Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Engineering Structures 153 (2017) 569–581

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Evaluation and comparison of ultimate deformation limits for RC columns MARK


b,⁎
Muhammed Alperen Özdemir , İlker Kazaz a
, Suat Gökhan Özkaya c

a
Dept. of Civil Engineering, Iğdır University, 76000 Iğdır, Turkey
b
Dept. of Civil Engineering, Erzurum Technical University, 25050 Erzurum, Turkey
c
Dept. of Civil Engineering, Ardahan University, 75000 Ardahan, Turkey

A R T I C L E I N F O A B S T R A C T

Keywords: The ultimate deformation capacity assigned to structural elements of reinforced concrete structures is an es-
RC column sential parameter in the determination of their structural performance especially under seismic attack. Various
Performance limit ultimate deformation capacity limits were proposed for RC columns in the previous studies and were accom-
Strain limit modated in the current building codes. However, the reliability of the existing deformation limits is still a matter
Rotation limit
of considerable debate. This study mainly focuses on the evaluation of the accuracy of the existing damage limits
Ultimate deformation
and aims to develop a new definition with a higher reliability in comparison to the existing limits. In this
Performance assessment
purpose, the study was composed of four major steps. Firstly, the existing building codes, standards, regulations
and previous studies were reviewed and evaluated in terms of ultimate deformation criteria for RC columns.
Secondly, actually tested sixty-nine RC columns were selected from PEER Structural Performance Database and
were numerically modelled by using finite element method. The selected RC columns have different dimensions,
aspect ratios, concrete and steel strength, longitudinal and transverse reinforcement amounts and axial load
ratio. Third stage includes the comparison of the results obtained from the experimentally verified numerical
models with the existing ultimate deformation limits to reveal the shortcomings of existing criteria. The nu-
merical modelling facilitated the consistent comparison of strains and curvatures that are rarely available and
difficult to measure in tests. A new concrete compressive strain prediction equation was proposed to determine
the ultimate deformation capacity of rectangular RC columns. It was concluded that there is need for further
comprehensive analytical and experimental studies on deformation limits of reinforced concrete columns.

1. Introduction deformation measures when fiber models are used to estimate inelastic
seismic deformation demands in concrete members in nonlinear ana-
Predicting the deformation capacity of reinforced concrete beams, lysis. However, a limited number of column test results were used for
columns and walls is vital for a thorough seismic evaluation of existing verifying these expressions and limit states specified at the codes,
or new buildings in performance based earthquake engineering. The especially which are based on strain criteria. Acun and Sucuoğlu [9]
ability to remain without significant loss of load-carrying capacity of conducted twelve full-scale column tests to evaluate performance limits
reinforced concrete (RC) structural members under seismic loads is a and they found that proposed deformation limits in Eurocode-8 [1] and
necessary criteria for life safety. Understanding the behavior of col- ASCE/SEI 41 [2] are very conservative. Bae and Bayrak [10] found
umns, which are the primary components of structure, is important for similar results for drift capacity of RC columns. Due to the effort, time
the evaluation of entire structural system. Although behavior of RC spent and limitations of the test set-ups, limited tests were conducted on
columns is well understood and researched, the problem of determi- large-scale RC columns with different design parameters, such as con-
nation of the ultimate deformation capacity of RC columns under finement, concrete strength, longitudinal reinforcement ratio, axial
seismic loads is not fully resolved. load ratio and cross sectional dimensions. In addition, test results of
There are code and performance based expressions that are re- columns with similar properties tested by different researchers indicate
commended for estimating the ultimate deformation capacity of RC different drift capacities. Therefore, proposed expressions and accep-
columns. Codes and previous studies specifies deformation limits for RC tance criteria in seismic action for RC columns need a deeper in-
columns in terms of rotation angle (θ) [1–4], strain (ε) [5–7] and drift vestigation using more extensive and consistent column databases to
ratio at shear failure (δ) [8]. The local strains are convenient improve limit state definitions and their corresponding values.


Corresponding author.
E-mail addresses: m.alperen.ozdemir@igdir.edu.tr (M.A. Özdemir), ilkerkazaz@erzurum.edu.tr (İ. Kazaz), suatgokhanozkaya@ardahan.edu.tr (S.G. Özkaya).

http://dx.doi.org/10.1016/j.engstruct.2017.10.050
Received 15 June 2017; Received in revised form 17 September 2017; Accepted 18 October 2017
0141-0296/ © 2017 Elsevier Ltd. All rights reserved.
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Analytical studies can be conducted to minimize the dispersion and


Ve
discrepancy among the tests related to experimental uncertainties by v
doing large number of analyses from a single hand using reliable bw df ctm
models.
Evaluation of these code limits and proposed equations by analyzing
large number of column database using a well-calibrated finite element
model is easier and more feasible than conducting experiments in to-
day’s technology. In recent decades, the problem of accuracy in the
analysis is further reduced by the developments in the field of model-
ling. Previously, Kazaz et al. [11–13] evaluated deformation limits for
reinforced concrete shear walls by using nonlinear finite element pro-
cedures (FEM). They proposed reliable and improved prediction equa-
tions for the deformation capacity of shear walls. Using a similar ana- Fig. 1. Reduction factor (ϕ)-shear stress ratio (ν) relation.
lytical framework, this study utilizes a verified modelling tool to
investigate the deformation measures defined in terms of plastic rota- to idealized shear cracking model where the shear crack in a member is
tion, curvature and local concrete and steel strains at the cover and core assumed to be inclined at 45°. Here Ve is shear force demand on column,
fiber of RC columns of which test results were collected from the ex- bw is width of cross section, d is effective section height and fctm is
isting literature. tensile strength of concrete. The reduction factor is plotted in Fig. 1.
In this study, sixty-nine experimental column studies that were se-
lected from PEER Structural Performance Database [16] were analyzed 2.2. ASCE/SEI 41-13 plastic rotation limits
with finite element method. To conduct a comprehensive study of
column behavior, finite element models of tested columns were gen- ASCE/SEI 41 [2] is commonly used throughout the US and inter-
erated to estimate the local moment-curvature (strains, rotations as nationally for the seismic assessment and retrofit of existing concrete
well) and global lateral load-displacement behavior. In addition, buildings. Recently, the concrete provisions of ASCE/SEI 41 were up-
crushing and spalling of concrete cover, bond failure, buckling of dated to better reflect the observed performance of concrete compo-
compression bars, strain profile, strain of outer fiber of confined con- nents from quasi-static cyclic tests [14]. These updates were in-
crete were evaluated at ultimate limit state. The adequacy of de- corporated in the latest release of the seismic rehabilitation standard,
formation limits specified by codes and researchers were investigated ASCE/SEI 41.
and the reliability of previously proposed equations was evaluated. The discrete plastic rotation limits defined in terms of axial load
ratio (P/Agfc), volumetric ratio of the confinement (transverse) re-
2. Current regulations and previous studies inforcement (ρs) and normalized shear stress in ASCE/SEI 41 [2] are
expressed with continuous functions using the same variables as given
2.1. Turkish Seismic Code (TSC 2007 and TSC-Draft 2016) limit states in Table 1 for easy calculation of the defined limits. The failure modes
depend on transverse reinforcement condition and shear capacity ratio.
The Turkish Seismic Code (TSC 2007, [5]) specifies limits for To determine probable failure mode of column Table 2 is recommended
structures to be built in disaster areas that are applicable both to newly in ASCE/SEI 41. In Table 2, Vo is the shear strength of column member
constructed buildings and to existing buildings. The Turkish Seismic without modification factor for flexural ductility according to Eq. (10-3)
Code, when compared with equivalent codes, recognizes the use of the in ASCE/SEI 41 and Vp is the shear force at the development of the
concrete and steel strain values as damage limit instead of rotation flexural capacity of a concrete element.
angle or drift ratio.
Concrete and steel strain limits at the fibers of cross section are 2.3. Eurocode 8 Part-3 (CEN, 2005) limits
given in Eqs. (1)–(3).
According to CEN [1] the limit states are based on total chord ro-
For Minimum Damage Limit(MD); (εcu)MD = 0.0035; (εs )MD = 0.010
tation capacity of beams, columns and walls. The chord rotation is the
(1) angle between the tangent to the axis at the yielding end and the chord
connecting that end with the end of shear span at the point of contra
For Safety Limit(SL); (εcg )SL = 0.0035 + 0.01(ρs / ρsm ) ⩽ 0.0135; (εs )SL
flexure. The chord rotation is also equal to the element drift ratio,
= 0.040 (2) which is the deflection at the end of the shear span divided by the
length.
For Collapse Limit(CL); (εcg )CL = 0.004 + 0.014(ρs / ρsm ) ⩽ 0.018; The value of the total chord rotation capacity (elastic plus inelastic
(εs )CL = 0.060 (3) part) at ultimate (Near Collapse, NC), θu, of concrete members under
cyclic loading is calculated from the following expression:
In Eqs. (1)–(3), εcu is concrete strain at the outer fiber of the cross
0.225 f yw
section, εcg is concrete strain at the outer fiber of the confined core, εs is 1 max(0.01;ω′) ⎤ ⎛min ⎛9; L v ⎞ ⎞
0.35
(αρsx )
θu = 0.016(0.3v ) ⎡ fc 25 fc
longitudinal reinforcement strain, ρs is volumetric ratio of transverse γel ⎢
⎣ max(0.01;ω) ⎥ ⎦ ⎝ ⎝ h ⎠⎠
reinforcement that exists in the cross section and arranged as “special
(1.25100ρd ) (4)
seismic hoops and cross-ties”, ρsm is volumetric ratio of the transverse
reinforcement necessary to exist in the section as required by the code. where γel is equal to 1.5 for primary seismic elements and to 1.0 for
The plastic hinge length to relate the local strains to section plastic secondary seismic elements, h is the depth of cross-section, Lv = M/V is
rotation is recommended as 0.5h in TSC. the ratio of moment/shear at the end section, ν = N / bhfc (b width of
TSC-Draft [6] utilizes the same expressions given in Eqs. (1)–(3) in compression zone, N axial force positive for compression), ω' and ω are
the calculation of strain limits. However, for taking into account the reinforcement ratio of the longitudinal compression (including the web
aggravating effect of high shear force on ductility of RC members, strain reinforcement) and tension reinforcement, fc, fyw are the concrete
limit is multiplied by an emprical reduction factor (ϕ), which depends compressive strength (MPa) and the stirrup yield strength (MPa),
on the member shear stress ratio v = Ve /(bw dfctm) . The theoretical shear ρsx = Asx / bw sh is ratio of transverse steel parallel to the direction x of
cracking strength of concrete taken as approximately 0.65fctm according loading (sh = stirrup spacing), ρd is the steel ratio of diagonal

