of the cantilever column, was determined by the
movement of the central stub which was measured at points B, C and D shown in
Fig.9.25. The central stub was assumed to be a rigid body in these calculations.
The measurement at point B was a string pot with one end fixed to a point B on
the ground and the other to a point P at the bottom of the stub, as shown in Fig.9.32. The
movement of point P, from P to P' as shown, was measured by the change in length of
the string, from LBs to LB. The measurements at points C and D were given by two
LVDTs that were mounted to the supporting frame which was considered to be fixed with
respect to ground. These two LVDTs give the horizontal movements of the stub atthese
column, were first derived in terms of LB, AC and AD. The lateral displacement of the
-245-
CHAPTER 9 EXPERIMENTAL V/ORKS
d'
I
o
-l
DY
I
d'
I
e d'
d,
cl D
/f
MN a
Fig. 9.32 Calculation of lateral displacement
LC + A,D =24-tan9
or
-246 -
CHAPTER 9 EXPERIMENTAL V/ORKS
DX =DY.Ian?+(LD-LC)/2 (e.8)
and
where
and
AD- AC
b- dl + dl .sin(ø + 0) - d3 + (e.1s)
2
-247 -
CHAPTER 9 EXPERIMENTAL WORKS
One of the two solutions given by Eq.9.l6 is false. It can be found by letting
DY = LBo+ LB (9.17)
Obviously, the correct answer is DY - LBo - LB . Therefore, the only reasonable solution
of Eq.9.l6 is
From Fig.9.33, the lateral displacement at top of the cantilever column can be
calculated as
reproduced in Fig.9.34. Lines AB, CD and BC in Fig.9.34 represent the positions of the
top, bottom and diagonal LVDTs, respectively. Deformation within the rectangle ABCD
-248 -
CHAPTER 9 EXPERIMENTAL WORKS
is measured by these three LVDTs. In this section, the shear deformation of the measured
0.5 0.5
A A' B
before deformation
h
after deformation
C'C ED D'
Bt'
C D
W
Under pure flexural deformation without shear, the deformation of the measured
rectangle is shown by Fig.9.34(a), where the un-deformed shape is given by ABCD and
the deformed shape is shown by A'B'C'D'. The flexural deformation changes the length
of the top and bottom sides as well as the diagonal line, causing a change of readings in
all three LVDTs. Under pure shear deformation, as shown by Fig.9.34(b), the un-
-249 -
CHAPTER 9 EXPERIMENTAL V/ORKS
deformed rectangle ABCD is skewed to the deformed shape AB"CD". In this case, the
side length of the deformed shape is still equal to its original length, i.e. AB":AB and
CD":CD. In other words, the readings from the top and bottom LVDTs measure only the
flexural deformation. However, readings of the diagonal LVDT include both flexural and
shear deformation.
diagonal LVDT reading { is equal to the summation of diagonal length change due to
-250 -
CHAPTER IO MATERIAL TESTS
Tests were conducted to obtain the actual (average, or mean, values as opposed to the
characteristic, or design, values) material properties for the concrete, reinforcing bars,
steel plates, anchor bolts and for the full interaction gluing used in the fabrication of the
test specimens. These properties were needed for the analysis of the test results,
especially in the numerical simulations of the test columns. The methodology and results
10.1 CONCRETE
All four double-column test specimens were cast on 7th July 2000 from a single batch of
concrete supplied by CSR Readymix. The required design concrete strength was
specified to be 25N/mm2 with the expectation that the actual strength would be higher.
Concrete samples of cylinders (þl00mm x 200mm high) were taken from the same batch
of concrete. To estimate the actual compressive strength, tensile strength and Young's
modulus of the concrete for each column specimen as accurately as possible, cylinders
were tested around (but not exactly on) the test date of each column test.
Typically, two tests were conducted on each cylinder: (l) a Young's Modulus test
and (2) a compressive strength test. The Young's Modulus tests were performed first,
which involved application of a compression load to a cylinder in the elastic stress range
to establish the stress vs. strain properties of the concrete. The maximum stress applied in
the Young's Modulus test was approximately 40%o of the compressive strength of the
-251 -
CHAPTER IO MATERIAL TESTS
concrete. The slope of a linear trend line that gave the best fit to the stress-strain curve
was considered as the Young's Modulus. The results of the Young's Modulus tests are
given in column 9 of Table l0.l and plots of the tests are shown in Figs.Al to 417 in
Appendix A. After each Young's Modulus test, the cylinder was loaded in compression
until failure in order to determine the compressive strength of the concrete. The results of
To estimate the tensile strength of the concrete, Brazilian tests, or indirect tensile
tests, were conducted. As the tensile strength was not a major concern in this work, only
two such tests were performed. The test results are given in Table 10.2, in which the
tensilestrengthwascalculatetlbyr,=ffiandislistedincolumn8.
-252 -
CHAPTER IO MATERIAL TESTS
Three tensile tests were conducted to determine the tensile strength for both the Yl6 and
R6 reinforcing bars. The detailed results are given in Table 10.3.
Typical machine displacement vs. tensile force plots for the Y16 and R6 bars are
shown in Figs.l0.l and 10.2, respectively. The "yield force" in Table 10.3 is the y
(vertical) co-ordinate of the yield plateau in the displacement vs. tensile force chart. The
The above tensile tests give the yield and maximum strength of the reinforcing
bars, as listed in Table 10.3. However, it was not possible to determine the stress-strain
relation, or the Young's modulus and strain hardening stiffness, from this data, as the
displacement or horizontal co-ordinate of Figs.10.1 and 10.2 did not give the true
-2s3 -
CHAPTER IO MATERIAL TESTS
elongation of the bar due to slip of the bars in the end grips. To determine the stress-strain
relation, another three specimens of the Yl6 bars were tested with strain gauges installed
on the surface of the bars. The results of these tests are given in Figs.l0.3-10.5.
Force (kN)
¡1(,3
kâ(
I 00.
50
Di
20
Displacement (mm) 1
-254-
CHAPTER IO MATERIAL TESTS
600
500
N
tr
Ë 4oo
ã soo
.n
o
õ 2oo
100
0
0 0.05 0.1 0.15
Strain
600
534.3
500
N
E 400
E
z first yield strain 0.0026
300
o
tt,
o
200
Ø
y= 190004x+ 0.3531
100
0
0 0.001 0.002 0.003 0.004 0.005
Strain
-255 -
CHAPTER IO MATERIAL TESTS
600
500
À-_-
N
E 400
E
ào soo
U,
Ë
U'
zoo
100
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain
600
500
E 400
E first yield strain 0.0025
z 300
o
aD
o /
200
U'
y= 190290x- 0.184
100
0
/
0 0.001 0.002 0.003 0.004 0.005
Strain
-256-
CHAPTER IO MATERIAL TESTS
600
500
E 400
E
ào .oo
o
o
.b
U'
200
100
0
0 o.o2 0.04 0.06 0.08 0.1 0.12
Strain
600
532.8
500
N
E 400
E first yield strain 0.0026
z 300
o
o
o
200 ,/
a y = 191182x+ 1.2371
100
,/
0
,/
0 0.001 0.002 0.003 0.004 0.005
Stra in
-257 -
CHAPTER IO MATERIAL TESTS
From the stress-strain tests, the yield strength of the three Y16 specimens were
found to be 534.3 , 521.4 and 532.8 MPa, respectively, giving an average value of 529.5
MPa. The Young's Modulus is the gradient of the linear part, as shown by the coefficient
of x in the linear equation of the trend lines. The average value of the Young's Modulus
It is noted that the yield strength values from the stress-strain tests are consistently
lowerthan those given in Table 10.3 that have an average value of 548.8MPa. The reason
is that the cross-sectional area of the reinforcing bar was slightly reduced by grinding the
surface of the bar to make a smooth surface for strain gauge installation. The reduction in
the cross-sectional area was difficult to measure. Therefore, the nominal cross-section
area of 201.06 mm2 without reduction was used to calculate the yield strength and
estimated using the average yield strength from the stress-strain tests, 529.SMPa, and the
average yield strength from the normal tensile tests, 548.8Mpa. If 548.8MPa is
considered as the actual yield strength of the Y16 bar, then the reduced cross-sectional
more "accurate" cross-sectional area, the Young's Modulus from the stress-strain tests
The strain hardening stiffness of the Yl6 bar was estimated from the three stress-
strain curves, as shown in Fig.l0.6. The simplified strain hardening line was drawn
between a point in the yield plateau and a point near the peak of the curve as shown in
Figure 10.6, which gave a strain hardening stiffness of 798.4 MPa. For consistency, it
-258-
CHAPTER IO MATERIAL TESTS
was also revised using the more "accurate )) cross-sectional area to give
100
0
0 o.02 0.04 0.06 0.08 0.1
Strain
In summary, the average yield strength from the Yl6 bar tensile tests without a
strain gauge was used as the yield strength of the Yl6 bars. The Young's Modulus and
strain hardening stiffness values were obtained from the average stress-strain test results,
which were modifred with the "accurate" cross-sectional area. For the R6 bar material,
the average yield strength given in Table 10.3 was used. The final test results are
-259 -
CHAPTER IO MATERIAL TESTS
Three types of steel plates were used for plating the columns in the experimental work.
To obtain the steel plate properties, tensile tests, similar to those performed on the
results for the 12mm thick mild steel plate (MSl2), 6mm thick mild steel plate (MS6) and
6mm thick high tensile steel plate (HT6) are given in Figs.10.7, 10.8 and 10.9,
respectively. Three specimens were tested for each type of steel plate, corresponding to
For the mild steel plates, i.e. MS12 and MS6, the yield strength was taken as the
averagey co-ordinate of the yield plateau from Figs.l0.7 and 10.8, respectively. For the
high strength plate HT6, the yield strength was taken at the first yielding point shown by
"A" in Figl0.9. The Young's modulus was taken as the gradient of the linear segment, as
shown by the x coefficient for the linear trend lines. The average values of the three
specimens were used for the steel plate properties and are summarized in Table 10.5. The
strain hardening stiffness depends on the position of the last point on the stress-strain
curve. As this value may not critical in this study, an arbitrary value of 600MPa was
used.
-260-
CHAPTER IO MATERIAL TESTS
400
ôt
tr
300
È
z
u,
Ø
200
o
L
ö 100
0
0 0.02 0.04 0.06 008
Strain
300
N
E
E
> 200
y = 198056x - 0.5291
o
(t, y = 198246x+ 0.8171
Ë loo
U'
y = 199734x+ 0.771
0
0 0.0005 0.001 0.0015 0.002
Strain
-261 -
CHAPTER IO MATERIAL TESTS
400
N
tr
300
Ë
=o
o
200
o
¡-
õ 100
0
0 o.Qz 0.04 0.06 0.08 0.1 0.12
Strain
400
Ç soo
€
3o zoo
o
o = 198714x + 0.5602
1oo
ã y=200111x+1.0084
y=201269x+0.9476
0
0 0.001 0.002 0.003
Strain
-262 -
CHAPTER IO MATERIAL TESTS
800
N
E 600
E
z 400
o
U'
o
L
tt 200
Ø
0
0 o.o2 0.04 0.06 0.08
Strain
800
ôt
E 600
E
z ,/
-Ø 400
Ø
o v = 198909x + 0.7471
L v = 195924x + 2.0889
õ 200
y=195561x+0.7625
0
0 0.002 0.004 0.006
Strain
-263 -
CHAPTER IO MATERIAL TESTS
The shear force vs. deflection (slip) relation is needed to define the properties of the shear
connectors on the interface of the plated columns. The shear connector properties are not
only affected by the properties of the bolt but also by the properties of the concrete and
steel plate. Therefore, bolt shear tests, as illustrated by Fig.10.10, were conducted to
determine the shear connection properties. The actual test set up is shown in Fig.l0.1l.
The concrete blocks were cast from the same batch of concrete that was used to fabricate
the column specimens. Hence, the material properties were considered to be the same.
displacement
steel transducer
slip s
force 2Fo
,-
\
concrete block
-264-
CHAPTER IO MATERIAL TESTS
To identify a suitable bolt for bolting plates to the test columns, three types of bolt
were tested. These are shown in Fig.10.12. Two of the bolts are common Hilti bolts: the
Hilti HSL $12 bolt and the Hilti HIS Ql2 bolt. The Hilti HSL 012 bolt is a mechanical
bolt that relies on mechanical interlocking from the end expansion to provide the anchor
force. The Hilti HIS 0l2bolt is a chemical bolt. It relies on chemical bond to provide the
anchor force. Hilti HIT HY 150 adhesive is used as the bonding agent for this bolt.
The third type of bolt considered was a threaded rod bolt. It was cut from Q12
\
threaded rod of high tensile steel. The internal diameter of the threaded rod was 9.853mm
with an effective diameter of 10.863mm. Tensile tests similar to those conducted on the
reinforcing bars without strain gauges were conducted to determine the yield strength of
the threaded rod, which is given in Fig.l0.l3. This bolt was anchored by chemical bond
'-.1,j A,.
/'(
-265 -
CHAPTER IO MATERIAL TESTS
900
800
700
o
o- 600
= 500
o
o 400
o
at,
300
200
100
0
05'10 1520
Displacement (mm)
The installation procedure for the two Hilti bolts followed the standard procedure
as per specifications. The installation procedure for the threaded rod bolt was described in
Section 9.2.2. All types of bolt were tightened with 25 Nm torque using atorque wrench.
The embedment length of all the bolts was 70mm from the face of the concrete to the end
of bolts.
In the testing as shown in Fig.lO.ll, a pull force was applied by the hydraulic
actuator to the steel plate in a displacement control manner. The displacement was
applied at a (slow) rate of about l-3mm/per minute. The applied force and the movement
of the steel plate (slip) were then recorded using a load cell and displacement transducer,
respectively. This data was then used to produce the load vs. slip plots.
