Magazine of Concrete Research Magazine of Concrete Research, 2015, 67(3), 121–132

Volume 67 Issue 3 http://dx.doi.org/10.1680/macr.14.00228

Paper 1400228
New procedure for determining the Received 09/07/2014; revised 21/08/2014; accepted 22/08/2014
moment–curvature relationship of a Published online ahead of print 14/10/2014

reinforced concrete section ICE Publishing: All rights reserved

Chen, Zhou, Wang and Zhang

New procedure for

determining the moment–
curvature relationship of a
reinforced concrete section
Xu Chen Peng Wang
PhD Graduate, Department of Civil Engineering, Kunming University of Lecturer, Department of Civil Engineering, Kunming University of Science
Science and Technology, Kunming, China and Technology, Kunming, China
Dong-hua Zhou Sheng-ping Zhang
Professor, Department of Civil Engineering, Kunming University of Science Lecturer, Department of Civil Engineering, Kunming University of Science
and Technology, Kunming, China and Technology, Kunming, China

The moment–curvature relationship is fundamental for inelastic analysis of structures to predict sectional strength,
flexural stiffness, sectional ductility and energy dissipation capacity. Here, a new procedure for determining the
moment–curvature relationship is developed through calculating the internal force by deformation. This method
improves on the traditional iterative procedure by presenting and using the regulations of strain distributions on
the moment–curvature curve as well as employing more analytical approaches. Based on possible strain distribu-
tions in the ultimate state, three possible strain distribution cases are considered, giving a clear view of the change
process of strain distributions over the whole curve. Algorithms are then proposed to calculate the strain
distributions of the first and last points on the curve, which greatly benefits the rate of convergence. Furthermore,
according to constitutive relationships, equilibrium and deformation compatibility, analytical expressions with
respect to two strains are derived to calculate the internal forces of an arbitrary stress state, which may include the
ultimate; this is normally done by numerical methods. By computer programming, the influences of axial force on
the moment–curvature relationship are analysed, along with the influences of curvature on the moment–axial force
relationship.

Notation
As cross-sectional area of reinforcement mcr, my, mu crack, yield and ultimate moment
a distance between topmost fibre and resultant Ms moment of reinforcement
force of concrete ms dimensionless moment of reinforcement

as dimensionless parameter (¼ 1  d/h) N axial force
b overall width of cross-section n dimensionless axial force (¼ N/bhfcd)
d effective depth of cross-section n1, n2 parameters to determine sectional damage type
Es design value of modulus of elasticity of nc, nc1, nc2 resultant force of concrete
reinforcing steel Ns axial force of reinforcement
e distance between centroidal axis and resultant ns dimensionless axial force of reinforcement
force of concrete ny stress resultant when mean strain is yd
fcd design value of concrete compressive strength r radius of curvature
fyd design yield strength of reinforcement x neutral axis depth
h overall depth of cross-section z local coordinate
i1, iy, iend first, yield and last point on moment–curvature Æc, ka, Æd, kd parameters to determine resultant force of
curve respectively concrete and its position
kx parameter to determine neutral axis depth  strain
M moment c, c1 concrete strain at topmost and lowest fibre
m dimensionless moment (¼ M/bh2fcd) respectively
Mc moment of concrete r concrete strain at distance 3h/7 from the topmost
mc dimensionless moment of concrete fibre