570
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Table 1
Ultimate plastic rotation angle equations and failure modes.

Failure Modes Ultimate Plastic Rotation Angle (rad)

Condition i (Flexure Failure) P ⎞


0.0292−0.047 ⎛ + 1.625(ρs )
⎝ Ag fc′ ⎠
Condition ii (Flexure-Shear Failure where yielding is expected before shear failure) P ⎞ V
0.016 + 2.8636ρ−(4.5455ρs + 0.00117) ⎛ + (0.3636ρ−0.00202) ⎛⎜ ⎞⎟
⎝ Ag fc′ ⎠ ⎝ bw d fc′

Condition iii (Shear Failure) P ⎞
0.00678−0.013 ⎛ + 1.5454ρs
⎝ Ag fc′ ⎠

Table 2 α is confinement effectiveness factor, ρd is ratio of diagonal reinforce-


Transverse Reinforcement Details: Condition to be used for columns in Table 1 (adapted ment.
from ASCE/SEI 41 [2]).
In their study Grammatikou et al. [7] claim that the code limits are
Shear Capacity Ratio Transverse Reinforcement Details not safe and very general. They investigated the analytical relation
between moment and curvature to estimate strains in the bars and the
ACI 318 Closed Other (Including Lap- extreme concrete fibers of section. They derived strain limit formula for
Conforming Hoops with Spliced Transverse extreme fiber of concrete and steel bars as written in Eq. (7). They
Seismic Details 90-Degree Reinforcement)
concluded that strains depend on geometric features of the section. In
with 135-Degree Hooks
Hooks addition, they found that rebar size has an effect on ultimate strain of
concrete and confined concrete; number of bars in tension and com-
Vp/ Vo ⩽ 0.6 ia ii ii pression zone has effect on ultimate steel strain. They proposed the
1.0 ⩾ Vp/ Vo > 0.6 ii ii iii following expression for the ultimate strain of confined concrete
Vp/ Vo > 1.0 iii iii iii
f yw
a
To qualify for condition i, a column should have Av / b w s ⩾ 0.002 and s / d ⩽ 0.5 within εcu = 0.0035 + 0.04 αρs
fc′ (7)
flexural plastic hinge region. Otherwise, the column is assigned to condition ii.

Elwood and Moehle [15] re-evaluated 50 laboratory column tests


reinforcement (if any), in each diagonal direction, α is the confinement with low transverse reinforcement ratio (ρs < 0.007). As a result of
effectiveness factor. their study, a drift capacity model based on experimental results was
In addition, CEN [1] states a deformation limit for concrete ultimate proposed. The following relationship estimates drift ratio at shear
strain of the extreme fiber of the cross section. The expression for ul- failure:
timate strain of the compression zone reads as
3 1 v 1 P 1
αρs f yw δs = + 4ρs − − ⩾
εcm = 0.004 + 0.5 100 40 fc′ 40 Ag fc′ 100 (8)
fcc (5)
where ρs is transverse steel ratio, v is nominal shear stress, fc′ is concrete
where fcc is the confined concrete strength. compressive strength, P is the axial load on the column, and Ag is the
Examples of other deformation limits for existing RC columns are gross cross-sectional area, δs drift at shear failure.
Panagiotakos and Fardis [3], Elwood and Moehle [8], Haselton et al. Haselton et al. [4] used 255 column test results to create empirical
[4] and Grammatikou et al. [7]. Panagiotakos and Fardis [3] stated an equation that predicts deformation capacity. In their study, statistically
equation, which is based on acceptance criteria in CEN [1], to calculate significant design parameters were determined and correlation between
deformation limit as chord rotation of RC columns. Elwood and Moehle plastic rotation and design parameters were identified. They proposed
[8] provided a formula to estimate column drift ratio at shear and axial the following equation.
load failure. Haselton et al. [4] worked on a database of 255 quasi-static
cyclic column tests and they recommended an expression to calculate θcap,pl = 0.12(1 + 0.55asl )(0.16)v (0.02 + 40ρsh )0.43 (0.54)0.01cunits fc′ (0.66)0.1sn (2.27)10.0ρ
plastic rotation capacity of RC columns. Recently Grammatikou et al. (9)
[7] used 168 experimental data to propose an equation to determine
where asl is a bond-slip indicator (asl = 1 where bond-slip is possible), ν
deformation limits based on outer fiber strain of column. These studies
is the axial load ratio, ρsh is the area ratio of transverse reinforcement in
are summarized shortly next for complete evaluation of existing limits.
the plastic hinge region spacing, sn is a rebar buckling coefficient
Panagiotakos and Fardis [3] studied on a comprehensive set of ex-
((s / db)(cunits f y /100)0.5 ), s is stirrup spacing, db is the longitudinal rebar
perimental tests results to reveal deformation behavior of RC members.
diameter, fy is the yield strength of the longitudinal rebar, and cunits is a
They reported that conventional section models predict curvature at
units conversion variable that equals 1.0 when f′c and fy are in MPa
yield satisfactorily, yet ultimate curvature results display significant
units and 6.9 for ksi units.
scatter. For this reason, alternative expression for ultimate chord rota-
tion capacity (θu) given in Eq. (6) was proposed,
3. Methodology and parametric study
0.275
a max(0.01;ω2) ⎤ L 0.45
θum = ast acyc ⎛1 + sl ⎞ a wall (0.2v) ⎡ fc′ ⎛ s⎞
⎝ 2.3 ⎠ ⎢
⎣ max(0.01;ω1) ⎥ ⎦ ⎝h⎠ 3.1. Experimental database
(1.1100αωwx )(1.3 ρd ) (6)
The column dataset used in this study were taken from PEER
where ast is coefficient for the steel of longitudinal bars, acyc is coeffi- Structural Performance Database User’s Manual [16] and related pub-
cient for the type of loading, asl is coefficient for the bond-slip, awall is lications. Table 3 identifies the names as referred in original studies and
coefficient for wall, ν = N/Acf‘c is axial load ratio, ω1, ω2 are re- dimensions of 69 column specimens used in this study to evaluate de-
inforcement ratios, fc is uniaxial concrete strength (MPa), Ls/h = M/Vh formation capacities of rectangular RC columns. Material and geo-
is shear span ratio at the member end, ωwx = (Asx/bwsh)fyw/f′c is me- metric features of each experimentally evaluated rectangular RC
chanical ratio of transverse steel parallel to the direction (x) of loading, column tests and their measured values such as Δy, Vmax, Δu, and failure