The Hilti HSL bolt was tested on the 2nd November 2000 when the concrete age
was 116 days. The Hilti HIS bolt was tested on the 1Oth November 2000 at a concrete age
of 124 days. The threaded rod bolt test was conducted on 7th November 2000 at a
concrete age of l2l days. Figures 10.14, 10.15 and 10.16 show the bolt slip test results
for the Hilti HSL bolt, Hilti HIS bolt and threaded rod bolt, respectively.
-266-
CHAPTER IO MATERIAL TESTS
80.0
00
0.0 5.0 10 0 15.0 20.o 25.0
Slip (mm)
50
40
z-Yõ
ËË'
€o¡
LL')A
ooLv
Ês
U'
10
0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.O
Slip (mm)
-267 -
CHAPTER IO MATERIAL TESTS
80.0
?o^ 600
f) tt
.¡f o 4o.o
\
.ct
L
.! ñl
OL
fi $ zoo
0.0
0.0 5.0 10.0 15.0 20.0
Slip (mm)
In all the three tests, the bolt on the far side of the loading arm was slightly pulled
up in the final stage of loading, as shown in Fig.l0.l7 where the Hilti HSL bolt was
pulled up about 5mm prior to failure. The Hilti HSL bolt test ended with the breaking of
the concrete block, as shown in Fig.l0.l8 and Fig.l0.l9. Since most of the important
curve for this bolt had already been obtained, this test was not repeated. However,
additional measures were taken to prevent concrete block failures in the remainder of
these tests. Two strong steel rods were added between the two corner anchor points that
The Hilti HIS bolt test failed with the shearing off of the bolts below the plate in
the concrete layer as shown in Fig. 10.21. From Fig.10.l 5, it can be seen that the two bolts
did not fail simultaneously as suggested by the two drops on the curve. It seems that the
268 -
CHAPTER IO MATERIAL TESTS
first bolt (closer to the load cell) snapped first and the second bolt broke about a minute
later. The failure mode of the threaded rod bolt was similar, as shown in Fig.10.22.
269 -
CHAPTER IO MATERIAL TESTS
Fig. 10.19 Hilti HSL bolt and concrete block after test
Aclditional steel
+
concrete block
-270-
CHAPTER IO MATERIAL TESTS
-271 -
CHAPTER IO MATERIAL TESTS
From the bolt test results of Figs.10.l4-10.16, it can be seen that the threaded rod
bolt is the strongest and stiffest. From a ductility point of view, the Hilti HSL $12 bolt is
the best. From the theoretical studies in this thesis, it has been shown that the stiffness
and strength of bolts are more important to plated columns than the ductility which is not
critical since columns fail (concrete crushes) before excessive slip takes place. Therefore,
it was decided to use the threaded rod bolts for column plating. Furthermore, the threaded
As discussed in Section 4.1.4, the load vs. slip relation is represented by a bi-
linear model in the numerical simulations. The bi-linear "best fit" curve for the original
test data of the threaded rod bolts is shown in Fig.10.23, from which the properties of the
bolt shear connector were obtained and which are summarized in Table 10.6.
40
-272-
CHAPTER IO MATERIAL TESTS
As one test column was retrofitted by gluing and bolting a plate to the face of the column,
this connection was also tested. The same setup as described in Section 10.4 for the bolt
connector tests was used for this test. The actual test specimen is shown in Fig.10.24
where the cross-hatched area indicates the section of the plate (200mm x l00mm) that
was glued. The brand of glue used was Ciba Performance Polymers Araldite LC 340 with
Hardener LC340. Before gluing, the face of the steel plate was sand blasted to provide a
better bond. The concrete face was roughened with a needle gun. Threaded rod bolts,
applied as described in Section 10.4, were used for bolting. The glued area and the two
bolts were equivalent to one row of bolting in the actual plated column at a same cross-
section.
-273 -
CHAPTER IO MATERIAL TESTS
The test was conducted on 9'h November 2000 at a concrete age of 123 days. To
get the softening branch of the load vs. slip response curve, a displacement control
procedure was adopted for loading. Because there is essentially no slip with a glued plate
before the bond breaks down on the interface, the movement of the loading arm of the
hydraulic actuator had to be very slow. A loading rate of 3mm/per hour (in contrast to l-
3mm/minute for the unglued bolting tests) was tried initially but this was eventually
The movement of the actuator actually did not produce any slip on the interface
before the bond break down. Instead, itproduced an elongation of the steel plate and the
lateral deformation of the supporting frame that fixed the concrete block to the ground.
Actually, it is the increase in deformation of the test set up that gives a continuous
increase in the test load. As the external force from the hydraulic actuator must be
balanced by the reaction force from the supporting frame, the more the suppofting frame
deforms, the more reaction load is applied. The test result is shown in Fig. 10.25.
120.O
100.0
80.0
zJ
!ttú 60.0
o
J 40.0
20.0
0.0
00 5.0 10.0 15.0 20.0 25.0 30.0
Slip (mm)
-274-
CHAPTER IO MATERIAL TESTS
In the test, the bond broke down suddenly at a load of l08kN. Up to this point
negligible slip had been measured at the interface. The test then continued at an increased
loading rate of 3mm/per minute until the bolts failed in shear as observed previously.
After the glue bond broke, the system behaved similarly to the threaded rod bolting
system giving a similar result as can be seen by comparing Fig.l 0.25 with Fig.10.16.
The failed specimen is shown in Fig.10.26 and 10.27. The failure plane shows that the
gluing on the front area near the loading side is partly debonded, as indicated in
Fig.10.27. The rest of the plane failed due to concrete peeling off.
-275 -
CHAPTER IO MATERIAL TESTS
The conclusion from this test was that the average bond stress by this type of
This is a very rough but simple estimation. However, it is sufficient for estimating the
minimum glue length (or area) for full interaction plating. At full yielding of the steel
plate, the minimum glue length for the 12mm thick steel plate can be estimated by
As the full glue lengh is 950mm as shown in Fig.9.8, this bond length is considered
adequate. The column test result that will be presented in Section ll.2 confirmed this
-276-
CHAPTERIICOLUMNTESTS
Six column tests, consisting of three monotonic static tests and three quasi-static cyclic
tests, were conducted to investigate the effectiveness of the new retrofitting scheme as
well as to verify the results of numerical simulations. The details of the testing
procedures are given in Chapter 9. The test process, observations and test results are
reported in this chapter. The discussion of the test results will be presented in Chapter 12.
The notation used to identify each of the six test specimens in the following sections is
defined in Section 9.1. The test data, such as displacements and strains, were collected at
specific locations that are indicated in Section 9.4 where it is also noted that the strain
reading shall be treated as a reference only and may not be accurate. A constant axial
The first test consisted of a monotonically loaded test of the "benchmark" reinforced
concrete column without any plating. It was tested on22"d February 2001 at a concrete
age of 228 days. The applied load versus deflection response of the column is given in
Fìg.11.1, where it can be seen that the maximum load (peak point) was 36kN that
occurred at a lateral displacement of 25.7mm (2.1% drift). The measured strain values of
the longitudinal reinforcing bars and stirrups are shown in Fig.l1.2, where it can be seen
that the yield point of the tension reinforcement coincided with the peak point of
-277 -
CHAPTERIICOLUMNTESTS
Fig.B.l to Fig.B.6 in Appendix B. The number marked on the small black board and on
the cracks of the specimen in the photos, e.g.4mm in Fig.B.l, is the string pot reading at
the bottom of the stub that was not the actual lateral deflection of the column at top. (As
noted in Section 9.4,the actual column deformations were calculated using the data from
three separate displacement measurements at pol,tfn, C and D.) The actual lateral
V
During the test, fine tension cracks in the concrete were noticed at about 5mm of
lateral displacement (0.4%o drift). First signs of significant distress (early signs of
crushing and spalling) of the concrete in the compression side occurred at about 20mm of
lateral displacement (1.6%o drift). The specimen failed due to concrete crushing at a
displacement of approximately l20mm (10% of drift), when the axial load could no
Front and rear views of the failed specimen are shown in Figs.l1.3 and 11.4,
respectively. The final crack pattern and extent of concrete crushing are highlighted in
Fig.l 1.3 where it can be seen that the tension crack opening \¡/as concentrated in the first
crack about l00mm away from the face of the stub, and the extent of concrete crushing
-278-
CHAPTERIICOLUMNTESTS
240
-v
CL
o30
/
/A*
t!
8zo /
o
E10 {
o
o
Jo
0 20 40 60 80 100 120
0.o2
0.016
.E 0.012 bar
tú
0.008 stirups
o -compression
bar
IE 0.004
^ct
o
É. 0 -tension
-0.004
6ô 80 100 1
-0.008
0.003
0.002
T
.=
t!
0.001
t
Èo
g -o.oo1
o
É. 4.002 bar
stinups vieldin
-0.003
-compression
þ¿¡
-0.004
-fs¡sie¡
Lateral displacement at top (mm)
-279 -
CHAPTER II COLUMNTESTS
__r_v
'a
71 a<,
-280-
CHAPTERIICOLUMNTESTS
This was a monotonically loaded test of the column with a l2mm thick mild steel plate
glued and bolted on the compression face (full-interaction plated column). It was tested
on 13th March 2001 ata concrete age of 247 days. The load versus deflection response for
the column is given in Fig.l 1.5 where the maximum load (peak point) reached 55.3kN at
a lateral displacement of 20.7mm (1.7% drift). It is noted that the initial displacement of
the column was a negative value. As mentioned in Section9.5.4, the first data point was
recorded after the application of the axial load and before any movement of the vertical
load jack. Therefore, the negative displacement of the first data point was caused by the
rotation of the stub due to flexure in the member. The flexural moment was caused by the
eccentricity of the axial load that was applied at the centroid of the RC cross-section
The measured strain values of the longitudinal reinforcing bars and stirrups are
shown in Fig.l1.6, where the tension reinforcement yielded (a.r:0.0026) at about 3Omm
of lateral displacement which occurred after the peak point of Fig.l1.5. The possible
reasons for this delay in yielding will be discussed in Section l2.l.l. It can also be seen
in Fìg.l 1.6 that the compression reinforcement was loaded into tension at about 22mm of
lateral displacement, which was predicted for columns with a strong plating system by
the theoretical study in Section 5.2.2. The compression bars yielded in tension after the
76mm (a,r:0.0035).
-281 -
CHAPTERIICOLUMNTESTS
Figure I 1.7 shows the strains that were recorded in the steel plate on the external
face and internal face (interface between concrete and plate) near the base of the column.
The steel plate started yielding on the external face at a lateral displacement of about
65mm, after which the strain on the external face increased quickly. The internal face of
the steel plate remained elastic before the strain gauge broke at about 56mm of lateral
displacement. The shear deformation measured from the bottom 300mm long segment
direct proportion to the lateral displacement and reached a maximum value of about
8.5mm at the end of the test that was equal to about 6.8% of the total lateral displacement
of the column.
Appendix B. As noted earlier, the numbers marked on the small black board and on the
cracks of the specimen in the photos were the string pot readings at the bottom of the
stub. The actual lateral deflections are given in the respective title of figures. The overall
crack pattern (rear view) at the completion of testing is shown in Fig.ll.9. A close-up
\'+/
view of the major cracks (front view) fi{shown in Fig.l 1.10.
During the test, fine tension cracks in the concrete were first noticed at about
2mm of lateral displacement. Early signs of shear cracking were observed at about lOmm
displacement. Signs of plate debonding at the top of the column occurred at about 50mm
throughout the remainder of this test. Testing was stopped when the maximum travel
distance that the setup allowed was reached at the displacement of l25mm. No excessive
-282 -
CHAPTERII COLUMNTESTS
sign of distress in terms of instability was observed at any stage during this test except for
one major tension crack of more than 5mm width, as shown in Fig.B.l2.
Extension of tension cracks into the stub, or column base, was evident, as shown
in Fig.ll.l0. This observation indicates that yield penetration into the support and
rotation of the bottom cross-section as relative to the stub occurred in the test. Minor fìne
cracks between the steel plate and concrete were found at the bottom of the column (see
enlarged corner view in Fig.l l.ll) which indicated possible slip at that region, but no
\Àlð-'å
sign of concrete crushing wøEnoticed.
Therefore, it was concluded that the ultimate displacement of this test specimen
was not reached during the test. The specimen was still intact in the end of the test.
-283-
CHAPTER I I COLUMNTESTS
zlÉ
CL
o
G
o 30
o
o 20
E
o
(ú
.s -0
30 "". 50 90 110 1
(!
ID
-0
bar
o
.c¡ stirrups
o -0
É. -***lension bar
-compression
-0
.s
(!
at,
t! compresston
¡¡ stirrups
o
É,
----- -- tension bar
-284-
CHAPTER II COLUMNTESTS
0. side
-external
----_ internal side
.s
ag
6py
-10
.10
30 50 70 90 110 130
.E
(!
o
o
IU
side
CL
-external
----internal side
-50 5 10 '15 20 25 30 35 40
E
E
c
o
(E
E
o
o
E
tú
o
at,
0 10 30 50 70 90 1 10 130
-285 -
CHAPTER II COLUMN TESTS
Fig. I I .10 Front view after test showing major cracks for specimen 2AMF12
-286-
CHAPTERIICOLUMNTESTS
f
.2
r2
Fig. I l.l I Enlarged view of corner for specimen2AMFI2
This was a monotonically loaded test of the column with a 6mm thick mild steel plate
bolted on the compression face (partial-interaction plated column). It was tested on 27th
March 2001 at a concrete age of 261 days. The load versus deflection response of the
column is given in Fig.ll.l2 where it can be seen that its peak strength was 39.3kN
The measured strain values of the longitudinal reinforcing bars and stirrups are
shown in Fig.ll.l3. Figure 11.14 gives the strains recorded on both sides of the steel
plate near the bottom of the column. From Fig.ll.l3, it can be seen that the tension
reinforcement yielded at the same lateral displacement as the peak load was achieved in
- 287
CHAPTERIICOLUMNTESTS
where the compression reinforcement changed from compression strain to tension strain
when the lateral displacement increased, the strain of the compression reinforcement in
this test stayed in compression and monotonically increased with the lateral displacement.