121
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

s, s1 strain of lower and upper reinforcement analytical expressions are derived with respect to two strains to
respectively calculate the internal forces of an arbitrary stress state.
yd strain of reinforcement
Ł angle Assumptions
 stress (a) Plane sections remain plane.
s, s1 stress of lower and upper reinforcement (b) The strain in bonded reinforcement, whether in tension or in
respectively compression, is the same as that in the surrounding concrete.
 dimensionless curvature (¼ h/r) (c) Tensile strain, stress and axial force are positive.
cr, y, u crack, yield and ultimate curvature (d ) The effect of creep is ignored.
ø mechanical reinforcement ratio (¼ Asfyd /bhfcd) (e) The stresses in the reinforcement are given by Equation 1, as
shown in Figure 1(a), where fyd and yd are the design yield
strength and strain of reinforcement respectively and Es is the
Introduction design value of the modulus of elasticity of the reinforcing
The moment–curvature (m–) relationship is used to determine steel. The strain limits for tensile and compressive
the load–deflection behaviour of a member or structure, the reinforcement are set as 20‰.
deflection is a double integration of the curvature on the small-
8
deflection hypothesis and the calculation of second-order effects >
>  f yd  , yd
is dependent on the m– relationship (Kordina and Quast, < f yd
1972). The governing differential equation of a beam–column
 ¼ Es  jj , yd yd ¼
>
> Es
:
problem always involves flexural stiffness (Chen and Atsuta, 1: f yd  . yd
1976), which is calculated as the ratio of moment to curvature.
Furthermore, for seismic design, non-linear analysis like the
pushover method needs an accurate input of m– curves for ( f ) The stresses in the concrete are given by Equation 2, as
reliable results, and new displacement-based methods are based shown in Figure 1(b), where fcd is the design value of
on yield curvature estimates (Choi et al., 2013; Montes and concrete compressive strength. The concrete compression
Aschleim, 2003; Sheikh et al., 2010). strain limit is set as 3 .5‰ and the concrete pure
compression strain limit as 2‰.
Two basic and effective methods are used to obtain the m–
relationship of a reinforced concrete (RC) cross-section. One is a
8
<0 >0
direct method, which is to calculate deformation (curvature and  ¼ f cd ( þ 0.252 ) 0 .  > 2‰
strain) from internal forces (moment and axial force) (Pfeiffer and :
2:  f cd 2‰ .  > 3.3‰
Quast, 2003). The other is an inverse method that calculates the
internal force by deformation (Park and Paulay, 1975), in which a
traditional iterative procedure is generally used by incrementing the
concrete strain at the extreme compression fibre and numerical
Possible strain distributions of moment–
integration is used to determine the internal forces. In the iterative
curvature relationship
procedure, however, a starting strain value is generally assigned
(Sheikh and Yeh, 1993) regardless of the level of axial force, the Possible strain distributions of moment–axial force
procedure is terminated when the sectional strength is reached and interaction curve
the last strain distributions cannot be known beforehand. An As is known, regular strain distributions exist in a strength
idealised m– relationship is normally used because of the incon- calculation of RC cross-sections. For a rectangular cross-section
venience of the real relationship (Jirásek and Bažant, 2001), such as with symmetric reinforcement, sectional parameters and strain
a tri-linear model (Löhning et al., 2009). Recently, analytical distributions in the ultimate state are denoted as in Figure 2(a);
solutions have been used to calculate the ultimate moment and here, all the possible strains are within the dashed border lines,
curvature (Colajanni et al., 2013), as well as the curvature ductility
(Chandrasekaran et al., 2011; Guler et al., 2013). σ σ
fyd
fcd
Based on possible strain distributions in the ultimate state, this 20‰ εyd
work obtains the regular change process of strain distributions on
0 εyd 20‰ ε
the m– curve when a prescribed RC cross-section is subjected
0 2‰ 3·5‰ ε
to a constant axial force, which may be all the possible values fyd
from ultimate tension to ultimate compression. Algorithms are
(a) (b)
proposed to determine the strain distributions of the first and the
last points on the m– curve, which greatly benefits the iterative Figure 1. Stress–strain diagrams: (a) reinforcement; (b) concrete
procedure used to construct the m– relationship. In addition,

122
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

0·5As possible strain distributions are known. Naturally, this raises the
3h/7 question of whether it is possible to find an analogous solution to
d the m– curve.
h εr
1 2
3 Possible strain distributions of the moment–curvature
4
curve
0·5As
εc ‰ Regular strain distributions also exist in determining the m–
b εs relationship, although the m– curves are dependent on the levels
20 εyd 0 2 3·5
of axial force.
(a)

fyd  435 N/mm2, d  0·9h Due to the three sectional damage types mentioned earlier, there are
1·2 also three possible strain distributions on the m– curve, which are
shown in Figures 3(a), 3(b) and 3(c) respectively. For a given cross-
2
section, the three diagrams present the m– curve with a prescribed
3
Moment, m

0·8 axial force on the left-hand side and three strain distributions and
values on the right-hand side, which are of the first point (i1), the
1
yield point (iy) and the last point (iend) of each curve.
0·4
4
2