571
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Table 3
Dimensions and summary of test results for specimens selected from PEER Structural Performance Database [16]

No Test Series h (mm) b (mm) L (mm) P/Po fyt (Mpa) fc (Mpa) Δy (mm) Δu (mm) Vmax (kN) Failure type

1 Tanaka and Park [39], No. 5 550 550 1650 0.1 325 32 13.55 74.30 386 F
2 Tanaka and Park [39], No. 6 550 550 1650 0.1 325 32 11.96 111.77 409 F
3 Tanaka and Park [39], No. 7 550 550 1650 0.057 325 32.1 9.68 82.81 588 F
4 Park and Paulay [40], No. 9 600 400 1784 0.1 305 26.9 10.71 105.11 393 F
5 Ohno and Nishioka [31], L1 400 400 1600 0.032 325 24.8 9.49 81.39 119 F
6 Ohno and Nishioka [31], L2 400 400 1600 0.032 325 24.8 9.05 85.92 111 F
7 Imai and Yamamoto [38], No. 1 500 400 825 0.07 336 27.1 4.20 18.61 471 SF
8 Saatcioglu and Ozcebe [29], U1 350 350 1000 0 470 43.6 14.00 69.20 295 F
9 Saatcioglu and Ozcebe [29], U3 350 350 1000 0.14 470 34.8 20.78 51.10 267 F
10 Saatcioglu and Ozcebe [29], U4 350 350 1000 0.15 470 32 13.08 89.40 326 F
11 Saatcioglu and Ozcebe [29], U6 350 350 1000 0.13 425 37.3 13.56 87.34 343 F
12 Saatcioglu and Ozcebe [29], U7 350 350 1000 0.13 425 39 13.61 87.60 342 F
13 Wehbe et al. [42], A1 610 380 2335 0.1 428 27.2 23.60 122.09 368 F
14 Wehbe et al. [42], A2 610 380 2335 0.24 428 27.2 22.00 101.42 400 F
15 Wehbe et al. [42], B1 610 380 2335 0.09 428 28.1 27.30 160.52 380 F
16 Wehbe et al. [42], B2 610 380 2335 0.23 428 28.1 26.86 129.50 424 F
17 Lynn et al. [43], 3CLH18 457.2 457.2 1473 0.09 400 27 9.47 23.10 277 SF
18 Lynn et al. [43], 2CLH18 457.2 457.2 1473 0.07 400 33.1 8.40 46.20 241 SF
19 Lynn et al. [43], 2CMH18 457.2 457.2 1473 0.28 400 25.5 4.48 15.50 306 SF
20 Lynn et al. [43], 3CMH18 457.2 457.2 1473 0.26 400 27.6 7.90 15.40 328 SF
21 Lynn et al. [43], 3CMD12 457.2 457.2 1473 0.26 400 27.6 6.39 25.70 355 SF
22 Lynn et al. [43], 3SLH18 457.2 457.2 1473 0.089 400 26.9 7.90 22.80 270 SF
23 Lynn et al. [43], 2SLH18 457.2 457.2 1473 0.073 400 33.1 7.18 31.30 229 SF
24 Lynn et al. [43], 3SMD12 457.2 457.2 1473 0.284 400 25.5 7.84 24.70 367 SF
25 Xiao and Martirossyan [44], HC4-8L19-T10-0.1P 254 254 508 0.1 510 76 6.20 47.80 324 F
26 Xiao and Martirossyan [44], HC4-8L19-T10-0.2P 254 254 508 0.2 510 76 5.30 39.80 378 F
27 Xiao and Martirossyan [44], HC4-8L16-T10-0.1P 254 254 508 0.096 510 86 5.20 36.50 275 F
28 Xiao and Martirossyan [44], HC4-8L16-T10-0.2P 254 254 508 0.192 510 86 6.20 35.00 319 F
29 Nosho et al. [41], No. 1 279.4 279.4 2134 0.339 351 40.6 11.30 34.70 55 F
30 Bayrak and Sheikh [34], ES-1HT 305 305 1842 0.5 463 72.1 9.50 42.30 136 F
31 Bayrak and Sheikh [34], AS-2HT 305 305 1842 0.36 542 71.7 11.30 45.30 149 F
32 Bayrak and Sheikh [34], AS-3HT 305 305 1842 0.5 542 71.8 9.30 33.90 144 F
33 Bayrak and Sheikh [34], AS-4HT 305 305 1842 0.5 463 71.9 14.00 42.00 139 F
34 Bayrak and Sheikh [34], AS-5HT 305 305 1842 0.45 463 101.8 5.67 29.70 182 F
35 Bayrak and Sheikh [34], AS-6HT 305 305 1842 0.46 463 101.9 14.20 28.30 170 F
36 Bayrak and Sheikh [34], AS-7HT 305 305 1842 0.45 542 102 14.60 23.10 159 F
37 Bayrak and Sheikh [34], ES-8HT 305 305 1842 0.47 463 102.2 9.60 29.40 178 F
38 Saatcioglu and Grira [35], BG-1 350 350 1645 0.428 570 34 9.99 33.00 195 F
39 Saatcioglu and Grira [35], BG-2 350 350 1645 0.428 570 34 9.64 50.01 186 F
40 Saatcioglu and Grira [35], BG-3 350 350 1645 0.2 570 34 15.40 116.02 164 F
41 Saatcioglu and Grira [35], BG-4 350 350 1645 0.462 570 34 11.01 50.00 204 F
42 Saatcioglu and Grira [35], BG-5 350 350 1645 0.462 570 34 13.76 99.52 204 F
43 Saatcioglu and Grira [35], BG-6 350 350 1645 0.456 570 34 11.31 81.51 222 F
44 Saatcioglu and Grira [35], BG-7 350 350 1645 0.462 580 34 11.97 99.52 209 F
45 Saatcioglu and Grira [35], BG-8 350 350 1645 0.231 580 34 20.66 118.00 198 F
46 Mo and Wang [45],C1-1 400 400 1400 0.113 460 24.9 14.95 82.28 250 F
47 Mo and Wang [45],C1-2 400 400 1400 0.158 460 26.7 14.79 96.57 268 F
48 Mo and Wang [45],C1-3 400 400 1400 0.216 460 26.1 14.87 88.10 305 F
49 Mo and Wang [45],C2-1 400 400 1400 0.111 460 25.3 16.71 98.02 248 F
50 Mo and Wang [45],C2-2 400 400 1400 0.156 460 27.1 15.62 101.84 261 F
51 Mo and Wang [45],C2-3 400 400 1400 0.21 460 26.8 13.45 77.00 310 F
52 Mo and Wang [45],C3-1 400 400 1400 0.107 460 26.38 17.91 101.59 235 F
53 Mo and Wang [45],C3-2 400 400 1400 0.154 460 27.48 17.89 104.21 260 F
54 Mo and Wang [45],C3-3 400 400 1400 0.209 460 26.9 15.43 120.00 300 F
55 Sezen and Moehle [48], No. 1 457.2 457.2 1473 0.151 476 21.1 12.80 40.41 303 SF
56 Sezen and Moehle [48], No. 2 457.2 457.2 1473 0.605 476 21.1 7.44 27.53 300 SF
57 Sezen and Moehle [48], No. 4 457.2 457.2 1473 0.152 476 21 14.00 38.57 295 SF
58 Bechtoula, Kono, Arai and Watanabe [46], L1D60 600 600 1200 0.567 524 39.2 4.50 35.12 1239 F
59 Bechtoula, Kono, Arai and Watanabe [46], L1N60 600 600 1200 0.567 524 39.2 3.50 31.27 1339 F
60 Bechtoula, Kono, Arai and Watanabe [46], L1N6B 560 560 1200 0.594 524 32.2 2.88 36.25 1200 F
61 Takemura and Kawashima, [47], Test 3 (JSCE-6) 400 400 1245 0.029 368 34.3 5.80 74.18 153 F
62 Takemura and Kawashima, [47], Test 6 (JSCE-9) 400 400 1245 0.027 368 35.9 6.40 105.73 171 F
63 Xiao & Yun [36], No. FHC1-0.2 510 510 1778 0.2 445 64.1 14.80 106.78 724 F
64 Xiao & Yun [36], No. FHC2-0.34 510 510 1778 0.333 445 62.1 11.10 72.32 804 F
65 Xiao & Yun [36], No. FHC3-0.22 510 510 1778 0.225 524 62.1 13.80 106.68 724 F
66 Xiao & Yun [36], No. FHC4-0.33 510 510 1778 0.324 525 62.1 11.40 71.18 797 F
67 Xiao & Yun [36], No. FHC5-0.2 510 510 1778 0.2 445 64.1 14.10 53.32 712 F
68 Xiao & Yun [36], No. FHC6-0.2 510 510 1778 0.2 524 64.1 16.00 72.00 704 F
69 Bayrak & Sheikh [34], No. RS-11HT 350 250 1842 0.51 542 70.8 9.25 86.50 190 F