The compression bar yielded at a lateral displacement of about 26mm. The stirrup did not
yield in Fig. I I . l3(a). Figure I I .l 4 shows that the external face of the steel plate yielded
at a lateral displacement of about 46mm, and the internal face of the steel plate remained
linear elastic up to the end of the test which meant that full yielding of the steel plate did
The shear deformation within the bottom 300mm long segment is depicted in
Fig.ll.15 but it is incomplete due to failure of the instrumentation in the middle of the
test. One of the fixing points for the instrumentation broke due to debonding of the glue
The slip between the steel plate and the RC column, which was also measured for
this specimen, is given by Fig.ll.l6. The slip measurement positions are detailed in
Section 9.4. This was the only test in which the slips were measured. The maximum
measured slip was about 2mm as shown in Fig.ll.l6. This slip was just over the linear
elastic limit and much less than the breaking slip of about lOmm, as can be seen from
Fig.10.23. The slip distribution was such that it increased from the bottom of the column
toward the top of the column. The slips at the measurement points monotonically
increased with the lateral displacement except at the bottom point that showed a negative
slip at the beginning of the test caused by the initial moment from the axial load
eccentricity.
-288-
CHAPTERII COLUMNTESTS
Appendix B. The numbers marked on the small black board and on the cracks of the
q
specimen in the photos were readinlOof the response "E" (refer Fig.9.25) from the steel
arm as described in Section 9.4. This reading was intended to give a direct recording of
the actual top displacement of the cantilever column. However, the steel arm was
mistakenly fixed to the strengthening frames instead of to the stub of the column in this
test. It was found during the test that there was substantial relative movement between the
stub and the strengthening frames. Therefore, this reading did not reflect the actual
column deflection. In fact, significant discrepancy was found between this direct
measurement and the true displacement calculated from readings B, C and D (refer to
Fig.9.25 and Section 9.6.2), which is shown in Fig.l 1.17. This problem was subsequently
resolved in the remaining tests by attaching the steel arm directly to the concrete of the
stub instead to the strengthening frames. The difference between the directly measured
displacement (response "E") in subsequent tests and the calculated displacement (using
As shown in Fig.B.13, two G-clamps were used at the bottom of the column to
prevent sudden buckling of the plate in the early stages of the test. The G-clamps were
provided due to safety concerns. They were removed at the column displacement of
93mm, however, when it was felt that sudden instability would not occur even if the plate
buckled. Another concern, and reason for the removal of the G-clamps, was that they
might strengthen the plating system and possibly affect the test results. Luckily, these
clamps were not effective at the time of removal, as the contact between the legs of the
-289 -
CHAPTERIICOLUMNTESTS
clamps with the plate was still loose, and buckling of steel plate did not occur until a
displacement of 106mm was reached. The G clamps were not used in the rest of tests.
The test column failed due to buckling of the steel plate, as shown by Figs.ll.lS and
11.19. The test was stopped immediately after the plate buckled and the column was
unloaded.
Cracks in the concrete at the corner where the steel plate was bearing down were
continuously through out the test. However, the concrete was able to provide sufhcient
support to the plate until the end of the test when the plate buckled and the maximum
compression zone of the plastic hinge was noticed at a displacement of about 40mm.
Significant concrete crushing in the plastic hinge region occurred after buckling of the
steel plate, as shown in Figs.ll.lS and ll.l9. Cracking was concentrated in 3 major
-290-
CHAPTERIICOLUMNTESTS
z.Y 40
CL
o
l!
o
o
o
õ
o
t!
-10 10 30 50 70 90 110
.Ê,
tú
L
fl
o
G
.c¡
o 30 0
É.
sYl- - -
bar
'õ stinups
--.-
- -tension bar
an -compression
IE 't5
.c¡ 5 10
o
É.
-291 -
CHAPTER II COLUMNTESTS
0.
face
-external
internal face
.=o
ll,
-
at
o
s-o
,|
0 0 30 50 70 90 0
-0
E
E
o
t!
É
o
o instrumentation break
tt
G
o
E
U'
-5 5 15 25 35
of plate
------ middleof plate
-lep þelle¡ of plate
-
E
CL
30 50 70 90 0
-292 -
CHAPTERII COLUMNTESTS
!rE
OE
J
a=
EÞ
çE
lio o
õ.8
o.9
Ït
-10 l0 30 50 70 90 110
T "1
U,
^'"-*-*r. >
¡c
551
11,1
2:n O
Fig. I I .18 Enlarged rear view at displacement of 106mm for specimen I BMP6
-293 -
CHAPTER II COLUMNTESTS
This test involved applying quasi-static cyclic loading (as described in Section 9.5.3) to
the benchmark column without plating. lt was tested from the l6th to lTth of May 2001 at
a concrete age of 3l I days. The load versus deflection response for the column is given
in Fig.ll.20, where the lateral load varied from -38.7KN to +32.5KN (as compared to
35.9kN for the monotonic specimen lAMR). A positive displacement shown in the figure
indicates pushing down toward the ground from the initial position and a negative
The measured strain values of the longitudinal reinforcing bars and stirrups are
Fig.9.27 . Figure 1 L21 shows that the upper reinforcement bar reached a tensile straìn of
-294-
CHAPTER II COLUMNTESTS
0.0025 at the negative peak response point (-38.7kN, -26.4mm) of Fig.ll.20. The strain
at this point was close to the monotonic yield strain of 0.0025-0.0026. However, it needs
to be noted that these strain values do not generally indicate yielding status or stress level,
as yielding depends on the strain history under cyclic load conditions (as seen in
Fig.4.12). A tensile strain of 0.00135 was recorded in the lower bars in Fig.ll.22 atthe
positive peak point (32.5kN, 23.9mm) of Fig.11.20. The reason was not exactly clear for
the tensile strain difference between the positive and negative peaks apart from the
possible variation in the strain gauge positions as noted in Section 9.4. The stirrups
apparently yielded (the monotonic yield strain was about 0.0035) in the last two cycles as
Appendix B and Figs.l1.24-11.26. The numbers marked on the small black board in
these photos were the actual lateral deflection of the column measured from the steel arm
E (see Fig.9.25). However, the "*" and "-" signs in the pictures were opposite to the sign
convention of Fig.11.20. The correct sign is given in the respective titles of the figures.
It was noticed that there are a few sudden, but small, drops in the response curve
of Fig.11.20. These drops were attributed to the frictional force exerted on the side of the
specimen from the lateral restraints shown in Fig.9.20, which were provided to prevent
the specimen from swaying out-of-plane. The sway of the specimen can be seen in a
photograph of the specimen in Fig.1l.26 which was taken after failure. This frictional
force from the lateral restraint blocks was reduced in the subsequent tests by greasing the
contact faces.
-295 -
CHAPTER II COLUMNTESTS
IAMR), when the axial load could no longer be maintained at which point the test was
stopped. The extent of concrete crushing at failure was more than 200mm as shown in
Figs.ll.24 and 11.25. Obviously, the damage to concrete was more severe under cyclic
-60 40 -20 0 20 40 60
40
z! 30
CL 20
o
10
o
o 0
o
o -10
-20
E
o -30
G
-40
-296-
CHAPTERIlCOLUMNTESTS
,60 40 -20 20 40 60
0.004
.=
tú
-¿ / 0.002
U'
t!
l¡
f
o
o
CL
CL
ø æ
_¿
7
v 0
-0.002
U -0.004
Fig. I I .21 Measured strains in upper reinforcing bar for specimen 3ACR
€0 40 -20 0 20 40 60
0.003
'Ë 0.002
t,
l-=-- 0.001
l! ñ >---
¡t
E
o
ìo
N \------è
\\\\ -\--\
0
-0 001
-0.002
-0.003
Fig. I 1.22 Measured strains in lower reinforcing bar for specimen 3ACR
-60 40 -20 0 20 40 60
0
: -0 0005
tr -0.001
(E \---\ -0 0015
\ -\.-
o
CL
-\..\ -0 002
-0.0025
-0 003
|t, \-, -0 0035
\ -0.004
\_]
-0.0045
-297 -
CHAPTER II COLUMNTESTS
Fig. I I .24Rear view at -56.3mm displacement of 1't cycle for specimen 3ACR
F)'..ìLURt"
Fig. I L25 Front view at -56.3mm displacement of I't cycle for specimen 3ACR
-298-
CHAPTERIICOLUMNTESTS
column with 6mm thick mild steel plates bolted on both the tension and compression
faces. This column was tested from the 6th to 8'h of June 2001 at a concrete age of 332
days. The load-deflection response of the column is given in Fig.ll.27,where the lateral
The measured strains of the reinforcing bars and steel plates are shown in
Figs.l1.28-11.36. Figure 11.28 shows that the upper reinforcing bars reached a tensile
strain of 0.0042 at the negative peak point (-45.9kN, -32.3mm) of Fig.11.27. At the
positive peak point (47.4kN, 26.lmm) of Fig.l1.27, the tensile strain of the lower
reinforcing bars was 0.0023 as shown in Fig.ll.29. Although the monotonic yield strain
-299 -
CHAPTERIICOLUMNTESTS
0.0025-0.0026 of the reinforcing bar cannot be used to identify the yielding of the
reinforcement directly in the cyclic test, the range of the strain variation in Fig.1l.28 and
Fig.l I .29 clearly shows that both the upper and lower reinforcement experienced
alternative tension and compression yielding. The stirrups were apparently yielded when
Fig.ll.30. The range of strain variation in Figs ll.3l to 11.36 also apparently suggests
that both faces of both plates experienced alternative tension and compression yielding
(the monotonic yield strain was 0.0015). The steel plate was completely yielded at the
end of the test as can be seen from Fig.l I .3 I and Fig. I I .33.
The progressive damage to the specimen is depicted in Figures 8.24 T.o B.48 in
Appendix B. As noted above, the "+" and "-" signs of the displacement on the small
black board were opposite to the sign convention of Fig.l I .27 . The correct sign is given
In testing, the first sign of concrete crushing was noticed at about +39mm of
lateral displacement (see Figs.B.28 and 8.29). The crushing of the concrete continued to
develop with further increases in displacement. However, the steel plate seemed to
excessive axial shortening of the specimen was noticed in the test. The integrity of the
specimen was also evident from the response curve where no excessive degradation was
The maximum upward travel distance of the vertical jack was reached at a
displacement of -82mm. Therefore, the last 2 cycles went from -82mm to +9lmm
instead of from the intended -9lmm to +9lmm, as shown in Figs.B.44 1"o8.47. Fig.ll.37
-300-
CHAPTERII COLUMNTESTS
illustrates damage to the concrete in the plastic hinge zone and the opening of the steel
The test was concluded at this point with monotonic downward pushing of the
specimen until the maximum downward travel distance was reached. Fig.ll.38 depicts
the compression zone of the plastic hinge at the final displacement of about +150mm,
from which it can be seen that significant damage to the concrete occurred and signs of
buckling of the steel plate was evident. However, because the buckling was toward the
concrete side, no instability of the steel plate was observed. Theoretically, the steel plate
should always buckle toward the concrete side because the curvature of the plate makes it
The test was then stopped at +150mm of displacement. The extent of the final
damage to the specimen is depicted in Figs.ll.39 and 11.40. These two photos were
taken after the specimen was removed from the test rig and the loose concrete was
brushed away. Buckling of reinforcing bars can also be clearly seen in these figures.
z! 25
o
o
o 0
(ú
o -25
IE
J
-50
-301 -
CHAPTERIICOLUMNTESTS
'õ 0.01
0 0075
tt,
(g 0 005
.c¡
o
o
CL
4- .>
0.0025
0
CL =7 I
-0.0025
f
-0.005
Fig. 1 1.28 Measured strains in upper reinforcing bar for specimen 4ACP6
'õ
0.0025
an
tú
.ct 0
o
o
ì
o
-0 0025
J
-0 005
Fig. 1 1.29 Measured strains in lower reinforcing bar for specimen 4ACP6
0
.E
\* -0.001
ø
o -0.002
CL
J
-0.003
(t,
\
-0.004
strain oauge brqak
-0.005
-302-
CHAPTERIICOLUMNTESTS
Fig. I I .31 Measured strain in steel plate on upper external side for specimen 4ACP6
-90 €0 -30 0 30 60 90
.= 0.003
tú
0.0025
o
o 0.002
p 0.0015
at,
õc
o
x
o
4 0.001
0.0005
0
-0.0005
o -0.001
CL
CL
f -0.0015
o / 0.0035
CL
CL
f -0.0015
Fig. I1.33 Measured strain in steel plate on upper internal side for specimen 4ACP6
-303-
CHAPTER II COLUMNTESTS
-0.0005
o -0.001
CL
CL
f, -0.0015
Fig. I 1.35 Measured strain in steel plate on lower external side for specimen 4ACP6
o 0
G -0.0005
c
o -0.001
.; -0.0015
o -0 002
o
J= -0.0025
Fig. I 1.36 Measured strain in steel plate on lower internal side for specimen 4^CP6
-304-
CHAPTERIICOLUMNTESTS
Fig. 1 I .37 Open up of steel plate at -82mm of 2"d cycle for specimen 4ACP6
{¡:
Fig. I 1.38 Rear view at +l50mm displacement for specimen 4ACP6
-305-
CHAPTERIICOLUMNTESTS
-\ ri
" '*lå .,
Èeo
I
é
Fig. I 1.39 Rear view of plastic hinge zone after testing for specimen 4ACP6
\
j
Fig. I 1.40 Front view of plastic hinge zone after testing for specinren 4ACP6
-306-
CHAPTERIICOLUMNTESTS
This test consisted of applying quasi-static cyclic loading to the column with 6mm thick
high tensile strength steel plates bolted on both the tension and compression sides (partial
interaction plating). Bottom gaps between the steel plate and the beam stub were also set
in this test on both plates as per Fig.9.l7. The design gap width was l.2mm on both sides.