1


In order to determine which case belongs to a given axial force n,


ω

0 two parameters, n1 and n2, are introduced: n1 is the stress

3 2 1 0 1 2 3
Axial force, n
resultant of s ¼ 20‰ and c ¼ 3 .5‰; n2 is the stress resultant
(b) of the concrete strain at the lowest fibre (c1) ¼ 0 and
c ¼ 3 .5‰, as shown in Figure 3(d). Accordingly, the strains
Figure 2. Rectangular RC section with symmetric reinforcement: change as in case 1 if n > n1, case 2 if n1 . n > n2 and case 3 if
(a) possible strain distributions in the ultimate limit state; (b) m–n n , n2.
interaction curves
For a prescribed RC cross-section subjected to a constant axial
force, methods were derived to directly calculate the strains of
including four deformed regions. Analysing the ultimate strains, point i1 and point iend. Figure 3 shows that the strains from point
it can be concluded that there are three damage types of a cross- i1 to point iend on these curves change regularly.
section according to the strain distributions of fibres along the
sectional depth. (a) For point i1, it is in a pure compression state due to m ¼ 0
and  ¼ 0. The strains of all fibres are equal and the strain
j The first is tensile damage of reinforcement in region 1 at the distribution shape of this point is a vertical line (see Figures
strain of the lower reinforcement (s) ¼ 20‰. 3(a), 3(b) and 3(c)). Analytical expressions were derived to
j The second is compressive damage of concrete in regions 2 calculate this strain, which is denoted as (i1) and the
and 3 at concrete strain at the topmost fibre (c) ¼ 3 .5‰. approach is shown in the next section.
j The third is damage of concrete in region 4 at a concrete (b) For point iend , it is in the ultimate limit state. The curvature
strain at the 3h/7 distance from the topmost fibre of this point at a maximum value means that the strain
(r) ¼ 2‰. distribution shape of this point is a line with the greatest
slope (see Figures 3(a), 3(b) and 3(c)). On the basis of the
Therefore, what is known for ultimate strain is the strain of one three sectional damage types, the known strain of one fibre of
fibre, which may be s ¼ 20‰, c ¼ 3 .5‰ or r ¼ 2‰, and this point is s ¼ 20‰ in case 1, c ¼ 3 .5‰ in case 2 or
the strains of the rest of the fibres are unknown, which may r ¼ 2‰ in case 3. An algorithm is proposed to calculate
change arbitrarily within the dashed border lines. the strain of another fibre of this point, which is also given
later. Furthermore, the m–n value of this point is the same as
To determine values of all the possible values on the moment– one corresponding point on the m–n interaction curve.
axial force (m–n) interaction curve (see Figure 2(b)), the ultimate (c) Point iy is the state of the first yielding of reinforcement. Two
strain of one fibre, besides the already known one, is taken as an possibilities exist – one is tensile yield in the lower
independent variable. Given all the possible values within the reinforcement in cases 1 and 2 and the other is compressive
dashed border lines, the corresponding stresses are then calcu- yield in the upper reinforcement in case 3.
lated by the constitutive laws of Equations 1 and 2. Finally, a (d ) During the process from point i1 to point iend , the strains of
series of values of m–n is obtained from equilibrium conditions. all fibres are unknown. Nevertheless, with an increase in
This procedure can be finished analytically or numerically. In curvature, c is always decreasing from (i1) to c(iend) and s
summary, the m–n interaction curve can be obtained if all the is always increasing from (i1) to s(iend), as shown in the

123
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

Point i1 iy iend Point i1 iy iend

n  0·5 n  0·2
εc 1·09 0·22 2·2 εc
0·3 0·6 0·14 2·17 3·5
εc εc

Moment, m
Moment, m

iy iend
0·2 0·4
iy iend

0·1 0·2
εs εc1: ‰ εs εc1: ‰
i1 i1
0 5 10 15 20 25
εs 20 2·175 1·09 0 3 6 9 12 15 εs 8·73 2·175 εc1  0·14
Curvature, φ: ‰ Point iend iy i1 Curvature, φ: ‰ Point iend iy Point i1

(a) Case 1: n  0·187 (b) Case 2: 0·187  n  1·39

Point i1 iy iend
n  1·8
0·10 εc 1·77 2·27 2·46
0·08 iy εc
iend
Moment, m

0·06 Point iy Point iend

εc1  2·175 εr  2
0·04

0·02 εs εc1: ‰ εs εc εc: ‰

i1 εc1 1·295 1·325 1·77 20 0 3·5 0 3·5

0 0·5 1·0 1·5 n1  0·187 n2  1·39
Point iend iy i1
Curvature, φ: ‰

(c) Case 3: n  1·39 (d) Strain distributions of n1 and n2

Figure 3. m– curves and strain distributions for a symmetric RC

cross-section of ø ¼ 1, fyd ¼ 435 N/mm2 and d ¼ 0 .9h

tables of strain values in Figure 3(a), 3(b) and 3(c). For a given compression, the design yield strain of reinforcement
Obviously, here, the stress resultants of m–n are not ultimate (yd) should be compared with 2‰. Firstly, when yd . 2‰, the
m–n, which are located inside the m–n interaction curve. reinforcements do not always yield, then the equilibrium is
Specially, arbitrary m–n is denoted for them in this paper to
distinguish from the ultimate m–n.
N ¼ As Es (i1 ) þ bhf cd ((i1 ) þ 0:252 (i1 ))
Analytical expressions to calculate the strain of point i1
(å(i1)) where b and h are the overall width and depth of the cross-
The mean strain (i1) is obtained by equilibrium and constitutive section, respectively. Then,
relations. According to the foregoing assumptions, the range of
the mean strain is yd > (i1) > 2‰. For simplicity, the values
ø(i1 )
of all strains and curvatures in this paper are multiplied by 1000 n¼ þ (i1 ) þ 0.25(i1 )2
before placement in the formulas. For a given tension, the stress yd
of concrete is zero, then N ¼ AsEs(i1), where As is the cross-
sectional area of the reinforcement. The dimensionless axial force
is given by Solving this quadratic equation generates

" 2 #1=2
As f yd Es (i1 ) ø(i1 ) ø ø
n¼  ¼ (i1 ) ¼ 2 þ1 þn 2 2
bhf cd f yd yd yd yd