h: Column Depth, b: Column Width, P/Po = Axial Load Ratio, fyt = Yield stress of transverse reinforcement, fc = Characteristic compressive strength of concrete, Δy = The nominal
displacement at yield, Δu = The maximum recorded column deflections prior to observing a particular level of damage, Vmax = The maximum recorded lateral force. F: Flexural failure,
SF: Shear-Flexure failure.

572
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

type information were given in Table 3. Chord rotation (θu = Δu/L) and uniaxial concrete stress-strain curve to model the concrete multiaxial
chord rotation at yield (θy = Δy/L) values were calculated by using stress behavior. The TSC [5] steel constitutive model was used to model
these measurements. the response of the longitudinal reinforcement. The bar buckling model
Most of the selected test specimens experienced flexure-critical proposed by Dhakal and Maekawa [21] was used to model the buckling
failure whereas only 12 of 69 specimen failed in flexure-shear critical behavior of longitudinal steel under compression. It was realized that
mode [16]. No column in the dataset failed under shear-critical mode. bond-slip might be a significant deformation component on force-de-
The selected columns satisfy the following criteria: column aspect ratio formation relation of columns so that Eligehousen et al. [22] bond-slip
as 1.65 < L/h < 7.63; concrete compression strength as model was added to FE models.
21 < fc < 102 MPa; longitudinal and transverse reinforcement nom- In nonlinear analysis of RC, a shear transfer coefficient must be
inal yield stress, fy and fyw, in the range of 300–580 MPa; longitudinal assumed. For closed cracks (βc), the coefficient is assumed to be 0.8,
reinforcement ratio, 0.01 < ρ < 0.036; transverse reinforcement whereas for open cracks (βt) it should be in the suggested range of
ratio, 0.00082 < ρs < 0.032 and axial load ratio, 0 < P/Po < 0.6. 0.05–0.5, rather than 0.0, to prevent numerical difficulties. In this
So, this database covering wide range of column properties may be study, a value of 0.05 was used, which resulted in acceptably accurate
assumed to represent the flexure-controlled columns’ behavior ade- predictions [18].
quately. SOLID65 utilized in the finite-element analysis is a basic brick ele-
ment with eight nodes and three translational degrees of freedom at
each node. When the softening property after peak response is used in
3.2. Finite Element Analysis (FEA) and validation constitutive models, the finite-element solutions are known to have
spurious sensitivity to the mesh size and to have difficulty in conver-
Finite element modelling strategy adopted in this study does not ging. Element has extra shapes functions capability, yet this function
require any special handling of shear or flexure type behaviors to have leads to strain localization problems in the softening region. Without
good estimations of experimental behavior. The purpose of the finite extra shape functions, to overcome localization and shear locking
element models used in this study is not to revive the complete response problems, section must be discretized with sufficient number of ele-
or all physical phenomenon (cover spalling, crushing, buckling) of an ments in bending direction. The optimum mesh density in each finite
actual test, but is rather to mimic or imitate the observed behavior with element model was determined by performing sensitivity analysis using
an acceptable margin of error. different mesh sizes. Then, lateral displacement-shear force response
Response of tested columns was calculated using nonlinear finite from finite element analyses were compared with experimental results.
element analysis program ANSYS [17]. The general modelling strategy As a rule of thumb, it was determined that for column and beam ele-
and calculation of deformation measures are illustrated in Fig. 2. This ments dividing core section to 5–6 elements along the depth and aiming
study follows the general rules of validated modelling strategy for for an element aspect ratio close to 1 yields reasonable estimation of the
which the detailed description of material models for concrete and steel experimentally measured response. One element thickness is used for
can be found in Kazaz et al. [11–13,18]. These studies include extensive cover concrete.
verification examples of reinforced concrete specimens tested under All type of reinforcement in the model was modelled with LINK180.
dynamic and cycling loading conditions. The effect of cycling loading LINK180 is a 3-D uniaxial tension-compression element with three de-
was accounted by using material models that incorporates such de- grees of freedom at each node: translations in the nodal x, y, and z
gradation effects. The uniaxial confined concrete stress-strain curve is directions.
calculated according to Saatcioglu and Razvi [19] model using the Lateral and axial loads were applied as in the experimental set-ups.
geometric and material properties of the columns. Willam-Warnke [20] Lateral displacement was applied to upper middle nodes of the columns
concrete failure surface model was combined together with multilinear and axial loads were applied as distributed at the tip surface of the
kinematic work hardening plasticity (MKIN) that is based on the

Fig. 2. Finite element model of RC column specimen tested by Ohno and Nishioka [31] displaying the necessary components of deformation calculations.