However, the maximum gap width measured immediately before the test when the test
specimen was placed in its test position was about 2mm for both plates, and was
apparently not equal for the two sides. The reason was discussed in Section9.2.3.lt also
seemed that part of the base plate touched the concrete face of the stub.
This test was conducted from 27'h to29th of June 2001 at a concrete age of 353
days. The load versus deflection response for the column is given in Fig.ll.41. The
The measured strains in the reinforcing bars and steel plates are shown in
Figs.l1.42-11.47. Figure 11.42 shows that the upper reinforcing bars reached a tensile
strain of 0.003 at the negative peak point (-51.4kN, -32.3mm) of Fig.ll.4l. At the
positive peak point (32.8kN, 26.1mm) of Fig.ll.4l, the tensile strain of the lower
reinforcing bars was 0.0009 as shown in Fig.11.43. The range of the strain variations in
Fig.ll.42 and Fig.l 1.43 suggests that both the upper and lower reinforcement
experienced alternative tension and compression yielding. The stirrup strain was very
small and did not yield throughout the test (e,r:0.00345) which varied in a range of
- 0.0006 to +0.0006. The stirrup strain was much smaller and very different from the
other five specimens that experienced only tensile strain, as seen from Fig.ll.44. The
compression strain in the stirrups started to occur in cycles with amplitudes greater than
-307 -
CHAPTERIICOLUMNTESTS
52mm and when the displacement decreased from these peak values. The reason for a
smaller stirrup strain may be due to the much higher yield strain of the plate that
prevented the concrete from excessive crushing and dilation. However, the reason for the
compression strain to occur was not clear. A possible explanation is that the stirrup
experienced flexural bending under expansion of concrete and the strain gauge was on
the compression face of the stirrup bar. Under an amplitude (peak) displacement, the
concrete dilation was maximum and the stirrups experienced a large axial elongation
which gave a tensile strain. 'When the column moved back from its amplitude (peak)
position, the concrete dilation was relieved but the bending in the stirrups remained
which gave a compressive strain in the strain gauge. The variation range of plate strains
in Figs I1.45 to 11.47 apparently suggests that the steel plates did not yield on both faces
at any displacement (the monotonic yield strain was about 0.004). The strain gauge for
the internal side of the lower plate was not working in the test.
A problem occurred in this test when the steel plate on the tension side lifted up,
as shown in Fig.11.48. This lifting up of the steel plate invalidated the displacement
measurement at point C, as it could no longer reflect the true rotation of the beam stub.
This problem was not discovered until the very late stages of the test. It had never been a
problem in other tests where two bolts on the side of the stub were used to anchor the
plate, as shown in Figs.9.l5 and 9.16, and the LVDT point C (or D) never moved closer
to the column face than the lower bolt. However, only one bolt on the far side of the stub
was installed in this test, as shown in Fig.9.l7 and Fig.l1.48. The inside bolt was
removed because of concerns that the tension resistance from the tension steel plate may
be too high for this type of high strength steel plate if the inside bolt was used. As
-308-
CHAPTERIICOLUMNTESTS
mentioned in Chapter 5, this increased tension resistance was not desirable from a
To calculate the lateral force for the response curve of Fig.l I .4l,it is necessary to
know the rotation of the stub (as discussed in Section 9.6.1). To solve this problem, the
formula discussed in Section 9.6.2, where measurement points B, C and D were used to
calculate the measurement at point E, as shown in Fig.9.25, were used to calculate the
stub rotation. In this case, measurement points E, B and D were used to calculate the
measurement at point C, when it was pulled up. When it was pushed down, points E, B
and C were used to calculate the displacement of point D. When both displacements at
points C and D are known, the stub rotation can then be calculated. The whole response
curve of Fig.I I .41 was produced using this calculated stub rotation.
On the third day (for the last cycle) of the test, measurement points C and D were
moved to the opposite face of the stub to avoid the problem of plate lifting. The specimen
was pushed and then pulled to the maximum allowable displacement of the test setup in
this last cycle. The result is shown in Fig.l1.49 with the solid lines. The previous cycles
obtained from the calculated stub rotation are shown in the figure by dotted lines. It can
be seen that the transmission from the previous cycles to the last cycle, which was
confirmed, indirectly, that the calculation of the stub rotation mentioned in the previous
Appendix B. The "+" and "-" signs of displacement on the small black board were
-309-
CHAPTERIICOLUMNTESTS
oppos¡te to the sign convention of Fig.1l.4l. The correct sign is given in the respective
During the test, the gaps for the upper and lower plates seemed to close at alaterul
displacement of about +l0mm. Signs of concrete crushing were first noticed at about
*39mm of lateral displacement. The crushing of the concrete continued to develop with
further increase in displacement. However, the steel plates were able to provide adequate
compressive resistance to replace the strength lost from the crushed concrete. In the
maximum pull position (-l00mrn displacement) of the last cycle, the positions of the
horizontal actuator and the verticaljack were held for about l0 minutes. No significant
drop of loading was observed in this period, which was taken to reflect the stability of the
specimen at the final position. The integrity of the specimen was also evident from the
response curve (Fig.ll.4l) where no excessive degradation was indicated up to the end of
the test.
Figuresll.50 and ll.5l show the damage in the specimen at the conclusion of
testing when loosened concrete was removed. It is clearly seen that significant concrete
crushing occurred in the plastic hinge zone. However, the degree of concrete crushing
was lower compared to the other two cyclic tests. The steel plates were still flat and
Figs.l 1.52 and 11.53 after removal of the steel plate. It can be seen that no excessive slip
had occurred in the bolts, as no sign of concrete crushing around the bolts is visible.
- 310 -
CHAPTER II COLUMN TESTS
-80 €0 40 -20 0 20 40 60 80
40
30
zJ 20
10
o
o 0
o -10
õ -20
o -30
(E
J -40
-50
€0
Lateral displacement (mm)
0 006
.=
(! -.-'
_-..- 0 004
+¡
.n
t!
: 0.002
¡t
o,
0
o
CL
CL -0.002
f
-0 004
Fig. I 1.42 Measured strain in upper reinforcing bar for specimen 2BCP6G
'õ 0 008
o 0.006
a!
¡r 0.004
o
o o 002
ìo ¿#
e-æ1
0
J
-0 002
Fig. I 1.43 Measured strain in lower reinforcing bar for specimen 2BCP6G
- 3l I -
CHAPTERIICOLUMNTESTS
-80 €0 40 -20 0 20 40 60 80
0 0006
0 0004
(ú 0 0002
tt 0
an
CL -0 0002
J
-0.0004
at,
-0.0006
0 0008
r
U'
0
o -0 001
x
o
-0 002
o
CL
CL
-0.003
f
Lateral d isplacement (mm)
Fig. I 1.45 Measured strain of steel plate on upper external side for specimen 2BCP6G
po 0.0015
tt,
0.001
õ
0 0005
o
'; 0
o -0 0005
CL
CL
f -0 001
Fig. I 1.46 Measured strain of steel plate on upper internal side for specimen 2BCP6G
-312-
CHAPTERIICOLUMNTESTS
6) -0.001
xo
-0.002
o
ì
o -0.003
J
Lateral displacement (mm)
Fig. 1 1.47 Measured strain of steel plate on lower external side for specimen 2BCP6G
- 313 -
CHAPTERIICOLUMNTESTS
Fig. I I .49 Last cycle of test for specimen 2BCP6G after changing instrumentation
-314-
CHAPTER II COLUMNTESTS
- 315 -
CHAPTERIICOLUMNTESTS
In summary, three monotonically and three cyclically loaded columns were tested
to investigate the plating effect. The test setup and procedures were considered to be
simple but successful and the target of the tests was achieved. Another feature of the
experimental work was that the cost was relatively low. The total cost for the materials
-316-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
The test results are analyzed and discussed in this chapter. ln Section 12.1, compansons
are made between columns with different plating details to see the effects of plating. The
test results are compared to the numerical results in Section 12.2 to verify and evaluate
The test results discussed in this section are the original test results without modification
to account for the self-weight of the test specimen and strengthening frames. It is
intended to keep the test results as original as possible in this part of discussion.
However, the self-weight, that affects the peak force by about 5-7yo, will be included in
Section 12.2 where the test results are compared to the numerical results.
To compare the test results, the three monotonic response curves are plotted together in
Fig.l2.l. The peak points and post-peak softening slopes of the response curves are
summarized in Table 12.1. From numerical simulations in Chapter 5, it was found that
the first yielding of the tension reinforcing bars usually occurs at the peak point of a
response curve (see Figs.5.6 and 5.7). This is the case for columns IAMR and lBMP6,
where it can be seen from column 4 of Table l2.l tha| the strains of the tensile bars at the
peak points were very close to the yield strain of the reinforcing bar, i.e. 0.0025-0.0026 as
-317 -
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
obtained from the reinforcing bar tests in Chapter 10. However, the peak point strain for
2AMF12 is smaller than the yield strain. This measured strain may not be accurate for the
reason given in Section 9.4, because the strain difference between the strain on the top
face and bottom face of a bar due to the curyature is estimated to be as big as 0.0003 at
^60
z 2AMF12
5so
CL
940
o 1BMP6
e30
P
e20
rú R
o10
o
Jo
0 20 40 60 80 100 120
- 318 -
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
lt: L,,lLr, where Ar. is the yield displacement (same as the displacement at the peak
point) of a response curve, and À,, is the ultimate lateral displacement at a point on the
softening branch where the lateral resistance force equals 80% of thícorresponding peak
force, as given in the seventh column of Table 12.1. The tested specimens are depicted in
Fig.12.2, where Ä is the final displacement at which the test stopped. It is clearly seen in
Ftg.l2.2 that the damage to the test specimens reduced with the increase in strength of the
plating systern.
\*"-
-319 -
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
l. Plating increases column strength. This strength enhancement is clearly seen in the
third column of Table l2.l. lt was also evident from the cyclic tests. The strength
enhancement was largely caused by the transferring of compression force from the
RC section to the steel plate, which increased the lever arm of the compression
2. Plating reduces the post-peak softening slope. The post-peak slopes for the plated
columns IBMP6 and 2AMF12 are less steep than that for the un-plated column
IAMR, as shown in the fifth column of Table 12.1. This reduction of slope is also
caused by the transmission of the axial force from the RC column to the steel plate,
3. The displacement ductility factor is increased by plating, as shown in Table 12.1. The
reason for this increase in ductility is that the plating reduces the axial load on the RC
column. The fact that increasing the axial load on the RC column reduces the ductility
of the RC columns is well documented in the literature (Watson. 1994; Watson and
Park 1994). From this point of view, tension plating is not beneficial as it increases
the total compression force on the RC column and hence reduces the ductility.
4. Plating can increase the deformation capacity. The displacement capacity of structural
members is critical for structures to survive a severe seismic attack. The ultimate
displacement of the test columns was clearly increased by platìng, as shown in the
-320-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
5. The amount of strength increase due to plating can be managed and, where necessary,
minimised. The strength increment of IBMP6 is much smaller than that of 2AMF12.
peak strength of the column but only reduce the degradation slope in the softening
For comparison, the cyclic test results are superimposed in Fig.12.3.The purpose of the
cyclic tests was to identify the toughness and suitability of the proposed plating system
under cyclic or dynamic loading. However, the same observations as previously made for
the monotonic tests can also be made from the cyclic test results. In addition, the
Fig.11.20, Fig.l1.27 and Fig.ll.4l that the strength reduction between cycles at the
2. Plating increases the energy absorption capacity of the members. This is evident by
comparing the enclosed area of the response curves.
The conditions of the tested specimens are depicted inFig.l2.4, in which "failure"
means the final loss of axial load carrying capacity where the test could not continue any
further. The damage to column 4ACP6 which was plated by 6mm thick mild steel plates
was less severe as compared to column 3ACR which lost the axial load carrying capacity
under a much smaller displacement excursion than that for column 4ACP6. The damage
to column 2BCP6G which was plated by 6mm thick high yield steel plates was further
-321 -
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
Fig.l2.4.
It is noted in Fig.l2.3 that column 2BCP6G that was plated with high strength
steel plates has a smaller positive peak strength than the normal steel plated column
4ACP6. Furthermore, the negative peak strength is much greater than the positive peak
strength for 2BCP6G itself. The reason for this will be further discussed in Section
12.2.7.
-322-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
The numerical response of the test columns was determined usingthe numerical program
"PLTCOL" described in Chapter 4 and compared to the test results. This comparison can
verify the accuracy of the numerical analysis. It can also give a deeper insight into the
In the numerical simulations, the input data are based on average geometrical
dimensions measured from the test specimens and average material properties obtained
from Chapter 10. The compressive strength and Young's modulus of the concrete are
-)zJ-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
based on cylinder CLYI0 as summartzedin Table 10.1 (Chapter 10). The reason for not
using other cylinders is that some of the cylinder test results were considered as
unreasonable because of their lower strength compared with other cylinders tested earlier.
It was also found after these cylinder tests that there was excessive honeycombing inside
that downgraded the strength. After comparing all the test results in Table 10.1, CLYl0 is
considered to be the most representative of all the concrete at an age of 200 days and
above.
As mentioned before, the test results excluded the self-weight of the test specimen
and the attached strengthening frames. This gravity load was about 5-7%o of the peak
possible, the test results are adjusted to include this gravity load. Details of the
As indicated in Section 9.5.4,the readings of the initial test point were set to zeros for all
instrumentation when no vertical and horizontal loads were applied. However, the self-
weight of the specimen and the attached strengthening frames had already applied an
initial lateral force and, hence, an initial lateral displacement to the test column. This self-
weight for an unplated column with the strengthening frames attached was measured to
be 6kN, which was estimated to give a 2.5kN additional lateral force to the test column.