4: n < 0, yd . 2‰
where the mechanical reinforcement ratio ø ¼ Asfyd /bhfcd. Then,

nyd
(i1 ) ¼ n.0 Secondly, when yd < 2‰, parameter ny is introduced to judge
3: ø
whether the reinforcements yield or not, and ny is the stress

124
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

resultant of (i1) ¼ yd, Similarly, one obtains ny ¼ 0:252yd ment (s1), expressed by c and s as given in Equation 6 in which
 ø  yd : dimensionless parameter as ¼ 1  d=h
(
When 0 . n > ny, the reinforcements do not yield, then the c1 ¼ c þ (s  c )=(1  as )
expression of (i1) is the same as Equation 4. When n , ny, the
6: s1 ¼ c þ (s  c )as =(1  as )
reinforcements yield and equilibrium is given by

N ¼ As Es yd þ bhf cd ((i1 ) þ 0:252 (i1 ))

The relationship between strain and curvature can be established
by the geometrical compatibility in Figure 4. Point O is the
Similarly, n ¼ ø þ (i1 ) þ 0:252 (i1 ): Then curvature centre, r is the radius of curvature and the relationship
between curvature 1/r and relative angle Ł of the two end sections
5: (i1 ) ¼ 2(1 þ ø þ n)1=2  2 n , ny , yd < 2‰ is 1/r ¼ dŁ/ds. Therefore

ds (1 þ c1 )ds  (1 þ c )ds (c1  c )ds

¼ ¼
Note that these expressions for (i1) are also suitable for other r h h
types of RC cross-section (e.g. circular and annular) because the
dimensionless forms are used.
With the dimensionless curvature  ¼ h/r and Equation 6, one
Analytical solution obtains
Previous studies on ultimate m–n have been done by calculating
(
the ultimate strains analytically or numerically and, in these  ¼ c1  c
studies, only one independent variable (unknown strain of one
7:  ¼ (s  c )=(1  as )
fibre) is needed. When it comes to the problem of the m–
relationship, what should be determined is arbitrary m–n, as
analysed above. Therefore, an analytical algorithm to obtain
arbitrary m–n was derived by taking two strains as independent Consequently, the strain distributions can be determined by only
variables, which can change arbitrarily within the dashed border two variables, which may be the strains of any two fibres along
lines of Figure 2. This is the main content of this section. the longitudinal direction or the curvature and strain of one fibre.
Note that Equations 6 and 7 are suitable for any section shape
Relationship between strain, neutral axis depth and and any material, including elastic and inelastic, because only the
curvature compatibility condition is used during the derivation.
A segment is taken from a member along the longitudinal
direction, and its flexural deformation and normal strain distribu- Assuming neutral axis depth x ¼ kxh, one obtains an expression
tions are shown in Figure 4. Since plane sections remain plane, for kx by geometrical relationships, as shown in Equation 8. This
the strain distributions are linear – that is, the strain value of a will be used to determine the stress resultant of the concrete and
fibre is proportional to the distance between the fibre and the can be used to judge the location of the neutral axis and the stress
neutral axis. Therefore, conversion relationships between the state (see Figure 5)
strains of all the fibres along the sectional depth can be obtained,
such as the formulas of c1 and the strain of the upper reinforce- c
kx ¼
8: c1  c
O
εc
εs1
θ When kx , 0, the neutral axis is located above the topmost edge
r of a cross-section, which is subjected to tension. When
x  kxh
0 < kx < 1, the neutral axis is located inside the cross-section,
m (1  εc)ds m which is subjected to partial tension and partial compression.
n n Neutral axis When kx . 1, the neutral axis is located below the lower edge of
εs a cross-section, which is subjected to compression.
ds
εc1

(1  εc1)ds Formulas of arbitrary m–n with respect to two strains

when kx < 1
Figure 4. Flexural deformation of a segment and strain According to the superposition principle, the stress resultants of
distribution reinforcement and concrete can be calculated separately. The
calculation of reinforcement is relatively simple because of its

125
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

kxh

εc εc

kxh

kxh

εc1 εc1

0
ε 0 ε ε 0 ε

(a) kx  0 (b) 0  kx  1 (c) kx  1

Figure 5. Neutral axis positions

unchanged working point. By substituting s and s1 into εc

0·5ωσs1/fyd
constitutive Equation 1, one obtains εs1 a  kax
xz
x z
zεc /x nc  ackx
z m
j the stress of the lower and upper reinforcement (s and s1
respectively) n
j the stress resultants of reinforcement
0·5ωσs /fyd
εs
N s ¼ 0:5As ( s þ  s1 ) εc1
(a)
εc
M s ¼ 0:5As  s (d  0:5h)  0:5As  s1 (d  0:5h)
(1  2/ac)x a  kax
x  kxh

2‰
by equilibrium z
2x /εc
j and then the dimensionless forms nc  ackx

ns ¼ N s =bhf cd ¼ 0:5ø( s þ  s1 )= f yd εc1

(b)
ms ¼ M s =bh2 f cd ¼ 0:5ø( s   s1 )(0:5  
as )= f yd
Figure 6. Calculation of internal forces when kx < 1: (a) parabola
stress distribution; (b) parabola–rectangular stress distribution
as shown in Figure 6(a).