573
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

column. Large deflection effects were considered wherever required. that the conventional finite element and the section analysis may
Each analysis of test specimens was performed under monotonic static overestimate deformation capacity of RC columns when bond-slip
loads. failure is not accounted in the analysis.
The rotation (θ) at the base of the columns were calculated using the In all the analyses, the ultimate deformation point was determined
axial elongation Δt at the tension side and shortening Δc at the com- on the basis of one of the criteria defined as the point, where strength
pression side of the column section. These displacements (Δt and Δc) drops abruptly or degrades to 85% of the ultimate strength on the load-
were taken over the plastic hinge length assumed to be Lp = h/2 from deformation curve, or the steel strain at the tension side exceeds εs=0.1
base and then divided by the column clear section depth (ho = h-2.cc, or the reinforcing bars at the compression side buckles (accompanied
where cc is the clear cover) to obtain the base rotation angle (θ) of by significant crushing of concrete). Fig. 5(a) and (c) display that there
column. Curvature of section (ϕ), ultimate average strains at core is a good agreement between calculated (FEA) and experimental (EXP)
(εcu,ave) and cover (εcu,ave_cover) fibers of concrete and tension steel ultimate tip displacements (Δu) and base shear force (Vmax) of the
(εsu,ave) over Lp length are calculated as displayed in Fig. 2. columns.
The “plastic hinge length, Lp” is actually a reference length to obtain The yield displacement was calculated according to the method
the base rotation of the column, not really the “plastic zone length, Lpz”, based on an equivalent yield curvature obtained from bilinearization of
which might be different in all models, as it results from the spread of the moment-curvature curve [30]. The curvatures are calculated ac-
plastic steel strains in the model. Eight different widely accepted and cording to Fig. 2 over h/2 height at the base. The bilinear curve is
used plastic hinge formulations [23–28] were evaluated in comparison obtained by drawing the first line from origin to the point on the curve
to the calculated plastic hinge length taken as 0.5Lpz assuming linear where the reinforcement yields for the first time, and extending up to
distribution of the curvatures within the plastic zone. The results are the nominal yield moment (EIy = My/ϕy). The nominal yield moment
plotted in Fig. 3. It is seen that the simple 0.5h formula yields reason- (MNy) is defined as the moment where the extreme fiber strain in
able estimation of the plastic zone length. In order to keep the proce- compression reaches 0.004 or the extreme strain in tension reaches
dure as simple as possible although complicated analyses were carried 0.015, whichever occurs first. The nominal moment closely approx-
out here, a “representative average ultimate strain” was associated with imates the global yield moment. The second line connects the first line
the ultimate damage state although it is not the “actual” value, yet at the top to the ultimate point on the moment-curvature curve. The
correlating with the “actual average strain” over h/2 length at the base nominal yield curvature becomes ϕNy = MNy/EIy. The displacement at
of column. By this way, it is aimed to obtain reasonable estimations of this nominal curvature is adopted as the yield displacement. As shown
the ultimate deformation capacity of the member using direct section in Fig. 5(b) and (d) respectively, there are some differences between
based moment-curvature analysis. experimental and FEA responses for tip displacement at yield (Δy) and
Fig. 4(a) displays the experimental load-deformation curve for a plastic rotation (θp) that is calculated as the ratio of plastic tip dis-
typical tested column by Saatcioglu and Ozcebe [29] and FEA analysis placement to the column length, (Δu − Δy)/L. This observation can be
results. It is also shown in Fig. 4(a) that the bond-slip has a significant attributed to the fact that the procedure depends on the flexural re-
effect on the element flexural rigidity, deformation and load capacity. sponse (moment-curvature curve), so it may give slightly smaller yield
Additionally, potential failure conditions such as bar buckling, crushing displacement than the global yield observed on the experimental load-
of cover concrete were evaluated at ultimate deformation point. deformation curve ignoring the shear deformations. Fig. 5(b) also dis-
Fig. 4(b) displays the comparison of moment-curvature graphs from plays the global yield displacements (blue circles) observed directly on
Finite Element Analysis (FEA) and Section Analysis (SA). SA does not the force-deformation curve. The latter correlates better with experi-
consider bar buckling, bond-slip, high strength concrete and size effects mental values. The other reasons are in general FEA yields slightly
in the calculation of column behavior. The result of this comparison is stiffer response than the experimental ones which may be attributed to
Lp(Model)

db f sy L
Lp Lv h Lp d d Lp d L Lp h
fc d

Lp h
P As
Lp L h
Po Ag
Lp(Model)

Lp h
Lp L db f sy

Lp(FEA) Lp(FEA) Lp(FEA) Lp(FEA)

Fig. 3. Comparison of eight different widely accepted and used plastic hinge formulations with the calculated plastic hinge length that was taken as 0.5Lpz assuming linear distribution of
the curvatures within the plastic zone.

574
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Fig. 4. (a) Comparison of force-displacement curves of experimental, finite element and section analyses results, (b) moment-rotation relationships obtained from SA and FEA.

Fig. 5. Comparison of experimental and FEA results: (a) ultimate tip displacement; (b) tip displacement at yield (the blue empty circles displays the apparent yield displacement on the
load-deformation curve, the filled circles are calculated according to Priestley et al. [30]); (c) maximum base shear force; (d) plastic rotation.

stiffer base fixity, inadequacy of modelling of bond slip, base rotation, typical column tested by Ohno and Nishioka [31]. Strain profiles of
etc. The comparisons given in Fig. 5 validate the finite element models’ columns were obtained at the ultimate deformation. The local strains
ability to represent the experimental study. When the FEA results are over column length were calculated from each element nodal dis-
compared with the experimental data, the agreement seems to be ac- placements, i.e. element strains are displayed in Fig. 6. Curvature and
ceptable for relying on FEA results. rotation diagrams were generated from strain profiles. Also, maximum
Fig. 6 shows calculated strain, curvature and rotation profiles for a local concrete strains (εcu,max) and curvature (ϕmax) at core were taken

h h

h/2 h/2

Fig. 6. Calculated (a) strain, (b) curvature, and (c) rotation profiles of a test specimen by Ohno and Nishioka [31]

575
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Fig. 7. Comparison of curvatures and average strains (εave) calculated over Lp region and from strain profile (εmax) in the core zone.

from strain profiles. As it is seen, the plastic region extends over a basis of finite element analyses results. SA strain values were compared
length approximately equal to column section height h. Strain pene- with average strains over Lp from FEA results. As seen in Fig. 8, there is
tration in the tension side can be also observed in Fig. 6(a). Although a good agreement between the FEA and SA strains. It can be concluded
only displayed for a single column in Fig. 6, severely damaged plastic that the strain calculated over h/2 gage length can be a used to evaluate
region covers a length slightly greater than column section height h in the structural damage, although it is not the actual strain as discussed in
majority of columns. Fig. 7(b).
The measurement of strain, curvature and rotation in experiments is
a crucial issue for reliable comparison of experimental results among
4. Results and discussion
different studies. The location for strain gage installation or the lvdt
gage length for displacement measurements affects the reported results
4.1. Evaluation of code and previously proposed limits
significantly. To shed more light on this issue the strains calculated
pointwise as element maximum and as average over a region equal to
ASCE/SEI 41 specify the plastic rotation limits for RC members. As
h/2 on the column length calculated from FE analyses were compared
shown in Fig. 9(a), comparison of calculated rotations and code plastic
for each deformation measure in Fig. 7. Fig. 7(a) and (c) shows that
rotation limits reveals that ultimate plastic rotation limits of ASCE/SEI
there are small differences between the maximum and average values of
41 corresponding to significant loss of lateral-force capacity provides
curvature and steel strains. However, concrete core maximum (εcu,max)
mostly conservative estimations. Ludovico et al. [33] and Acun and
and average strains (εcu,ave) have significant differences and compres-
Sucuoğlu [9] reached same result in their study on nonconforming RC
sive strains calculated at the cover outer fiber (εcu,ave_cover) are sig-
columns. However, for the columns, which have axial load ratios
nificantly larger than the strains from core outer fiber (εcu,ave) as dis-
greater than 0.45 and made of high strength concrete (HSC, fc
played in Fig. 7(d).
∼70–102 MPa), ASCE/SEI 41 limits remains on the unsafe side, espe-
cially observed in Bayrak and Sheikh [23] tests as displayed in Fig. 9(a).
3.3. Relating FEA and section analyses Haselton et al. [4] proposed Eq. (9) to determine deformation ca-
pacity of columns based on plastic rotation angle. Fig. 9(b) displays that
The load-deformation responses of 69 columns were calculated by the equation provides improved and more adequate estimations of the
using conventional section analysis (SA) in combination with plastic stated limits in ASCE/SEI 41 [2], although the variability of the model
hinge method to establish correlation between FEA and SA results, since error is still big (R2 = 0.56). On the other hand, proposed equation
the use of nonlinear finite element method might not be well suited for yields inadequate estimations for columns made of HSC, have aspect
engineers in practice due to its complication. A modified version of ratio greater than six, transverse reinforcement diameter greater than
CUMBIA [32], which is a reinforced concrete section analysis MATLAB 15 mm and have axial load ratio greater than 0.45 in a similar manner
code, was used for calculating moment-curvature response of columns. to ASCE/SEI 41. These nonconforming column tests were conducted by
However, as mentioned before SA does not consider bar buckling, bond- Saatcioglu and Grira [35], Saatcioglu and Ozcebe [29], Xiao and Yun
slip, high strength concrete and size effects in the calculation of column [36] and Bayrak and Sheikh [37].
behavior. Fig. 9 also displays the comparison of model estimations against the
Fig. 8 compares concrete and steel strains from section and finite experimental ultimate plastic rotations (blue-empty circles). Similar
element analyses at the same ultimate curvature determined on the trends observed for the comparisons of the model estimations against

Fig. 8. Comparison of ultimate strains from FEA and SA at the same


ultimate curvature (a) concrete ultimate strains and (b) steel ultimate
strains.