For plated columns, this additional lateral force was estimated to be approximately
2.7kN.
-324-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
Consequently, an initial lateral displacement due to the gravity load was also
applied to the test column before the start of loading. The displacement of an elastic
F.T:
l\
-
(l 2. l)
3EI
-
However, the initial displacement cannot be estimated using Eq.l2.l, although the system
is linear elastic atthat stage. The reason is that 8q.12.1 is only applicable to a fixed end
column, while the test column is not fixed at its base but is more like spring connected at
the end. The rotational stiffness of the bottom cross-section relative to the imaginary rigid
beam stub, which is considered as the fixed ground of the cantilever column, cannot be
estimated by a given formula. This rotational stiffness at the base cannot be ignored
either, as will be seen in the next section from the test and numerical results.
Nevertheless, this initial displacement can be easily estimated using the initial
stiffness of the response curve. For example, the initial lateral response stiffness is found
to be 4.8kN/mm for column IAMR, which is the slope of the straight line connecting the
first two test points. It is signif,rcantly different from the elastic stiffness of K = Y:r.t
L
kN/mm given by Eq.l2.l. Using the stiffness of 4.8kN/mm, the initial displacement by
the gravity load is simply given by 2.514.8:0.52mm. In this way, all the initial
displacements can be calculated using the initial force and initial stiffness of the
The adjustment to the test results is made, therefore, by shifting the original
The adjusted response curves for the monotonic tests are shown by solid lines in Fig.l2.5
-325-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
in comparison with the original results shown with dotted lines. The cyclic test results are
similarly adjusted. All the comparisons between the test results and numerical results in
60
z.Y 2AMF12
50
CL
o
40
1BMP6
o
o 30
o
o
20
t! lAMR
o 10
o
0
0 20 40 60 80 100 120
The input data for the specimen used in the numerical simulation are summarized below.
The definitions of the parameters are given in previous chapters and are summarized in
the Notation.
o Geometry
o Material
-326-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
The applied axial load is N: 360kN. The confined concrete strength inside the
l00mm, l00mm, l00mm, l00mm, l00mm, l00mm, and 200mm from the top to the
bottom of the column. Mander's model and Menegotto-Pinto's model are used for the
Curve A in Fig.12.6is produced assuming a fixed base and aplastic hinge length
obvious from curves A and B that increasing the plastic hinge length decreases the
descending slope of the column in the post peak softening part. However, it has little
which is caused by the yield penetration of reinforcing bars into the beam stub and the
subsequent pull out of the bars from the face of the stub. The plastic hinge lengths above
the base Ln, and below the base Lrt are calculated using 8q.4.56c and Eq.4.56d,
respectively:
L
r, = 0.08. ¿ :0.08* 121 8:97 .4mm*1 00mm
f
Lru = 0.022' ,r'du:0.022*548.8* l6:193.2x200mm
-327 -
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
2¿o
5
3¡o Test result
G
o20
o
o
õ10
o
-50 0 20 40 60 80 100 120
first stirrup at the base of the column was ignored. However, this first stirrup also takes
part in confining the concrete in practice. From the damage extent of the tested specimen
in Fig.ll.3, it can be seen that the concrete crushing zone is about l50mm long. There
were2 stirrups in this crushed zone including the first stirrup. If these two stirrups were
considered to be fully effective in confining the concrete during the test, then the
effective spacing of the stirrups should be adjusted to 75mm. Using a stirrup spacing of
75mm, the confined concrete strength inside the centerline of stirrups is 48.\MPa.
From Figl2.6, curve D gives the best fit to the test results. This part of the
numerical exercise supports the discussion in Section 4.2.4that the base rotation should
modcling thc basc rotation propcrly, the numerical model cannot represent the initial
-328-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
is given by Eq.12.1, which leads to the elastic column stiffness of K = *:l.l kN/mm.
t
Using the first two points of the numerical and test results, the initial elastic stiffnesses
are calculated from the slope of the line connecting the two points tobe 6.7,6.4,4.6,4.6
and 4.8 (kN/mm) respectively for curves A, B, C, D and the test curve. Obviously, the
initial stiffness of curves A and B, which assumes a fixed base without rotation, is close
to the theoretical value of 7.lkN/mm. The initial stiffness for curves C and D, which
consider a base rotation by allowing the bottom curvature to be extended into the base an
additional length of Lor,, is much closer to the experimental initial stiffness of 4.8kN/mm.
This analysis further justifies the base rotation model in Section 4.2.4.
Þå
For simplicity, only the numerical results¡urve D are compared to the test results
in Fig.l2.7 . lt should also be noted that the shear deformation is not consìdered in the
numerical models. The shear deformation due to shear cracking (not the elastic shear
deformation) was measured for specimen 2AMF12 in Fig.1 I .8 and found to be almost in
direct proportion to lateral displacement. This shear deformation was about 7mm at a
assumed for specimen lAMR, it will bring the numerical result closer to the test result.
-329-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
^40
z
5q30 test result
o
(E
820
o
E10
o
(!
Jo
0 20 40 60 80 100 120
All the input data of this specimen are the same as for specimen IAMR with the
:306.9MP a,
f ,,,,
E
n: 198 67 9MP a, and E rn:600MPa.
The cross-section of the steel plate is discretised into l5 layers in the numerical model.
The confined concrete strength inside the centerline of stirrups is 48.\MPs, the same as
that for curve D of Fig.l2.6.The plastic hinge arrangement is also the same as for curve
The numerical result is given in Fig.l2.8 and compared with the test result. The
initial stiffnesses for the numerical result and for the test result are 9.OkN/mm and
9.4kN/mm, respectively. The numerical column seems stiffer than the test column near
the top of the ascending branch in Fig.l2.8. The difference may be caused by the full-
-330-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
interaction assumption in the numerical model. In the test column, some minor cracks
were noticed on the interface at the bottom of the steel plate that signified some slip in
that small region. When slip occurs in a column, the plating system and hence the plated
60
z.Y 50
o
o 40
test result
o
o 30
o
o 20
E
o 1U
a!
Fig. 12.8 Comparison of test result with theory for specimen 2AMF12
12.2.4Column 1BMP6
-l'"ugf
The input data for this column are the same as {þf given in Section 12.2.3 but with the
The numerical result is shown in Fig.l2.9 (marked with "numerical l") together with the
test result. Obviously alarge difference exists between the numerical and test result even
- 331 -
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
The reason can be found by comparing the three monotonic test results shown in
Fig.l2.l. lt can be seen that this plated test column IBMP6 is even "softer" than the un-
plated benchmark specimen IAMR in the initial stage of loading. ln fact, the initial
that of the un-plated column IAMR of 4.8kN/mm. Apparently, the test result of IBMP6
is "unreasonable" as the plated column should be stiffer than the un-plated one.
nu rical 1
zT numerical2
CL
40
o
(ú 30
o
o test result
o 0
ro +0
o
J
-10 10 30 50 70 90 110
Fig. 12.9 Comparison of test result with theory for specimen IBMP6
Actually, the problem comes from the letter "8" in the name of the specimen
"IBMP6", as "lB" denotes that this was the second test on specimen "l" and it was
tested on the previously strengthened side of column IAMR. In the first test of a virgin
specimen, the un-tested column on the other side experiences some sort of unavoidable
damage under deformation, although the deformation is largely reduced by the very
-JJ¿-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
Therefore, in the second test of the same specimen, the bond condition of the reinforcing
bars inside the beam stub has been deteriorated. In other word, more slippage inside the
stub will be incurred for a same force in a reinforcing bar. This is the reason why the
plated specimen IBMP6 behaved "softer" than the unplated specimen IAMR.
Based on the above analysis, specimen IBMP6 would have a larger base rotation
than the virgin column for the same applied base moment due to the poorer base
this case, the deteriorated base anchorage of the bars can be simulated in the numerical
analysis by using a larger plastic hinge length below the base, which means a deeper
strain penetration into the column base. Another numerical analysis is conducted using
Lor,:450mm instead of the original 200mm (the length of L6 is not the yield penetration
length, it is a scale used to quantifu the base rotation). The result is shown in Fig.12.9 and
is marked with "numerical 2". As expected, it gives a much closer agreement to the test
result.
Specimen 2BCP6G had the same problem that will be discussed in Section 12.2.7,
-JJJ-
CHAPTER I2 COMPARISONS AND DÌSCUSSIONS OF TEST RESULTS
All the input data for this column is the same as for column IAMR. The confined
concrete strength inside the centerline of the stirrups is the same as that for curve D of
Fig.l2.6, i.e. 48.\MPa. The plastic hinge arrangement is also the same, i.e. Lr,:l00mm
and L,,r,:200mm. The numerical result is given in Fig.l2. I I and compared with the test
result. This numerical result is considered as very close to the test result, which further
i,ø9
veri\the accuracy of the computer program "PLTCOL".
Fig. 12.l I Comparison of test result with theory for specimen 3ACR
The input data for this column is the same as that used to get the result of "numerical l"
for column lBMP6. The numerical result is given inFig.l2.12 and is compared with the
test result.
The agreement between the test and the numerical results is not as good as in the
previous 4 columns, especially near the end of the unloading branch where all the
-334-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
unloading branches of the numerical results seem to be pinched toward the origin of the
chart.
zlt 25
o
o
o 0
G
o
t! -25
-50
Fig. 12.12 Comparison of test result with theory for specimen 4ACP6
Column 2BCP6G also presented some complications for the numerical modeling as it
was the second test on specimen 2 and base gaps were provided. For a second test on a
specimen, the bond condition of the longitudinal bars in the beam stub region has
deteriorated from the first test (as discussed in Section 12.2.4) so that the specimen
behaves softer than that of a first test. The test result reproduced in Fig.12.l3 shows that
resistance of 35.5kN and a negative peak resistance of 48.7kN. Compared with the result
of the plated column 4ACP6 in Fig.12.12, the positive peak of this column is
significantly smaller.
-335-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
-80 €0 40 -20 0 20 40 60 80
40
30
ztt 20
o 10
o 0
o
-10
E
o -20
o -30
J
-40
-50
There could be two reasons for this peak reduction. One is due to the gap effect as
the test result, where a 1.2mm base gap for the upper plate (compression plate at positive
response) and no gap for the lower plate are assumed. The input data for this numerical
analysis is the same as that used to get the result of "numerical l" for specimen IBMP6
(no bond deterioration assumed at base), except that the steel plate properties are changed
¡s .f,:734.4MPa, En:196798MPa. The un-symmetrical gap widths on the two sides are
assumed without experimental justification. As discussed in Section I 1.6, the design gap
width was l.2mm on both sides. However, the actual gap width was difficult to measure
-336-
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
t 40
-50
Fig. 12.14 Numerical simulation with un-symmetrical base gaps on two sides
While the respective positive and negative peak strengths of the numerical results
are closer to the test results by including the un-symmetrical base gap, the slope of the
test curves in the elastic response part on the positive side is still much smaller than that
of the numerical ones in Fig.l2.l4. This suggests a second possible reason for the
reduction of the positive peak. As discussed in Section 12.2.4, the deterioration of bond
for the tension bars caused by the previous test of column 2AMF12 on specimen 2 caused
the softening of the support condition. Careful inspection of specimen 2 revealed that the
tension side of column 2AMF12 was also the tension side of the positive response in
Fig.l2.l3. This is the most likely explanation for the un-symmetrical response: the
deterioration of bond in the tension bars from the previous test was more severe than that
in the compression bars. The tension bars of the monotonically loaded column 2AMFI2
slipped significantly under a maximum strain excursion up to 0.9%o (see Fig.l1.6), while
the deformation in the compression bars was much smaller (0.3% of strain in Fig.ll.6).
-337 -
CHAPTER I2 COMPARISONS AND DISCUSSIONS OF TEST RESULTS
Strengthened by the steel plate on the compression face, it can be imagined and can be
seen in Fig.ll.l0 that the slip of the compression bars inside the stub would be much
smaller than that of the tension bars. Because the bond deterioration to the tension bars of
the negative response in Fig.l2.l3 is much smaller, the negative peak is therefore much
bigger than the positive peak. However, this un-symmetrical support condition cannot be
In summary, comparisons between the test results confirmed the theoretical predictions
that the proposed new plating system can be used to increase both the strength and
comparisons between the tests and the numerical simulations showed that the computer
program "PLTCOL" is reasonably accurate. It also showed that the base rotation at the
support due to yield penetration and bond breakdown of reinforcement must be modeled
-338-
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
RECOMMENDATIONS
13.1 SUMMARY
Due to the deficiency of the rectangular shape in providing confinement, existing retrofit
columns that have a potential concrete crushing failure mode in the plastic hinge zone, by
composite partial interaction plating, was proposed and studied in this work. Numerical,
mathematical and experimental research was conducted to study the effectiveness of this
new technique. The results from these studies show that it is possible to use this
technique to achieve the following purposes of:
A practical design methodology that is suitable for engineer to use with a hand
calculator was developed to design the plating system to achieve the design requirement
a The discovery of key composite parameters that govern the degree of interaction
between the two elements connected by shear connectors. This result was obtained
also provides a theoretical background for related design guidelines. This part of the
paper.
a Development of a simple method to calculate the P-A effect. 'With this method, the
column response curve is first calculated without considering the P-À effect and then
converted to a new response curve by a very simply formula. The converted response
curve was proved theoretically to be very close to the actual response curve with the
which can be very different from the classic slip distribution that has been previously
used as design guides. The typical forms were mathematically derived, numerically
condition in the linear elastic analysis that effectively extended the linear theory to
-340-
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
o Introduction and substantiation of a new but very simple method of identifying the
the concrete.
o Proof and explanation of the existence of the "focal point" which was discovered by
Seracino et al. (2001) and regarded as a new concept in composite structure design.