The calculation for concrete is more complicated because of its mc ¼ N c (0:5h  a)=bh2 f cd ¼ Æc k x (0:5  k a k x )
non-linear constitutive relation and the changed neutral axis.
Parameters Æc and ka are introduced to determine the stress
resultant of concrete (Nc) and its location. Assuming as shown in Figure 6. Therefore, the internal forces m and n are
expressed as

N c ¼ Æc bxf cd 8 ø
>
> n¼ ( s þ  s1 )  Æc k x
>
< 2 f yd
   
one obtains >
> ø 1 1
>
:m ¼ ( s   s1 )  as þ Æc k x  kakx
9: 2 f yd 2 2
nc ¼ N c =bhf cd ¼ Æc k x

The expressions for Æc, ka and strain are then determined by

Assuming the height of Nc is a ¼ kax ¼ kakxh yields equilibrium and constitutive relationships. When kx , 0, we have

126
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

Æc ¼ 0, ka ¼ 0 because the stress in concrete is zero. When Formulas of arbitrary m–n expressed by strain when
0 < kx < 1, the shape of the strain distribution in the compression kx > 1
area is a triangle, in which local coordinate system OZ is estab- Similarly, for the calculation for concrete, parameters Æd and kd
lished, and the position of the neutral axis is set as the original are introduced. Assuming the stress resultant of concrete
point, as shown in Figure 6. For any one of z, the strain of it is zc / N c ¼ Æd bhf cd , one obtains nc ¼ Æd : The distance between Nc
x. From Equation 2, there are two possible stress distributions: a and the centroidal axis is supposed as e ¼ kdh, as shown in Figure
parabola distribution when 0 > c > 2‰, and a parabola–rectan- 7. By equilibrium, the internal forces m and n are
gular distribution when 2‰ > c > 3 .5‰. Accordingly, two
integrals are obtained from 0 to x in the OZ system, which are 8 ø
>
> n¼ ( s þ  s1 )  Æd
>
< 2 f yd
ð
 
N c ¼ b  dz >
> ø 1
>m ¼
:  as ( s   s1 ) þ Æd k d
12: 2 f yd 2

and
By the superposition principle shown in Figure 7, the trapezoid
ð shape of the strain distribution along the sectional depth becomes
M c ¼ b (x  z) dz a subtraction of two triangles and, in this way, the above results
can be used here. Thus, one obtains

with respect to the topmost edge of the cross-section. With

nc ¼ nc1  nc2

Æc ¼ N c =(  bxf cd ) e ¼ [nc1 (0:5  a1 =h)  nc2 (0:5 þ a2 =h)]=nc

and and two possibilities (parabola or parabola–rectangular stress)

should be considered. With Equations 10 and 11, one
ð obtains
a ¼ M c =N c ¼ b (x  z) dz=(  Æc bxf cd )

one obtains

8 ð x "   2 #
>
> 1 z 1 z (6 þ c )c
Ðx >
> c þ c dz ¼  0 > c > 2‰
b  dz < x 0 x 4 x 12
Æc ¼ 0 ¼ (ð "  2 #   )
bxf cd > 2x=c  
>
> 1 z 1 z 2 2
>
: c þ c dz  1 þ x ¼1þ 2‰ > c > 3.5‰
10: x 0 x 4 x c 3c

Ðx
a b 0 (x  z) dz
ka ¼ ¼
x xÆc bxf cd
8 "  2 #
>
> Ð z 1 zc
>
>
x c
>
> 0 (x  z) þ dz
>
> x 4 x 8 þ c
>
>
< ¼ 0 > c > 2‰
Æc x2 4(6 þ c )
¼ "
>
>  2 #  
>
>
>
Ð  2x= c
z c 1 zc x2 2 2
>
> (x  z) þ dz  1þ
>
>
0 x 4 x 2 c 2 þ (4 þ 3c )c
>
: ¼ 2‰ > c > 3.5‰
11: Æc x2 2(2 þ 3c )c

127
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

εc 0·5ωσs1 /fyd
εs1 Arbitrary m and n are obtained by the above equations, which are
a1
m also suitable for the ultimate m–n calculation. However, general
n nc  ad
h nc1 analytical expressions in the ultimate state (Leonhardt, 1984)
e
cannot be used to calculate arbitrary m–n in the case of kx .1.
0·5ωσs /fyd
εs
εc1
Two functions
xh a2 nc2
Set function 1 as the process to calculate internal forces by two
strains, namely
Figure 7. Calculation of internal forces when kx . 1