576
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Fig. 9. Comparison of calculated and proposed plastic rotation capa-


cities: (a) ASCE/SEI 41 limits; (b) Eq. (9) proposed by Haselton et al.
[4]. The blue-empty circles are the model estimations plotted against
the experimental plastic rotations at the horizontal axis.

the FEA based strain limits and experimental limits supports the relia- column database on the unsafe side. Fig. 11(c) shows that when ex-
bility of the acceptance of FEA results as a basis for this study as also treme concrete fiber strain of compression zone given by Eq. (7) were
validated in Figs.4 and 5. evaluated by using calculated strain values, majority of column speci-
CEN [1] and Panagiotakos and Fardis [3] suggest deformation ca- mens (83%) are in the safe zone. However, the problem for estimating
pacity limits for beams, columns and walls based on the total chord deformation capacity of columns under high axial load level and with
rotation. Panagiotakos and Fardis proposed an equation after ex- high aspect ratio still persists (P/Agfc > 0.45 and L/h > 6).
amining 878 RC structural element tests. Moreover, CEN states de-
formation limits similar to Panagiotakos and Fardis. As seen 4.2. Relationship of column design parameters with deformation criteria
Fig. 10(a), 73% of RC column specimens are on the unsafe side con-
cerning the chord rotation angle when CEN limit states were con- Scatter plots in Fig. 12 display the relation between the deformation
sidered. The equation proposed in Panagiotakos and Fardis yields in- measures at ultimate damage state and column design parameters such
significant improvement compared to CEN and just 36% of the columns as the transverse reinforcement ratio, column aspect ratio, axial load
are at safe region as shown in Fig. 10(b). The deflection at the tip of the ratio and concrete compressive strength. These column design para-
column divided by column height defines the drift ratio which can be meters are used to calculate column deformation capacities in codes
interpreted as total chord rotation also. Although the drift ratio ex- and proposed equations. As seen in Fig. 12, when transverse re-
pression given in Eq. (8) yields more conservative rotation estimations inforcement ratio, column compression strength and axial load ratio are
for majority of columns when compared to other definitions, it remains increased, the deformation capacities of RC columns decrease. This The
incapable of giving good correlation against actual values as displayed best correlation, although the relationship is not very strong, between
in Fig. 10(c). the design parameters and the deformation capacities were observed for
Another approach for predicting deformation capacity of RC col- column aspect ratio, which is also related to the shear stress on column,
umns is using the concrete and steel strains as deformation criteria. As and axial load ratio. The confinement reinforcement ratio does not
mentioned before, TSC [5,6] and CEN [1] set strain limits for structural display the expected correlation trend due to high values attained by
RC members. Recently, Grammatikou et al. [7] proposed ultimate this parameter, since above a threshold value the confinement effect
strains and comparing estimations computed from these strains with cannot be distinguished on the column response where other effects
experimental ultimate curvatures, stated that the proposed expression become more dominant.
exhibit much less scatter than those obtained from arbitrary ultimate It is well- known that level of confinement has an important role in
strains specified in the mentioned codes. They also stated that these determining deformation capacity of RC members. It is seen in the first
code predictions are, in general, unsafe. Fig. 11(a) compares the cal- column of Fig. 12 that for transverse reinforcement ratio (ρs) less than
culated core outer fiber strains with TSC-Draft [6] requirements. It is 0.007, ρs has an apparent effect on the concrete strain capacity. How-
concluded that 39% of specimens, which are subjected to high axial ever, in the cases where ρs greater than 0.007, no correlation can be
load and have aspect ratio greater than 6, are at the unsafe zone when observed between the ρs and deformation capacity of columns. For in-
code limits compared with calculated strain values. Moreover, concrete stance, a 500 × 500 mm column with 10 mm diameter bars at 100 mm
strain capacity of columns are significantly underestimated. As seen spacing providing four stirrup legs in the considered direction as con-
Fig. 11(b) CEN predicts the ultimate strain capacity of nearly half of the finement reinforcement has a confinement ratio of 0.007 and is

Fig. 10. Comparison of total chord rotation limits from (a) CEN [1], Eq. (4), (b) Panagiotakos and Fardis [3], Eq. (6) and (c) Elwood and Moehle [8], drift ratio using Eq. (8) with FEA
results.

577
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Fig. 11. Comparison between concrete strains of FEA and proposed limits: (a) Eq. (3), TSC-Draft [6]; (b) Eq. (5), CEN [1]; (c) Eq. (7), Grammatikou et al. [7].

considered as well confined column according to codes [5,6]. Nearly The average axial load ratio is 0.18 with standard deviation of 0.137. As
half of the columns used in this study have transverse reinforcement outliers, if specimens No38, 41, 44, 56 are excluded, the mean and
ratio more than 0.007 and because of that ρs become an insignificant standard deviation for axial load ratio becomes 0.14 and 0.0085 re-
regression variable when high values is attained by this parameter. spectively. So, the properties of columns in this range can be sum-
Aspect ratio (L/h) and axial load ratio (P/Po) have a negative effect marized as made of normal strength concrete, under low to moderate
on the deformation capacity. Column axial load ratio, confinement ratio axial load ratios (P/Po < 0.25) and have medium aspect ratios, L/
and aspect ratio are found to be the parameters that correlates best with h ∼ 3.5.
the deformation capacities for studied column set. It was also seen that the confinement effectiveness factor α is a good
Previous studies and this study show that rotation and curvature are parameter to estimate the concrete compressive strain capacity in
more reliable and stable deformation criteria than column external fiber ρs < 0.007. From the regression analysis, the coefficient of determi-
strains when Figs. 9–11 are compared among each other. Second and nation R2 of the trendline of α-εcu is 0.79 compared to 0.66 of ρs-εcu
third columns of Fig. 12 shows that the strongest correlation between trendline. This parameter besides including the transverse reinforce-
the parameters (axial load and aspect ratio) and the deformation ment spacing s, also reflects the effect of distance between longitudinal
measures is achieved for curvature and rotation as also mentioned in rebars on the confinement effectiveness, especially when the transverse
the discussion of existing limits given above. The trends in Fig. 12 reinforcement ratio is low. In ρs < 0.007 region, α takes values between
displays that steel tensile strains can be estimated satisfactorily. It is 0.19 and 0.83.
also concluded that recently proposed equations increased accuracy of In the ρs > 0.007 range the properties of data in each group is
compressive strain estimations, yet the equations caps the limits for summarized in Table 4. According to this table increasing concrete
large values. Another result of this study is that the calculated force- strength in the high strength concrete range, increasing axial load and
displacement curves matches perfectly with conforming confinement increasing column aspect ratio leads to decrease in the deformation
column tests under low-to-moderate axial load ratio. capacity of columns. Although Group I attains slightly higher values for
axial load ratio and concrete strength values compared to ρs < 0.007
range, it displays similar deformation characteristics, but the strain
4.3. Empirical expression for ultimate concrete strain
capacity is limited to 0.03 approximately. The primary reason of re-
duced strain capacity between Group I and II is seem to be increased
It was displayed that the existing relations for plastic rotation yields
concrete strength. When Group II and III is compared it is seen that both
conservative and satisfactory prediction of damage limits in perfor-
aspect ratio and axial load ratio increases leading to reduced strain
mance assessment of RC columns. On the other side, there is still a room
capacities. Among the column properties, only L/h displays a consistent
for further improvement for strain limits. Regression analysis used to
trend with strain values in all the groups. In ρs > 0.007 region, con-
develop an improved expression for ultimate concrete strains.
finement effectiveness factor α takes values between 0.61 and 0.82,
Confinement effectiveness factor, transverse reinforcement ratio, con-
covering a narrow range.
crete compressive strength, longitudinal reinforcement yield strength
Alternative equation for predicting ultimate compressive strain of
and aspect ratio are important parameters in predicting the deforma-
extreme fiber of reinforced concrete section is given by Eq. (10). The
tion capacity of members.
form of Eq. (7) was adopted for the proposed equation for predicting
Confinement effect on concrete ultimate compressive strain is re-
concrete strain capacity depending on the observed trends for
flected in Fig. 13(a) and an expression for concrete strain was derived
ρs < 0.007. For ρs > 0.007 since the best parameter seen to be the
on the basis of confinement effect. The dataset was organized graphi-
column aspect ratio, an expression is derived based on this parameter.
cally to reflect the effect of different parameters on the ultimate com-
Inclusion of concrete strength and axial load ratio did not make any
pressive strain capacity of reinforced concrete columns. As clearly seen
improvement over the estimation capacity of the basic equation. Pre-
in Fig. 13(a), the ultimate strain capacity increases with increasing
dictions are compared with FEM results in Fig. 13(b). The results show
transverse reinforcement ratio up to 0.007. After this value the ultimate
a clear dispersion around the mean trends. The errors in prediction
strain capacity seemed to be capped and when ρs > 0.007 the ultimate
normalized with respect to actual values (εcu_Eq-εcu_ave)/εcu_ave display
strain capacity displays considerable variation depending on other
nearly equal dispersion on both sides of the mean trend. The predictions
parameters. To perform detailed analysis of other coexisting conditions
deviate from the actual values by 23% and 18% in average on the
and discard possible interactions that leads to such observation the data
overestimation and the underestimation sides, respectively. If the pre-
is grouped according to ultimate strain value in the ρs > 0.007 range.
dicted values reduced by Cc = 0.80 (default value is 1.0 giving mean
In the ρs < 0.007 range (36 specimens) the average concrete
prediction), Eq. (10) can be used as lower bound codified performance
strength is 29.5 MPa with 43.6 MPa maximum. The average column
assessment limits.
aspect ratio (L/h) is 3.58 with standard deviation of 0.9 and only one
data point has an aspect ratio 7.64 in this region. Excluding this value
(Specimen No. 29) from dataset reduces the standard deviation to 0.6.