There were several other interesting but minor findings, such as minor modification to the
Priestley and Park (1987) model of the plastic hinge which substantially improved the
accuracy of numerical simulation results; and the identification and correction of a minor
problem in the famous Mander, Priestley and Park concrete stress-strain model.
column with steel or FRP plates attached by gluing or bolting to the faces of the column
(softening) parts of the response curve as well as cyclic loading without diffrculty. Non-
-341 -
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
geometric non-linearity. The P-A effect can be considered either through a rigorous
work which gives no additional computation load other than that for the response without
considering the P-A effect. An iterative procedure is developed to account for partial
interaction, i.e. slip, at the interface between the plates and the RC column. The compuLer
program can also model general variations in lateral confinement to the concrete as well
From the study of Chapter 5, the following conclusions on the ductility of plated columns
can be drawn:
l. The ductility factor of the column is largely affected by the axial load and the length
of the column. Columns with larger axial loads and/or smaller lengths will have
2. Partial interaction plating increases the ductility factor due to the "strength stiffening"
effect caused by transfening the compressive axial load from the RC column to the
3. A good plating system is less stiff, or with minimal stiffness, before yielding of the
column, but the stiffness of the system (plate and bolts) shall be sufficient to get
adequate strength stiffening after yielding of the column in order to minimise the
-342-
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
4. Yielding of the plating system (plate and bolt) stops further strength stiffening and
stiffening part, the yielding of the plating system should be deferued as late as
strength stiffening is achieved when the reinforcement at the compression face yields
Another important conclusion is that the classic slip distribution is applicable only
at the serviceability limit state. The slip distribution may be very different from the
classic one at the ultimate limit state where the maximum slip may occur in the plastic
The analytical work of Chapter 6 extended the classical linear theory of composite
beams. The unique features of this work were that it included an axial load to extend the
original theory from beams to columns and prestressed beams. By introducing the non-
zero boundary conditions associated with the plastic hinge and an elastic zone plus a
plastic hinge model, the theory was further extended from linear elastic (serviceability
limit state) analysis to encompass analyses at the non-linear deformation stage (ultimate
limit state). It was found that the response of composite members is governed by only a
few key composite parameters derived and introduced in this study, regardless of the
large number of variables involved. This finding revealed the fundamentals of composite
interaction and improved composite theory that can be used for the eventual formation of
design guidelines and procedures. This work also revealed a clearer insight into the
-343-
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
distributions of slip along the member and theoretically explained that the classic slip
distribution is only one of the general forms of slip distributions. The conditions at which
the largest slip occurs at the maximum moment position instead of at the zero moment
It was concluded from the study in Chapter 7 thatthe onset of the ultimate concrete strain
on the extreme compressive fibre can be used to identify the ultimate curvature of an RC
section with little confinement, such as in old RC columns with inadequately designed
l) Complete failure of a plain concrete section occurs when the ultimate curvature is
depends on the concrete material property t,,,, axial load level as well as cross-
section depth.
when the axial load level is lower than the critical load. However, it is vice versa for
an RC section with axial load levels higher than the critical load. The critical axial
A displacement based design procedure was developed in Chapter 8 to design the details
of the plating system. This design procedure makes sure that full yielding of the plating
system including the steel plates and the bolts does not occur at the design target
- 344
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
displacement (ultimate displacement). In the mean time, the concrete strain in the
extreme compressive fibre of the plastic hinge just reaches its ultimate strain (spalling
strain) at this target displacement. This design philosophy guarantees the plated RC
column to work in the "strength stiffening" range of its response curve, that was shown to
be a stable deformation stage in Section 5.2.2. The integrity of the concrete is also
guaranteed by equating the ultimate displacement of the RC column to the design target
This design procedure is simple and suitable for engineers to use with the help of
a hand calculator.
Six half-scale columns were designed, manufactured and tested. The results of these tests
confirmed the theoretical conclusions on the effectiveness of the new retrofitting system,
and verified the accuracy of the numerical simulations by the computer program
..PLTCOL".
The overall test design including the test set up, which is different from the test
rigs reported in the literature, was very successful in term of simplicity, low cost, stability
and safety.
From the study of this work, the effectiveness of the new retrofit system is beyond doubt,
at least for those circumstances considered in this research. However, the completed
study was only an initial step towards the new concept of using composite plating to
-345-
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
retrofit RC columns. Many issues and practical problems are yet to be addressed and
solved. For example, more experimental tests are needed on larger scale columns to
As this work concentrates on the concrete failure mode in a plastic hinge zone, other
failure modes, i.e. shear failure and lap slice failure, areyet to be studied. On the other
hand, only compression and tension face plating shown in Fig.13.1(a) was considered in
this study. Theref'ore, this technique in its cunent t'orm would be applicable primarily tbr
columns subject to one-way bending. It is applicable mainly for RC buildings that are
stiffened in one direction by strong shear walls or bridge columns that are restricted to
Further research is needed to investigate a plating scheme that can take load from
any direction or under two way bending. To do so, the other two cases shown in
,1, -l
+ <-
(a) (b) (c)
Actually the retrofit scheme studied in this work as shown in Fig.3.l is only one
way to make use of composite action to retrofit RC columns. Other schemes of composite
plating may also be investigated in the future. For example, the plating system shown in
-346-
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
Fig.l3.2 is also a partial interaction plating system. This system is almost exactly the
same as the traditional steel jacketing, except that the plate bears on the cap beam/footing
(or with a much smaller gap than that provided for jacketing). However, it is theoretically
a composite plating system instead of a jacketing system. The reason is that the
longitudinal force in the plate reduces the compression force in the concrete of the RC
column that inhibits the concrete from excessive crushing and dilating in the transverse
the hoop direction will be mobilised. Therefore, it does not rely on confinement to stop
concrete crushing, instead it relies on the composite action to increase the compressive
With the scheme of Fig.l3.2, The plate is in full interaction (no slip) with the
column in the bonded region as indicated, but it has no interaction with the column in the
and the column. As there is a large bonding area, the bond strength is expected to be
sufficient. If not, measures, such as stud shear connectors, can be used to provide a
sufficient interaction between the plate and the column. In the un-bonded region, the face
of the column could be painted with a coat of de-bonding agent before grouting. The
system shown in Fig.l3.2 works with the same mechanism as that in Fig.3.1. However, it
o It is applicable to columns under two way bending. The lateral load can be applied in
any direction.
-347 -
CHAPTER I3 CONCLUSIONS AND RECOMMENDATIONS
a It is effective for shear defective columns as the enclosed steel plate can be
considered as additional stir ups.
o Because the retrofit system is almost exactly the same as the traditional jacketing
system which has been widely used in engineering, it can be readily accepted by
N
F
RC plate bonded
column region
steel plate
plate un-bonded
re
column base
base gap
(a) Elevation (b) Cross-section
-348-
REFERENCES
REFERENCES
system". Proc. I2th lØorld Conf. on Earthquake Engineering', paper no. 0324, New
Zealand.
3. Aboutaha, R. S., Engelhardt, M.D., Jirsa, J. O., and Kreger, M. E. (1996). "Retrofit of
concrete columns with inadequate lap splices by the use of rectangular steel jackets."
4. Aboutaha, R.S. and Machado, R.l. (1999). "Seismic resistance of steel-tubed high-
125(5), 485-494.
5. Aboutaha, R.S. and Machado, R.I. (1998). "Seismic resistance of steel confined
260.
6. Abrams, D. (1996). "Effects of scale and loading rate with tests of concrete and
7. Ahmad, S.H. and Shah, S.P. (1982). "Stress-strain curves of concrete confined by
-349-
REFERENCES
9. Alcocer, Sergio M.; Jirsa, James O. (1993). "Strength of reinforced concrete frame
10. Almusallam, T.H. and Alsayed, S.H. (1995). "Stress-strain relationship of normal,
39-44.
strength concrete under simulated seismic loading". ACI Structural Journal, 95(5),
s47 -s57 .
13. Attard, M.M. and Setunge, S. (1996). "Stress-strain relationship of confined and
14. Bahn, B.Y. and Hsu, C.T.T. (1998). "Stress-strain of concrete under cyclic loading".
15. Balan, T.A.;Filippou, F.C. and Popov, E.P. (1998). "Hysteretic model of ordinary and
288-297.
16. Banthia, N. (2000). "Sprayed fiber reinforced plastics for repairs and strengthening".
-350-
REFERENCES
Transportation Industry, l4-18 Feb. 2000, University of New South Wales, Sydney,
Australia, 6l-67.
124(9),999-l0l 0.
18. Bayrak, O. and Sheikh, S.A. (1997). "F.arfhquake resistance of lOOMPa concrete
columns". Proc. I't International conference High Strength Concrete, July 13-18,
Institute, Detroit.
20. Biddah, A; Ghobarah, A. and Aziz, T.S. (1997). "Upgrading of nonductile reinforced
- 351 -
REFERENCES
23.Boyd, P.F.; Cofer, W.F. and Mclean, D.l. (1995). "Seismic performance of steel-
encased concrete columns under flexural loading". ACI Structural Journal, 92(3),
355-364.
24.Bracci, Joseph M.; Reinhorn, Andrei M.; and Mander, John B. (1995a). "Seismic
25.Bracci, Joseph M.; Reinhorn, Andrei M.; and Mander, John B. (1995b). "Seismic
26. Bradford, M.A.; Gilberl, R.l.; and Sun, S.C.-H. (1999). "Time-dependent analysis of
reinforced concrete structures using the layered finite element method". Slructural
Engineering and Mechanics, 3(6), Techno- Press Taejon South Korea, 561-578.
28. Bradford, M.A. and Gilbert, R.l. (1992). "Non-linear time-dependent behavior of
29.Bradford, M.A. (1991). "Design of short concrete filled RHS sections". Civll
pp.l89-194.
-352-
REFERENCES
30. Bradford, M.A. and Gilbert, R.l. (1990). "Time dependent analysis and design of
31. Bridge R.Q. (1976). "Concrete filled steel tubular columns". Civil Engineering
32. Bridge, R.Q., O'Shea M.D. and Zhang, J.Q. (1997). "Load moment interaction curves
for concrete filled tubes". Proceedings of the Fifteenth Australian Conference on the
pp.t4l-146.
33. Bridge, R.Q. and Roderick, J.W. (1978). "Behavior of built-up composite columns".
34. Bridge, R. and Webb, J. (1992). "Thin walled circular concrete frlled steel tubular
columns". ASCE, Composite construction in Steel and Concrete II, June 14-19, 1992,
pp.634-649.
35. Burnet, M.J. and Oehlers, D.J. (2001) "Fracture of mechanical shear connectors in
36.Cai, S.H. and Jiao, C. (1984). "Ultimate strength of concrete-filled tube columns".
37 . Calvi, G.M. and Kingsley, G.R. ( 1995) "Displacement-based seismic design of multi-
-353-
REFERENCES
38. Carol, I., Prat, P.C. and Bazant, Z.P. (1992). "New explicit microplane model for
39. CEB 1996. RC ELEMENTS UNDER CYCUC LOADING. CEB State of the Art
40. CEB 1997. FASTENNINGS FOR SEISMIC RETROFITTING. CEB State of the Art
42. Chai, Y.H. (1 996). "Analysis of the seismic characteristics of steel-jacketed circular
43.Chai, Y. H., Priestley, M. J. N., and Seible, F. (1994). "Analytical model for steel-
120(8), 2358-2376.
44.Chai, Yuk Hon; Priestley, M.J. Nigel; Seible, Frieder. (1991). "Seismic retrofit of
circular bridge columns for enhanced flexural performance." ACI Structural Journal,
88(5), 572-584.
45.Chai, Y.H.; Priestley, M.J.N. and Seible, F.(1990). "Retrofit of bridge columns for
-354-
REFERENCES
46. Cirtek, L. (2001). "RC columns strengthened with bandage - experimental program
349.
47. Crawford, John E.; Malvar, L. Javier; Wesevich, James W.; Valancius, Joseph;
48. Desayi, P.; Iyengar, K.T. and Reddy, T.S. (1979). "Stress-strain characteristics of
concrete confined in steel spirals under repeated loading". Mqterials and Structures,
49. Demers, M. and Neale, K.W. (1999). "Confinement of reinforced concrete columns
v.26, pp.226-241.
50. Dodd, L.L. and Restrepo-Posada J.l. (1995). "Model for predicting cyclic behavior of
51. Dritsos, Stephanos F,. (1997). "Jacket retrofitting of reinforced concrete columns".
52. El-Dash, K.M. and Ahmad, S.H. (1995). "A model for the stress-strain relationship of
53. El-Dash, K.M. and Ahmad, S.H. (1994). "4 model for the stress-strain relationship of
rectangular confined normal and high strength concrete columns". Materials and
-35s-
REFERENCES
54. Elhassan, Rami M.; Hart, Gary C. (1995). "Analysis and seismic strengthening of
56. Ersoy, U.; Tankut, A.T. and Suleiman, R. (1993). "Behavior of jacketed columns".
57. Fam, A. and Rizkalla, S. (2000). "Hybrid FRP/concrete structural members". Proc. of
Transportation Industry, 14-18 Feb. 2000, University of New South Wales, Sydney,
59. Foster, S.J. and Attard, M.M. (1997). "Ductility and strength in HSC columns". Proc.
I't International conference High Strength Concrete, July l3-l 8, 1997, Kona, Hawaii,
pp.201-214.
6l.Fujii, M.; Kobayashi, K.; Miyagawa, T.; Inoue, S. and Matsumoto, T.(1988). "'A'
62. Furlong, R.W. (1967). "Strength of steel-encased concrete beam columns". Journal of
-356-
REFERENCES
64. Gardner, N.J. and Jacobson, E.R. (1968). "Structural behavior of concrete filled steel
65. Gardner, J. and Jacobson, R. (1967). "Structural behavior of concrete filled steel
66. Ge, H. and Usami, T. (1992). "Strength of concrete-filled thin-walled steel box
column: experiment". Journal of Structural Engineering, ASCE, 118(l l), 3036-3054.