8 (m, n, ) ¼ fun1(c , s )
> (6 þ c )c
< kx 0 > c > 2‰
nc1 ¼  12 
> 2
: 1þ kx 2‰ > c > 3.5‰ in which the analytical expressions derived above are used.
13: 3c
Set function 2 as the algorithm to calculate the strain of point
iend , namely
8
>
> 8 þ c
< kxh 0 > c > 2‰
4(6 þ c )
a1 ¼ 2 þ (4 þ 3 )
> [c (iend ), s (iend )] ¼ fun2(n)
>
:
c c
kxh 2‰ > c > 3.5‰
14: 2(2 þ 3c )c
in which function 1 is used. Although this is a traditional iterative
Because of only one possibility of the parabola stress distribution, process, the rate of convergence is accelerated with the help of
and with Equations 10 and 11 the regulations of strain distributions of point iend as demonstrated
above.
(6 þ c1 )c1
nc2 ¼  (k x  1) The flow diagram for function 2 is shown in Figure 8. First, the
12
damage type can be judged by comparing the magnitudes of n, n1
8 þ c1 and n2, where n1 and n2 are determined by Figure 3(d) and
a2 ¼ (k x  1)h
15: 4(6 þ c1 ) function 1. Then, the strain of one fibre is known according to
the damage type, for example s ¼ 20‰ in case 1. Next, as an
example, the dichotomy method is used to determine the
unknown strain of the other fibre and the interval [1, 2] is
Thus Æd and kd are determined by
introduced. Specifically, as an example, in case 1 the strains
change in region 1 of Figure 2, and therefore we have: the
Æd ¼ nc1 þ nc2 minimum curvature is zero and this value is given to 1 as an
8 3 initial value; by Equation 7 the maximum curvature is
>
> c1 þ 62c1  3c  62c
>
>
< 0 > c > 2‰ (20  3 .5)h/d ¼ 23 .5/(1  
as ) (see Figure 3(d)) and this value is
12(c  c1 ) given to 2 as an initial value. Finally, the trial value of n9
¼
>
> (2 þ c1 )3 is obtained by  ¼ 0 .5(1 + 2) and function 1, and the
>
>
:1 þ 2‰ > c > 3.5‰ unknown strain will be found when the difference value |n  n9|
16: 12(c  c1 )
meets the requirement error .

   
a1 a2
nc1 0.5  þ nc2 0.5 þ
e h h
kd ¼ ¼
h nc1  nc2
8
>
> (c  c1 )2 (c þ c1 þ 4)
>¼
> 0 > c > 2‰
< 4(3c þ 62c  3c1  62c1 )
¼
>
> (2 þ c1 )3 (2c  c1 þ 2)
>
> 2‰ > c > 3.5‰
:¼
17: 4(c1  c )(8 þ 62c1 þ 3c1 þ 12c )

128
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

Input: n, ω, fyd, as, δ Input: n, ω, fyd, as, iend, δ

Equations 3–5
(n1)  fun1 (εc  3·5, εs  20)
(n2)  fun1 (εc  3·5, εs  3·5as ) [εc(iend), εs(iend)]  fun2(n) εc(1), εs(1)  εs1  ε(1)
εs2 εc(iend) m(1)  0, φ(1)  0

Case 1: n  n1 Case 3: n  n2
n, n1, n2
Δ  [εc(iend)  ε(1)]/(iend  1)
Case 2: n2  n  n1

εs(iend)  20 εc(iend)  3·5 εr(iend)  2 εc(i )  ε(1)  (i  1)Δ

φ1  0 φ1  3·5 φ1  0
φ2  23·5/(1  as ) φ2  23·5/(1  as ) φ2  3·5

εs(i )  0·5(εs1  εs2)

φ  0·5(φ1  φ2)
(n , m , φ )  fun1(εs(i), εc(i))
εs(iend), εc(iend)
No No
εs1  εs(i) |n  n |  δ εs2  εs(i)
n  fun1[εs(iend), εc(iend)] n n n n

Yes
No No
φ1  φ |n  n |  δ φ2  φ m(i)  m , φ(i)  φ
n n n n
Yes
ii1
Yes
i  iend εs1  εs(i  1)
Output: εs(iend), εc(iend) εs2  εs(iend)
No
Figure 8. Flow diagram of [c (i end ), s (i end )] ¼ fun2(n)
Output: n, m, φ