578
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Fig. 12. Scatter plots showing trends between different deformation measures and column design parameters.

⎧Cc (0.0035 + 0.11 αρs f yw reinforcement, α is the confinement effectiveness factor, L and h is the
) for ρs ⩽ 0.007 ⎫
εcu = fc < 0.035 distance from critical section to point of contra-flexure and section
⎨ ⎬ depth of the column, respectively.
⎩ Cc (0.0675e−0.355L / h) for ρs > 0.007 ⎭ (10)

where fc and fyw are the concrete compressive strength (in MPa) and the
stirrup yield strength (in MPa), ρs is volumetric ratio of transverse

579
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Fig. 13. (a) Effect of transverse reinforcement ratio on concrete strain; (b) comparison of calculated and estimated concrete strains using Eq. (10).

Table 4 concrete and steel strains, respectively). The use of gage length h/2
Mean and standard deviation (μ ± σ) of different column parameters for ρs > 0.007 in analyses and for measurement in tests is recommended for direct
range. interpretation of the experimentally determined section deformation
εcu
capacity with section analysis.

Group No L/h P/Po fc (MPa)
In general, plastic rotation appear to be more stable and predictable
I (13 specimen) 2.8 ± 0.65 0.27 ± 0.18 46.5 ± 19.9 0.0293 ± 0.0052 parameter than the concrete ultimate strain. Calculated results are
II (7 specimen) 3.25 ± 1.0 0.22 ± 0.14 65 ± 15.5 0.0170 ± 0.003 mostly conservative in terms of plastic rotation limits as re-
III (13 specimen) 5.45 ± 0.92 0.43 ± 0.12 66 ± 18.5 0.0084 ± 0.0029
commended in ASCE/SEI 41 [2] and by Eq. (9) [7] except columns
under high axial load and made of high strength concrete. Haselton
et al.’s [7] proposal significantly improve ASCE/SEI 41 limit defi-
5. Conclusion nitions, yet also increasing the variability and possibility of esti-
mating the plastic rotation capacity on the unsafe side. Their
Besides traditionally used plastic rotation limits, local deformation equation provides a larger room for the plastic rotation at the
measures like strain and curvature began to gain acceptance as per- capped region for ASCE/SEI 41. Chord rotation angle limits sug-
formance criteria due to developments in the computational field gested by CEN (Eq. (4)) and Eq. (6) [3] are ineligible to determine
especially owing to nonlinear modelling with fiber-based elements. deformation capacity of columns.
There are various studies investigating the predictability and reliability • TSC-Draft, CEN and Grammatikou et al. [7] set deformation limits in
of these measures as damage limit mainly employing statistical eva- respect of concrete strain for RC columns. Expression of Gramma-
luation of experimental data. However, the strains and curvatures are tikou et al. [7] given in Eq. (7) is the most accurate one, but for
rarely available from experiments impeding thorough evaluation of columns under high axial load, having large aspect ratio and made
strain-based criteria. This study using nonlinear finite element model- of high strength concrete the estimations need to be improved.
ling strategy investigates deformation measures calculated on 69 ex- • Investigating the effect of different design parameters on the column
perimental column specimens. Depending on the results of investiga- deformation capacity revealed that the volumetric ratio of the
tion, the following conclusions were derived, confinement reinforcement when become greater than 0.007 lose to
correlate with the increased ductility capacity. Factors such as the
• The finite element modelling strategy used in this study yields sa- applied loading regime, axial load level and distinct brittle behavior
of high strength concrete may be accountable for this observation,
tisfactory estimations of the peak experimental results. The models
predicted the lateral load capacity with a minor dispersion (R- nevertheless the existence of large amount of confinement re-
squared 0.99). The regression model for ultimate drift gives 0.80 inforcement should not be interpreted as the indication of high
variance while for plastic rotation this value is 0.77. ductility capacity of the member. The column aspect ratio L/h found
• The plastic zone over the column length extends over a region ap- to be the most effective parameter correlating with the compressive
strain capacity in the high confinement range.
proximately slightly higher than the column depth h. The plastic
hinge length taken as h/2 yields acceptable estimations of plastic • It is clearly seen that the damage limit expressions in existing reg-
rotations and tip displacements. ulations and previous studies partially satisfy the need for being a
• Another important conclusion is in regards to the determination of reliable deformation measure as a damage limit. Based on a form
similar to the Eq. (7) proposed by Grammatikou et al. [7] a concrete
strains in experimental and analytical studies. The actual strains
calculated or measured at a single point or on a very limited region core ultimate strain prediction equation is developed to assess the
is obviously greater than the average values of strains calculated structural performance of the RC columns. It is observed that the
over a larger region. The average core outer fiber concrete strains new formula is highly efficient when compared to existing strain
are approximately 1.3 times higher than the pointwise maximum equations and proposed formula can be used as a code performance
values at the same location. The difference between the core and assessment limit using a reduction factor of 0.8.
cover average strains is εcu,ave_cover = 1.17εcu,ave. Variation of strain
in different locations and lengths exhibits the need of using con- As a concluding remark, the authors of this study think that the
sistent length in analyses and experiments in order to directly experimental studies are indispensable to the understanding of the
compare the results. behavior of reinforced concrete members. On the other hand, compu-
• Comparison of average steel and concrete strains from finite element tational methods may produce considerable insight to the behavior of
structural elements where it is difficult to measure local response and
analyses calculated over plastic hinge length (h/2) and section
analyses yields very close correlation between the results of both enables obtaining results that are more consistent among themselves by
analyses (R-squared displaying the variance equals 0.92 and 0.98 for conducting large amount of analyses using several variables.