67. Gergely, Ioan; Pantelides, Chris P.; Nuismer, Ralph J.; and Reaveley, Lawrence D.
68. Ghobarah, A.; Aziz, Tarek S. and Biddah, Ashraf. (1997). "Rehabilitation of
reinforced concrete frame connections using corrugated steel jackefing." ACI
69. Ghobarah, A.; Aziz,TarekS. and Biddah, Ashraf. (1996). "Seismii rehabilitation of
reinforced concrete beam-column connectio ns" . Earthquake Spe ctra, l2(4), 7 61 -7 80.
71. Green, M.F. (2000). "Canadian perspective on rehabilitation of structures with FRP
-357 -
REFERENCES
73.Hajjar, J.F. and Gourley, B.C. (1996). "Representation of concrete filled steel tube
1336.
75.Hanna, S. and Jones, R. (1997). "Composite wraps for aging infrastructures: concrete
77.Hwang, Thomas T.; and Wehnes, William C. (1997). "California refinery meets
seismic codes using new wrapping technique." Oil and Gas Journal,95(13), 70-72.
78. Jasim, N.A. (1999). "Deflections of partially composite beams with linear connector
79. Johnson, R.P. and Molenstra, N. (1991). "Partial shear connection in composite
beams in buildings." Proc. Instit. of Civil Engineers, London, Patt2, Vol. 91,679-
704.
-358-
REFERENCES
81. Karbhari, V.M. (2000). "FRP composites for infrastructure renewal - status and
challenges for the 2l't century". Proc. of the ACUN-2 International Composites
82. Karbhari, V.M.; Eckel II, D.A. (1994). "Effect of cold regions climate on composite
83. Karbhari, V.M. and Gao, Y. (1997). "Composite jacketed concrete under uniaxial
9(4),185-193.
84. Karsan, I.D. and Jirsa, J.O. (1969). "Behavior of concrete under compressive
85. Katsumata, H.; Kobatake, Y.; and Takeda, T. (1988). "Study with carbon fiber for
86. Kendall, E. Atkinson. (1978). An Introduction to Numericol Analysis. John Wiley &
87. Kent, D.C. and Park, R. (1971). "Flexural members with confined concrete". I
88. Kilpatrick, A.E. and Rangan, B.V. (1999). "Tests on high-strength concrete-filled
-359-
REFERENCES
89. Kilpatrick, A.E. and Rangan, B.V. (1997). Deformation-control analysis of composite
90. Kloppel, V.K. and Goder, W. (1957a). "An investigation of the load carrying capacity
Yol.26,No,l, pp.l-10.
91. Kloppel, V.K. and Goder, W. (1957b). "An investigation of the load carryingcapacity
92. Knowels, R. and Park, R. (1969). "Strength of concrete filled steel tubular columns".
93. Knowels, R. and Park, R. (1970). "Axial load design for concrete filled steel tubes".
94. Kwon, Y.B.; Song, J.Y. and Kon, K.S. (2000). "The structural behavior of concrete-
f,rlled steel piers". Proceedings of the l6'h Congress of IABSE, Lucerne, Switzerland,
Sept. l 8-2 1.
95.Lau, K.T. and Zhou, L.M. (2001). "The mechanical behavior of composite-wrapped
96. Liddell, D.P., lngham, J.M. and Davidson B.J. (2001), "lnfluence of loading history
-360-
REFERENCES
98. Liu, A. and Park, R. (2001). "seismic behavior and retrofit of pre-1970's as-built
exterior beam-column joints reinforced by plain round bars". Bulleîin of the New
99. Liu, H.K.; Tai, N.H. and Chen C.C. (2000). "Compression strength of concrete
100. Liu, J. and Foster, S.J. (1998). "Finite-element model for confined concrete
101. Lohr, W.S. (1934). "Concrete encased in steel shells proposed". Engineering
102. Lynn, 4., Moehle, J.P., Mahin, S.A. and Holmes, W.T. (1996). "Seismic
t2(4),71s-739.
103. Machida, A.; Mutsuyoshi, H. and Adhikary, B.B. (2000). "Recent developments
-361 -
REFERENCES
106. Mander, J.8., Priestley, M. J. N., and Park, R. (1988a). "Theoretical stress-strain
1804-1826.
107. Mander, J.8., Priestley, M. J. N., and Park, R. (1988b). "Observed stress-strain
1827-1849.
f 08. Marino, S.; Kedgwick, J.; Yasuzaki, C. and Kanazawa, S. (1992). "Behavior of
1 09. Marsh, Lee. ( 1992) "Seismic retrofits for R/C column bar splices" . Proceedings of
the 1992 Structures Congress, ASCE, San Antonio, Texas, Aprill3-l 5,1992.
I10. Martinez-Rueda, J.E. and Elnashai, A.S. (1997). "Confined concrete model under
lll. Melchers, R.E. (Ed.) (1990). Newcastle Earthquake Study. The Institution of
Engineers, Australia.
ll4. Mirmiran, 4., Shahawy,M., Samaan, M., El Echary, H., Mastrapa, J.C. and Pico,
-362-
REFERENCES
I15. Moehle, J.P. (1996) "Displacement based seismic design criteria". Proceedings of
ll6. Moehle, J.P. and Mahin, S.A. (1991) "Observations on the behavior of reinforced
Structures: Inelastic Response and Design, ed. S.K. Ghosh, American Concrete
117. Monti, G.; Nistico, N. and Santini, S. (1998). "Design equations for FRP
the Netherlands.
ll8. Muguruma, H.; Watanabe, S.; Katsuta, S. and Tanaka, S. (1980). "4 stress-strain
model of confìned concrete". Proc. JCA Cement and Concrete,v.34, Japan Cement
I19. Nakanishi, K.; Kitada, T. and Nakai, H. (1999). "Experimental study on ultimate
strength and ductility of concrete filled steel columns under strong earthquake".
120. Nanni, A. and Bradford, N.M. (1995). "FRP-jacketed concrete under uniaxial
l2l. Neville, A.M. (1981). Properties of Concrete. Third Edition, Longman Scientific
122. Newmark, N.M.; Siess, C.P. and Viest, I.M. (1951). "Test and analysis of
-363 -
REFERENCES
123. Newmark, N.M.; Siess, C.P. and Viest, I.M. (1952). "Studies of slab and beam
highway bridges, Part III - Small scale tests of shear connectors and composite T-
124. Oehlers, D. J. and Bradford, M.A. (1995). Composite Steel and Concrete
U.K.
columns infilled with ultra high strength concrete hardened by centrifugal force."
126. Orito, Y.; Sato, T.; Tanaka, N.; and \üatanabe, Y. (1989). "Study on the unbonded
804.
127. O'Shea M.D. and Bridge, R.Q (2000). "Design of circular thin-walled concrete
128. O'Shea M.D. and Bridge, R.Q (1999). "The effects of local buckling and
confinement in concrete filled circular steel tubes". Proceedings of the 16th
129. Pantelides, C.P.; Marriott, N.; Gergely, J. and Reaveley, L.D. (2000). "Seismic
rehabilitation of damaged bridge piers with FRP composites". Proceedings of the ldh
-364-
REFERENCES
l3l. Park, R.; Kent, D.C. and Sampson, R.A. (1972). "Reinforced concrete members
132. Park, R., Priestley, M.J.N. and Gill, W.D. (1982). "Ductility of square confined
133. Parvin, A. and Wang, W. (2001). "Behavior of FRP jacketed concrete columns
under eccentric loading". Journal of Composiles for ConstrucÍior, ASCE, 5(3), 146-
152.
134. Paulay, T. and Priestley, M.J.N. (1992). Seismic Design of Reinforced Concrete
and Masonry Buildings. John Wiley and Sons, Inc. New York.
135. Pekau, O.A.; Zhang, Z.X.and Liu, G.T. (1 992). "Constitutive model for concrete
136. Pessiki, S.; Harries, K.A.; Kestner, J.T.; Sause, R. and Ricles, J.M. (2001). "Axial
137. Pham, L. and Griffith, M. (1995). Report on the January 17, 1995 Great Hyogo-
Ken Nanbu (Kobe) Earthquake. Joint report from CSIRO Division of Building,
139. Popovics, S. (1973). "4 numerical approach to the complete stress-strain curve of
-365-
REFERENCES
140. Popovics, S. (1970). "A review of stress-strain relationships for concrete". ACI
142. Priestley, M.J.N. (1998) "Displacement based approaches to rational limit states
143. Priestley, M. J. N.and Park, R. (1987). "Strength and ductility of concrete bridge
columns under seismic loading". ACI Structural Journal, 84( l), 6l-76.
144. Priestley, M.J.N.; Seible, F. (1995). "Design of seismic retrofit measures for
concrete and masonry structures." Construction and Building MaÍerials, 9(6), 365-
377.
145. Priestley, M. J. N., Verma, R., and Xiao, Y. (1994a). "Seismic shear strength of
2310-2329.
146. Priestley, M. J. N., Seible, F., Xiao, Y., and Verma, R. (1994b). "Steel jacket
147. Priestley, M. J. N., Seible, F., Xiao, Y., and Verma, R. (1994c). "Steel jacket
test results and comparison with theory." ACI Structural Journal, 91(5), 537 -551.
-366-
REFERENCES
148. Rangan, B.V. (1991). "Design of slender hollow steel columns hlled with
concrete". Proceeding, Inlernational conference on steel and aluminum structures,
149. Rangan, B.V. and Joyce, M. (1992). "Strength of eccentricity loaded slender steel
tubular columns filled with high-strength concrete". ACI Structural Journal, 89(6),
676-681.
150. Ramirez, J.L.; Barcena, J.M.; Urreta, J.l.; Sanchez, J.A. (1997). "Efficiency of
short steel jackets for strengthening square section concrete columns". Construction
l5l. Ramirez, Jose L. (1996). "Ten concrete column repair methods." Construction
columns with welded wire fabric". ACI Structural Journal, 86(6), 615-623.
153. Restrepo-Posada, J.l; Dodd, L.L.; Park, R. and Cooke, N. (1994). "Variables
154. Richart, F.E., Brandtzaeg, A. and Brown, R.L. (1928).*A study of the failure of
155. Richart, F.E., Brandtzaeg, A. and Brown, R.L. (1929). "The failure of plain and
-367 -
REFERENCES
156. Rochette, P. and Labossière, P. (1996). "4 plasticity approach for concrete
columns confined with composite materials". Adv. Composite Mat. in Bridges and
Struct., M. M. El-Badry, ed., Canadian Soc. for Civ. Engrg., Montreal, Canada,359-
366.
4(3), 129-136.
158. Rodriguez, Mario E. and Santiago, Sergio A. (1998). "Simulated seismic load
159. Rodriguez, M.; and Park, R. (1994). "Seismic load tests on reinforced concrete
col umns strengthened by jacketi ng." AC I Stru c tural Journ al, 9l(2), 15 0- 1 5 9.
Housing Research paper, No.l2, Housing and Home Finance Agency, Oct. 1953.
l6l. Saadatmanesh, H., Ehsani, M.R., and Jin, L. (1996). "Seismic strengthening of
circular bridge pier models with fiber composites". ACI Structural Journol, 93(6),
639-647.
162. Saadatmanesh, H., Ehsani, M.R., and Li, M. W. (1994). "Strength and ductility of
concrete columns externally reinforced with fibre composite straps." ACI Structural
163. Saafi, M., Toutanji, H.A. and Li, Z. (1999). "Behavior of concrete columns
confined with fiber reinforced polymer tubes". ACI Material Journal,96(4),500-509.
-368-
REFERENCES
164. Saatcioglu, M. and Razvi, S.R. (1992). "strength and ductility of confined
165. Saatcioglu, M. and Razvi, S.R. (1998). "High-strength concrete columns with
166. Sakai, J. and Kawashima, K. (2000). "An unloading and reloading stress-strain
model for concrete confined by tie reinforcements". ". Proc. I2'h World Conf. on
167. Sakino, K.; and Sun, Y.P. (2000). "Steel jacketing for improvement of column
strength and ductility". Proc. l2'h llorld Conf. on Earthquake Engineering, paper no.
168. Sakino, K.; Sun, Y.P. and Aklan, A. (1996). "Effects of wall thickness of steel
tube on the behavior of square tubed R/C column s". I lth llorld Conf. On Earthquake
Engrg.
square steel tubular short columns subjected to cyclic shearing force and constant
axial force' . Transactions of the Architectural Institute of Japan, No.353, July, pp.81-
89.
170. Samaan, M., Mirmiran, A. and Shahawy, M. (1998). "Model of concrete confined
171. Seible, Frieder; Priestley, M.J. Nigel; Hegemier, Gilbert A.; Innamorato, Donato.
-369-
REFERENCES
172. Seracino, R., Oehlers, D.J. and Yeo, M.F. (2001) "Partial-interaction flexural
173. Shah, S.P. and Ahmad S.H. (1994). High Performance Conuete: Properties and
174. Shah, S.P.; Fafitis, A. and Arnold, R. (1983). "Cyclic loading of spirally
reinforced concrete". Journal of Struclural Engineering, ASCE, v.l 09, pp. 1695- l 7l 0.
175. Shahawy, M.; Mirmiran, A. and Beitelman, T. (2000). "Tests and modeling of
480.
176. Shams, Mohammad; Saadeghvaziri, M. Ala. (1997). "State of the art of concrete-
177. Sheikh, S.A. and Khoury, S.S. (1997). "A performance-based approach for the
design of confining steel in tied column s" . ACI Structural Journal,94(4), 421-431 .