Figure 9. Flow diagram for m– curve with fixed n

Moment–curvature relationship
Algorithm
Figure 9 shows the flow diagram for construction of the Application
m– curve. With the possible strain distributions, we clearly By computer programming, an m– curve is obtained as shown
know the first and the last strain distributions and how the in Figure 10. This is a monotonically increasing curve from the
strain process proceeds, which greatly benefits the rate of origin, and the entire curve shows an almost tri-linear shape
convergence. divided by the crack point ic and yield point iy. The slope is
greatest at the start of the tri-linear shape and then decreases
It is a traditional iterative procedure. Giving a regular interval ˜ gradually towards the end due to concrete cracking and reinforce-
to c, we obtain a definite array of c, namely, c(1), c(2), . . ., ment yielding. Especially note that the first yielding point of the
c(i), . . ., and c(iend). For each c, as an example, the dichotomy reinforcement creates a big tangent change, which means that
method is also used to determine the s that should meet the flexural stiffness decreases greatly once the cross-section enters
equilibrium requirement of the axial force, in which s1 and s2 the inelastic phase.
are introduced to be the minimum and maximum. Hence, the
corresponding array of s is obtained as s(1), s(2), . . ., s(i), . . ., Figure 10 shows that points iy and iend are essential for the m–
and s(iend). Finally, the two-dimensional array m– determined relationship of a doubly reinforced concrete section. On the one
by the two strain arrays is the final result needed to draw the hand, the moment my basically represents the sectional strength,
curve, namely, m(1)–(1), m(2)–(2), . . ., m(i)–(i), . . ., and because there is only a slight increment for the moment and using
m(iend)–(iend). this increment makes little sense as it takes a lot of curvature,

129
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

mu increasing n. Under small compression, however, the axial

0·5 iend force is favourable: flexural capacity increases slightly with
my iy
an increase in n. Nevertheless, the curvature (deformation
0·4 Ultimate point
Yield point capacity) of all three types decreases with increasing n and
Moment, m

this is especially significant in the case of compression (i.e.

0·3
greater n results in smaller  and smaller n results in
Plastic zone greater ).
0·2
(b) In the plastic zone, the curve is longer in length and presents
0·1 an almost horizontal line under small compression and
ic Crack point φy φu tension, meaning that, here, the increment of flexural capacity
i1
is limited although curvature has a much larger increment.
0 1 2 3 4 5 6 7 8 9 10 11
Curvature, φ: ‰ Under large compression, the curve length becomes shorter,
meaning worse sectional ductility; here, the curve is no
Figure 10. m– curve of rectangular RC section: n ¼ 0 .3, longer horizontal, which is an apparent difference from the
ø ¼ 1, fyd ¼ 435 N/mm2 and d ¼ 0 .9h steel section.

Influence of ö on m–n curve

which is a typical characteristic of flexural capacity of a doubly The same analogy is extended to determine the m–n curve with
reinforced concrete section (i.e. the bending capacity is almost constant curvature . For the array of c(1), c(2), . . ., c(i), . . .,
exhausted once the reinforcement yields). On the other hand, the and c(iend), the array of s(1), s(2), . . ., s(i), . . ., and s(iend) is
relationship between point iy and point iend represents the plastic obtained by Equation 7, and then the two-dimensional array of
flexure capacity of the cross-section. The cross-section is capable m(1)–n(1), m(2)–n(2), . . ., m(i)–n(i), . . ., and m(iend)–i(iend) is
of a certain deformation capacity after the reinforcement yields, obtained by function 1.
which can be reflected by the horizontal length of the plastic zone
(see Figure 10) and quantified by the curvature ductility factor, Figure 12 shows such m–n curves for two different mechanical
which is the ratio of the ultimate curvature u to the yield reinforcement ratios (ø ¼ 0 .5 and ø ¼ 1). Obviously, the two
curvature y. diagrams present almost the same characteristics apart from size.
The influence of  on the m–n curve is obvious: the greater
Influence of n on m–ö curve curvature, the closer is the curve to the envelope (the m–n
The m– relationship depends on the existence of n. According interaction curve). For small curvature, the diagrams are loose
to the different responses under all possible n from ultimate and the m–n curves change almost linearly with curvature. For
tension to ultimate compression, there are three types, as shown large curvature, the diagrams become dense: the curves almost
in Figure 11. The differences between the three types can be overlap and form a narrow shadow band in the lower right-hand
summarised as follows. part of the diagram – this means that with a further increment in
curvature, sectional strength capacity will not increase but will
(a) Generally, the axial force is unfavourable. Because the slightly decrease. In addition, Figure 12 shows the position of the
existence of n makes the cross-section enter the plastic maximum curvature, max ¼ (20 + 3 .5)/0 .9 ¼ 26.11 from Equa-
stage early, flexural capacity decreases greatly with tion 7, which is independent of ø.