580
M.A. Özdemir et al. Engineering Structures 153 (2017) 569–581

Acknowledgements [23] Corley WG. Rotational capacity of reinforced concrete beams. J Struct Div, ASCE
1966;92(ST5):121–46.
[24] Mattock AH. Rotational capacity of hinging regions in reinforced concrete beams.
This research is supported financially by Turkish National Science Farmington Hills, MI: American Concrete Institute; 1964;SP-12:143–81.
Foundation, TUBITAK Project No: 113M442. [25] Park R, Priestley MJ, Gill WD. Ductility of square confined concrete columns. J
Struct Div, ASCE 1982;108(ST4):929–50.
[26] Paulay T, Priestley MJN. Seismic design of reinforced concrete and masonry
References buildings. New York: John Wiley and Sons NY, USA; 1992.
[27] Bayrak O, Sheikh SA. Confinement reinforcement design considerations for ductile
HSC columns. J Struct Eng 1998;124(9):999–1010.
[1] CEN (European Committee for Standardization). Design of structures for earthquake
[28] Bae S, Bayrak O. Plastic hinge length of reinforced concrete columns. ACI Struct J
resistance. Part 3: Assessment and retrofitting of buildings. Eurocode 8, EN 1998-3.
2008;105(3):290–300.
Brussels: Comite Europeen de Normalisation; 2005.
[29] Saatcioglu M, Ozcebe G. Response of reinforced concrete columns to simulated
[2] ASCE. Seismic rehabilitation of existing buildings. ASCE/SEI 41–13; 2013,
seismic loading. ACI Struct J 1989;86(1):3–12.
Reston, VA.
[30] Priestley MJN, Calvi GM, Kowalsky MJ. Displacement-based seismic design of
[3] Panagiotakos TB, Fardis MN. Deformations of RC members at yielding and ultimate.
structures. Pavia, Italy: IUSS Press; 2007.
ACI Struct J 2001;98(2):135–48.
[31] Ohno T, Nishioka T. An experimental study on energy absorption capacity of col-
[4] Haselton CB, Liel AB, Taylor LS, Deierlein GG. Beam-column element model cali-
umns in reinforced concrete structures. Proc JSCE, Struct Eng/Earthquake Eng JSCE
brated for predicting flexural response leading to global collapse of RC frame
1984;1(2):23–33.
buildings. PEER Rep. No. 2007/03. Berkeley: Pacific Engineering Research Center,
[32] Montejo LA. Set of codes for the analysis of reinforced concrete members Ph.D.
Univ. of California; 2008.
dissertation Raleigh, NC: Dept. of Civil, Construction and Environmental
[5] TSC. Turkish seismic design code for buildings: Specification for structures to be
Engineering, North Carolina State Univ.; 2007.
built in disaster areas. Ministry of Public Works and Settlement, Ankara, Turkey,
[33] Ludovico DM, Verderame G, Prota A, Manfredi G, Cosenza E. Cyclic behavior of
2007.
nonconforming full-scale RC columns. J Struct Eng 2014;140(5):04013107.
[6] TSC-Draft. Turkish seismic design code for buildings. Ankara, Turkey: Ministry of
[34] Bayrak O, Sheikh, S. Confinement steel requirements for high strength concrete
Public Works and Settlement; 2016.
columns. In: Proc., 11th World Conf. on earthquake engineering, Aculpulco,
[7] Grammatikou S, Biskinis D, Fardis MN. Ultimate strain criteria for RC members in
Mexico, Paper No. 463; 1996.
monotonic or cyclic flexure. J Struct Eng 2016;142(9). 04016046-1-13.
[35] Saatcioglu M, Grira M. Confinement of reinforced concrete columns with welded
[8] Elwood KJ, Moehle JP. Idealized backbone model for existing reinforced concrete
reinforcement grids. ACI Struct J 1999;96(1):29–39.
columns and comparisons with FEMA 356 criteria. Struct Des Tall Spec Build
[36] Xiao Y, Yun HW. Experimental studies on full-scale high-strength concrete columns.
2006;15(5):553–69.
ACI Struct J 2002;99(2):199–207.
[9] Acun B, Sucuoğlu H. Performance of reinforced concrete columns designed for
[37] Bayrak O, Sheikh S. Design of rectangular HSC columns for ductility. Special Publ
flexure under severe displacement cycles. ACI Struct J 2010;107(3):364–71.
ACI Struct J 2003;213:61–82.
[10] Bae S, Bayrak O. Drift capacity of reinforced concrete columns. ACI Struct J
[38] Imai H, Yamamoto Y. A study on causes of earthquake damage of izumi high school
2009;106(4):405–15.
due to miyagi-ken-oki earthquake in 1978. Trans Japan Concr Inst 1986;8:405–18.
[11] Kazaz İ. Analytical study on plastic hinge length of structural walls. J Struct Eng
[39] Tanaka H, Park R. Effect of lateral confining reinforcement on the ductile behavior
2013;139(11):1938–50.
of reinforced concrete columns. Rep. 90-2. Christchurch, New Zealand: Dept. of
[12] Kazaz İ, Gülkan P, Yakut A. Deformation limits for structural walls with confined
Civil Engineering, Univ. of Canterbury; 1990.
boundaries. Earthq Spectra 2012;28(3):1019–46.
[40] Park R, Paulay T. Use of interlocking spirals for transverse reinforcement in bridge
[13] Kazaz İ, Gülkan P, Yakut A. Performance limits for structural walls: an analytical
columns. Strength and ductility of concrete substructures of bridges, RRU (Road
perspective. Eng Str 2012;43:105–19.
Research Unit). Bulletin 1990;84(10):77–92.
[14] Elwood KJ, Matamoros AB, Wallace JW, Lehman DE, Heintz JA, Mitchell AD, et al.
[41] Nosho K, Stanton J, MacRae G. Retrofit of rectangular reinforced concrete columns
Update to ASCE/SE 41 concrete provisions. Earthq Spectra 2007;23(3):493–523.
using tonen forca tow sheet carbon fiber wrapping. Rep. No. SGEM 96-2.
[15] Elwood KJ, Moehle JP. Drift capacity of reinforced concrete columns with light
Washington: Department of Civil Engineering, Univ. of Washington, SE; 1996.
transverse reinforcement. Earthq Spectra 2005;21(1):71–89.
[42] Wehbe N, Saiidi MS, Sanders D. Confinement of rectangular bridge columns for
[16] Berry M, Parrish M, Eberhard M. PEER Structural performance database user’s
moderate seismic areas. National Center for Earthquake Engineering Research
manual. Berkeley, CA: Pacific Engineering Research Center, Univ. of California;
(NCEER) Bulletin 1998;12(1).
2004.
[43] Lynn AC, Moehle JP, Mahin SA, Holmes WT. Seismic evaluation of existing re-
[17] ANSYS 14.5 [Computer software]. Canonsburg, PA, Ansys.
inforced concrete building columns. Earthq Spec 1996;12(4):715–39.
[18] Kazaz I. Dynamic characteristics and performance assessment of reinforced concrete
[44] Xiao Y, Martirossyan A. Seismic performance of high strength concrete columns. J
structural walls Ph.D. Thesis Ankara, Turkey: Civil Engineering Dept., Middle East
Struct Eng 1998;124(3):241–51.
Technical University; 2010.
[45] Mo YL, Wang SJ. Seismic behavior of RC columns with various tie configurations. J
[19] Saatcioglu M, Razvi SR. Strength and ductility of confined concrete. J Struct Eng
Struct Eng 2000;126(10):1122–30.
(ASCE) 1992;118(6):1590–607.
[46] Arai Y, Hakim B, Kono S, Watanabe F. Damage assessment of reinforced concrete
[20] Willam KJ, Warnke ED. Constitutive model for the triaxial behavior of concrete. Int
columns under high axial loading. ACI Special Publ 2006;237:165–76.
Assoc Bridge Struct Eng Proc 1975;19:174–203.
[47] Takemura H, Kawashima K. Effect of loading hysteresis on ductility capacity of
[21] Dhakal RP, Maekawa K. Modeling for postyield buckling of reinforcement. J Struct
bridge piers. J Struct Eng 1997;43A:849–58. [in Japanese].
Eng (ASCE) 2002;128(9):1139–47.
[48] Sezen H, Moehle J. Strength and deformation capacity of RC columns with limited
[22] Eligehausen R, Popov EP, Bertero VV. Local bond stress-slip relationships of de-
ductility. In: Proc., 13th WCEE, Vancouver, Canada; 2004.
formed bars under generalized excitations. Rep. No. UCB/EERC 83–23. Berkeley,
CA: Earthquake Engineering Research Center, Univ. of California; 1983.

581

You might also like