178. Sheikh, S.A. and Khoury, S.S. (1993). "Confined concrete columns with stubs".
179. Sheikh, S.A. and Uzumeri, S.M. (1982). "Analytical model for concrete
180. Sheikh, S.A. and Uzumeri, S.M. (1980). "Strength and ductility of tied concrete
l8l. Sinha, B.P.; Gerstle, K.H. and Tulin, L.G. (1964). "Stress-strain relations for
-370-
REFERENCES
182. Spoelstra, M.R. and Monti, G. (1999). "FRP-confined concrete model". Journal
184. Tamai, S.; Sato, T. and Okamoto, M. (2000). "Hysteresis model of steel jacketed
185. Tan, T.H. and Yip, W.K. (1999). "Behavior of axially loaded concrete columns
confi ned by el liptical hoops"" ACI Structural Journal, 96(6), 967 -97 1 .
186. Taylor, A.W.; Stone, W.C. (1994). "Jacket thickness requirements for seismic
Rehabil Apr 5-6 1993 1994 Sponsored by: Iowa State University. Iowa State Univ
Press p 249-258.
187. Teh, K.F.; Tan, J.; Lin, S.H. and Yaw, S.Y. (1999) "Retrofitting existing
reinforced concrete columns by bolting steel plates to their sides". 4'h year
undergraduate project report, Dept. Civil & Environmental Engng, The University of
Adelaide.
188. Teng, J.G.; Chen, J.F.; Smith, S.T. and Lam, L. (2000). RC Structures
189. Thériault, M. and Neale, K.W. (2000). "Design equations for axially loaded
reinforced concrete columns strengthened with fibre reinforced polymer wraps". Can.
-371 -
REFERENCES
190. Timoshenko, S., StrengÍh of Materials - Part I &II, 3rd edn, D. VAN
NOSTRAND COMPANY, INC., New York, 1955.
192. Tomii, M. (1991). "Ductile and strong columns composed of steel tube, infilled
193. Tomii, M.; Sakino, K. and Xiao, Y. (1987). "Ultimate moment of reinforced
concrete short columns confined in steel tube'. Proceedings of the Pacific Conference
square steel tubular beam-columns". Trans. Arch. Inst. Japan, VoI.280, ppl I I -120.
Stobility of Structures Under Static and þnamic Loads, Washington DC. MaylT-19,
1977, pp.7l8-741. ii
196. Tomii, M., Matsui, C. and Sakino, K. (1973). "Concrete filled steel tube
Tokyo, Japan.
-372-
REFERENCES
externally confined with advanced fiber composite sheets". ACI Material Journal,
96(3),397-404.
jacketed viaduct using microtremor measurement & numerical simulation". Proc. l2th
199. Uy, B. (2001). "strength of short concrete f,rlled high strength steel box columns".
200. Uy, B.(2000). "Strength of concrete filled steel box columns incorporating local
201. Valluvan, Raj; Kreger, Michael E.; Jirsa, James O. (1993). "Strengthening of
column splices for seismic retrofit on nonductile reinforced concrete frames." ACI
202. Wang, Y.C. and Restrepo, J.l. (2001). "lnvestigation of concentrically loaded
203. Warner, R.F.; Rangan, B.V.; Hall, A.S.; and Faulkes, K.A. (1998). Concrete
204. Watson, S. and Park, R. (1994). "Simulated seismic load tests on reinforced
205. Watson, 5., Zahn, F.A. and Park, R. (1994). "Confining reinforcement for
-)t)-
REFERENCES
206. Xiao, Y., Wu, H., and Martin, G. R. (1999). "Prefabricated composite jacketing of
RC columns for enhanced shear strength." Journal of Structural Engineering, ASCE,
125(3),255-264.
207. Xiao, Y., Ma, R. (1997). "Seismic retrofit of RC circular columns using
123(10), 1357-1364.
208. Xiao, Y., Ma, R. (1995). Full scole testing of a typical parking structure column
retrolìtted with carbon fiber composite wrapping. Research Report No. USC-SERR-
I 995.
209. Yankelevsky, D.Z.and Reinhardt, H.W. (1987). "Model for cyclic compressive
210. Yao, J.; Li X. and Li, Z. (2001). "Experimental study of a circular concrete
pp. l88l-1887.
-374-
APPENDIX A YOUNG'S MODULUS TEST CHARTS
14
12
N
E 10
E
z 8
o 6
tt
o y=22671x+0.5755
4
al,
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006
Strain
12
10
N
E 8
E
z 6
U'
ø
o 4
.D y=22300x+0.6688
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
Strain
-375-
APPENDIX A YOUNG'S MODULUS TEST CHARTS
"t2
10
E
E
I
z 6
o
,t
o
L 4 y=2%9"lx+0.6691
tt,
2
ì
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
Strain
14
12
E 10
E
z I
U'
t, 6
atl
L
(r, 4 y=24457x+0.9355
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006
Stain
14
12
E l0
E
2 I
o 6
o
o
L
4 y=28790x+0.3811
v,
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
Sbain
-376-
APPENDIX A YOUNG'S MODULUS TEST CHARTS
14
12
N
Ê 10
E
z 8
ao
at
6
E
(t, 4
y=27468x+0.8611
2
0
0 0.0001 0.0002 0.0003 0 0004 0.0005
Strain
14
12
N
E 10
z I
E
tt 6
o
o
L
4 y=31764x+0.9139
o
2
0
0 0.0001 0.0002 0.0003 0.0004
Stra i n
14
12
N
10
E
z I
o 6
o
o
4
U' y=30168x+0.9785
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
Strain
-377 -
APPENDIX A YOUNG'S MODULUS TEST CHARTS
14
12
E 10
E
z I
at,
th
6
o
L
(r, 4
y=29&l0x+0.8197
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
Strain
25
20
N
E
E 15
z
U'
U' 10
o
L
aD
y=27941x+0.9475
5
0
0 0.0002 0.0004 0.0006 0.0008
Strain
25
20
E
E 5
z
o
U' 0
o
Þ
(t,
5 y=29407x+1.1562
0
0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006 0.0007
Strain
-378-
APPENDIX A YOUNG'S MODULUS TEST CHARTS
25
20
N
E
tr 5
z
t,
o
(¡)
10
È
(r, y=26462x+0.9791
5
0
0 0.0002 0.0004 0.0006 0.0008
Strain
18
16
N 14
E 12
E
z 10
tt, 8
o
E 6
(r, 4 y=32263x+0.7616
2
0
0 0.0001 0.0002 0.0003 0 0004 0.0005
Strain
18
16
.î 14
\12
E
¿10
t¡,
at,
8
E6 y=32343x+0.662
U'4
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
Strain
-379-
APPENDIX A YOUNG'S MODULUS TEST CHARTS
18
16
c14
Ftz
tr
¿10
o8
tt
E6
(t, 4 y=32029x+0.6482
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
Strain
18
16
,c 14
ltz
t
¿10
.o8
g6
att
y=34833x+0.6826
U'4
2
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
Strain
18
16
14
"í^
En
tr
>10
o8
Ø
0,A y=32912x+0.7517
U'4
2
0
0 0.0001 0.0002 0 0003 0 0004 0.0005
Strain
-380-
APPENDIX B TEST PHOTOGRAPHS
s
STUB
RC COLTIMN
T'
\4
,l
-381 -
APPENDIX B TEST PHOTOGRAPHS
!'
-')
I
(
a'}
+
,a+
6 +F"'
-382-
APPENDIX B TEST PHOTOGRAPHS
6
+
-383-
APPENDìX B TEST PHOTOGRAPHS
(,.7
-384-
APPENDIX B TEST PHOTOGRAPHS
r\1,
?.J
!.,¿'.r--t
1L'\
t7
ila-
-385-
APPENDIX B TEST PHOTOGRAPHS
rt)
SO.+ 33ì, ?..r
S.Z
-386-
APPENDIX B TEST PHOTOGRAPHS
-387 -
APPENDIX B TEST PHOTOGRAPHS
r):
-388-
APPENDIX B TEST PHOTOGRAPHS
-389-
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.20 Rear view at+24.5mm displacement of 2''d cycle for specimen 3ACR
-390-
APPENDIX B TEST PHOTOGRAPHS
.Eä
CL "25.Õ u,*
Fig. 8.21 Front view at -25.8mm displacement of 2nd cycle for specimen 3ACR
Fig.B.22 Rear view at -39mm displacement of 1't cycle for specimen 3ACR
-391 -
APPENDIX B TEST PHOTOCRAPHS
Fig.B.23 Rear view at +39mm displacement of 2nd cycle for specimen 3ACR
Fig.8.24 Rear view at +26mm displacement of I't cycle for specimen 4ACp6
-392-
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.25 Rear view at-26mm displacement of I't cycle for specimen 4ACP6
Fig.8.26 Rear view aT +26rnm displacement of 2nd cycle for specimen 4ACP6
-393-
APPENDIX B TEST PHOTOGRAPHS
Fig. B.27 Rear view at -26mm displaceme nt of 2"d cycle for specimen 4ACp6
Fig. 8.28 Rear view at +39mm displacement of l't cycle for spccimcn 4ACp6
-394-
APPENDIX B TEST PHOTOGRAPHS
Fig.B.29 Rear view at -39mm displacement of I't cycle for specimen 4ACP6
Fig. B.30 Rear view at +39mm displacement of 2nd cycle for specimen 4ACP6
-395 -
APPENDIX B TEST PHOTOGRAPHS
Fig. B.3l Rear view at -39mm displacement of 2nd cycle for specimen 4ACp6
Fig. 8.32 Rear view at +52mm displacement of I't cyclc for specimen 4ACp6
-396-
APPENDIX B TEST PHOTOGRAPHS
Fig. B.33 Rear view at -52mm displacement of I't cycle for specimen 4ACP6
Fig. 8.34 Rear view at +52mm displacement of 2nd cycle for specimen 4ACP6
-397 -
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.35 Rear view at -52mm displacement of 2nd cycle for specimen 4ACp6
Fig. 8.36 Rear view at -65mm displacement of l't cycle for specimcn 4ACp6
-398-
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.37 Rear view at +65mm displacement of 2nd cycle for specimen 4ACP6
Fig. 8.38 Rear view at -65mm displacement of 2nd cycle for specimen 4ACP6
-399-
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.39 Rear view at +78mm displacement of I't cycle for specimen 4ACp6
Fig. 8.40 Rear view at -78mm displacement of I't cycle for specimen 4^cp6
-400-
APPEND]X B TEST PHOTOGRAPHS
Fig. B.4l Rear view at +78mm displacement of 2nd cycle for specimen 4ACP6
-401 -
APPENDIX B TEST PHOTOGRAPHS
Fig. B.43 Front view after 2 cycles of +78mm displacement for specimen 4ACp6
Fig. 8.44 Rear view at +91mm displacement of I't cycle for specime n 4ACp6
-402-
APPENDIX B TEST PHOTOGRAPHS
l
l
Fig. 8.45 Rear view at -83mm displacement of 1't cycle for specimen 4ACP6
Fig. 8.46 Rear view at +9lmm displacement of 2nd cycle for specimen 4ACP6
-403-
APPENDIX B TEST PHOTOGRAPHS
Fig. B.47 Rear view at -82mm displacement of 2nd cycle for specimen 4ACp6
() ;¡,t,t l.l Ii i
":
*q rz-r -v?j'.!-
/l-û R l$,7c
L"'Vl),{ r-r"l
t'
I
-,
Fig. 8.48 Front view after 2 cycles of -82mm to +9lmm for specimen 4ACp6
-404-
APPENDIX B TEST PHOTOGRAPHS
t .'lc )-
:-lf ¡f6 -{'Ì6rnr*
Fig. 8.49 Rear view at -26mm displacement of 1't cycle for specimen 2BCP6G
Fig. 8.50 Rear view aT +26mm displacement of 2"d cycle for specimen 2BCP6G
-40s-
APPENDIX B TEST PHOTOCRAPHS
Fig. B.5l Rear view at -26mm displacement of 2''d cycle for specimen 2BCP6G
Fig. 8.52 Rear view at +39mm displacement of I't cycle for specimen 2BCp6G
406 -
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.53 Rear view at -39mm displacement of I't cycle for specimen 2BCP6G
Fig. 8.54 Rear view at +39mm displacement of 2nd cycle for specimen 2BCP6G
-407 -
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.55 Rear view at -39mm displacement of 2nd cycle for specimen 2BCP6G
Fig. B.56 Rear view at+52mm displacement of I't cycle for specimen 2BCP6G
-408-
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.57 Rear view af -52mm displacement of l't cycle for specimen 2BCP6G
Fig. 8.58 Rear view at+S2mmdisplacement of 2nd cycle for specimen 28CP6G
-409-
APPENDIX B TEST PHOTOGRAPHS
-- \ -:--
Fig. 8.59 Rear view at -52mm displacement of 2nd cycle for specimen 2BCP6G
Fig. 8.60 Rear view at +65mm displacement of l't cycle for specimen 2BCP6G
-410-
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.6l Rear view at -65mm displacement of l't cycle for specimen 2BCP6G
Fig.B.62 Rear view at +65mm displacement of 2nd cycle for specimen 28CP6G
- 411 -
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.63 Rear view at -65mm displacement of 2nd cycle for specimen 2BCP6G
Fig. 8.64 Rear view at +78mm displacement of 1't cycle for specimen 2BCP6G
-412-
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.65 Rear view at -78mm displacement of I't cycle for specimen 2BCP6G
Fig. 8.66 Rear view at +78mm displacement of 2nd cycle for specimen 2BCP6G
-413-
APPENDIX B TEST PHOTOGRAPHS
Fig. 8.67 Rear view at -78mm displacement of 2nd cycle for specimen 2BCP6G
-414-
APPENDIX B TEST PHOTOGRAPHS
-415-
APPENDIX B TEST PHOTOGRAPHS
-416-