0·50 0·4
0·5 0·5 0·3 0·5
0·5 0·2
0·75 0·1 n0
0·4 0·4 0·4
1·00 n0 0·2
m 0·3 1·25 m 0·3 m 0·3
0·4
0·2 1·50 0·2 0·2
0·6
0·1 n  1·75 0·1 0·1
0·8
0 0 0
0 2 4 6 8 0 5 10 15 20 0 5 10 15 20 25
φ: ‰ φ: ‰ φ: ‰
(a) (b) (c)

Figure 11. Influence of n on m– curves: (a) under large

compression; (b) under small compression; (c) under tension

130
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

2·0
expressions with respect to two strains. Accordingly, including
the ultimate strength, the internal forces of an arbitrary stress
1·5 state can be obtained. The numerical integration method is used
m–n interaction curve in the current study. The proposed method could also be extended
to consider the effects of stress hardening and non-symmetric
1·0
5 distribution of reinforcement.
Axial force, n

φ  0·5 1·5 6
0·5 The m– values of all the points can then be calculated through
1 2 3 4 7
the analytical formulas expressed by two strains. By computer
8
programming, an m– curve with constant axial force and an
0
m–n curve with constant curvature can be obtained efficiently.
φmax  26·11 The results show that
0·5
j axial force may be favourable because the ultimate moment
may increase with increasing axial force under small
1·0
0 0·1 0·2 0·3 0·4 0·5 compression
Moment, m j sectional ductility becomes poor under large compression
(a)
j the greater the curvature, the closer is m–n to the ultimate
2·0 value.
m–n interaction curve
1·5 Acknowledgement
This work was financially supported by the National Natural
Science Foundation of China (51468026, 51268022 and
1·0 5 51308269).
Axial force, n

φ  0·5 6
1·5
0·5 REFERENCES
1 2 3 4 7 Chandrasekaran S, Nunziante L, Serino G et al. (2011) Curvature
8 ductility of RC sections based on Eurocode: analytical
0
procedure. KSCE Journal of Civil Engineering 15(1): 131–
144.
φmax  26·11 Chen WF and Atsuta T (1976) Theory of Beam Columns: In-
0·5
Plane Behaviour and Design Vol. 1. McGraw-Hill, New York,
NY, USA.
1·0
0 0·1 0·2 0·3 0·4 0·5 Choi E, Chung YS, Park C et al. (2013) Seismic performance of
Moment, m circular RC columns retrofitted by prefabricated steel
(b) wrapping jackets. Magazine of Concrete Research 65(23):
Figure 12. Influence of  on m–n curves: (a) ø ¼ 0 .5; (b) ø ¼ 1 1429–1440.
Colajanni P, Fossetti M and Papia M (2013) Analytical prediction
of ultimate moment and curvature of RC rectangular sections
in compression. Bulletin of Earthquake Engineering 11(6):
2055–2078.
Conclusions Guler S, Çopur A and Aydogan M (2013) Axial capacity and
For a given axial force and a given cross-section, the sectional ductility of circular UHPC-filled steel tube columns.
damage type can be judged by introducing two parameters n1 and Magazine of Concrete Research 65(15): 898–905.
n2. The strain of the first point on the m– curve is analytically Jirásek M and Bažant ZP (2001) Inelastic Analysis of Structures.
derived, which is a mean compressive strain. The strain of the last Wiley, New York, NY, USA.
point, which is in the ultimate state, can be obtained by the Kordina K and Quast U (1972) Bemessung von schlanken
proposed algorithm and the strains of the remaining points are Bauteilen – Knicksicherheitsnachweis. In Beton-Kalender.
determined using a traditional iterative procedure. Therefore, all Ernst & Sohn, Berlin, Germany, pp. 695–793 (in German).
the possible strain distributions on the m– curve can be Leonhardt F (1984) Grundlagen Zur Bemessung Im
determined. It should be noted that this procedure is widely Stahlbetonbau. Springer, Berlin, Germany (in German).
applicable and is not confined to the assumptions in this paper. Löhning T, Schenk J and Starossek U (2009)
Querschnittsberechnung von Stahlbetonstäben – Stand der
Based on foregoing assumptions, the stress resultants of concrete Forschung. Beton- und Stahlbetonbau 104(4): 236–242 (in
and their position can be determined by the proposed analytical German).

131
Magazine of Concrete Research New procedure for determining the
Volume 67 Issue 3 moment–curvature relationship of a
reinforced concrete section
Chen, Zhou, Wang and Zhang

Montes EH and Aschleim M (2003) Estimates of the yield 529–538 (in German).
curvature for design of reinforced concrete columns. Sheikh MN, Tsang HH, McCarthy TJ et al. (2010) Yield curvature
Magazine of Concrete Research 55(4): 373–383. for seismic design of circular reinforced concrete columns.
Park R and Paulay T (1975) Reinforced Concrete Structures. Magazine of Concrete Research 62(10): 741–749.
Wiley, New York, NY, USA. Sheikh SA and Yeh CC (1993) Analytical moment–curvature
Pfeiffer U and Quast U (2003) Nichtlineares Berechnen relations for tied concrete columns. Journal of Structural
stabförmiger Bautale. Beton- und Stahlbetonbau 98(9): Engineering 118(2): 529–544.

WHAT DO YOU THINK?

To discuss this paper, please submit up to 500 words to
the editor at [email protected]. Your contribution will
be forwarded to the author(s) for a reply and, if
considered appropriate by the editorial panel, will be
published as a discussion in a future issue of the journal.

132
View publication stats