ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 86-S9

Creep Model for Reinforced Concrete Columns

by Raed M. Samra

A rational approach for calculating the serviceability structural ef- are much too small compared to the failure strain of
fects of creep on reinforced concrete columns and beams is pro- the materials involved. However, a verification of the
posed. The approach uses as input creep properties, exemplified by
internal stresses created by creep when the steel is
the creep coefficient, and produces as output deflections of members
and stresses in concrete and steel. The approach defines an effective placed unsymmetrically and its contribution to the
modulus of elasticity for the reinforced section and uses a trial and load-carrying capacity is very large must be considered.
error approach for its solution. The application of the method is While precautionary measures are useful and would in
made simple by the use of a microcomputer. The results of a specific many instances allow the problem to be designed
example, solved by means of this approach and compared with pro-
around, they are not helpful in providing information
cedures given in the ACI Building Code (3/8M-83) and British Stan-
dards Institution Standard BS 8/10, point toward good correlation about how the volume of concrete reacts to load, time,
and, considering the degree of uncertainty associated with the prob- and environment.
lem, confirm that the method is reasonably accurate.
RESEARCH SIGNIFICANCE
Keywords: beams (supports); columns (supports); creep properties; deflection'; This paper studies the behavior of reinforced con-
reinforced concrete: serviceability.
crete columns under sustained service load stresses.
Such a problem has not been addressed satisfactorily by
The importance of creep in concrete structures is al- design codes. The Building Code Requirements for
ready well established. Creep can affect internal stresses Reinforced Concrete (ACI 318M-83) (metric code), 2 for
in two ways: it can reduce existing internal stresses so instance, proposes a procedure for incorporating the
that they are of a continuous nature, and it can create effects of creep on the strength of columns, particu-
its own new internal stresses when materials with dif- larly slender columns. It introduces a creep factor {3d
ferent creep characteristics interact in a given cross sec- whose function is to magnify the moments in slender
tion. Such internal stresses created by creep itself occur columns. The design is then carried out for the axial
primarily in composite materials. One example would load and magnified moment. This approach does not
be a reinforced concrete column under permanent load. provide information about the nature of the behavior at
Creep causes contraction of concrete but not of steel. sustained working loads, a case significant in the study
Since plane sections remain plane, the stresses decrease of serviceability requirements.
in the concrete over time as the stresses increase in the
steel. The same redistribution is produced by shrink- PROCEDURE
age; these internal stresses are, in turn, reduced by In plain concrete members, the effects of creep have
creep. As a result of this redistribution caused by creep often been accounted for by using a modified (also
and shrinkage, compression reinforcement in under- called reduced or effective) modulus of elasticity. This
reinforced columns may reach the yield point even un- reduced modulus considers the reduction in stiffness of
der working loads. To prevent the columns from buck- a plain concrete member as a result of the occurrence
ling, an adequate concrete cover and an appropriate use of long-term strains caused by creep. The modulus is
of lateral ties are required. 1 In the dimensioning of given by
structural reinforced concrete elements, these internal
stresses created by creep are never taken into account. Ec
E =-- (1)
They may, however, require structural precautions such " I + cf>,
as close spacing of stirrups in members that will be
Received Oct. I, 1987, and reviewed under Institute publication policies.
subjected to high compressive stresses. Nonetheless, this Copyright © 1989, American Concrete lnstiiUie. All rights reserved, including
cannot have a measurable effect on load-carrying ca- the making of copies unless permission is obtained from the copyright propri-
etors. Pertinent discussion will be published in the November-December 1989
pacity since the strains caused by the internal stresses ACI Structural Journal if received by July I, 1989.

ACI Structural Journal I January-February 1989 77

 !::../;
AC! member Raed M. Samra is an assistant professor of civil engineering at the Eel Eci(1 + c/> 1)
r~ !::../;de/>; (2)
University of Jordan, Amman, Jordan. He holds engineering degrees from I he o Ec Ec
American Universily in Beiru/ and /he Universi/y of lllinois a/ Urbana-Cham-
creep
paign. His research interests are analysis and design of reinforced concrete initial plus short-term
s/ruc/ures, wilh a special focus on time-dependent behavior and computer ap- recovery
creep strain recovery
plications in structural engineering. strain
strain

in which

14 -

IZ

10-
2000

1800

1600

1400
rJ
As : 6000mm•

F~ : 21 MP0
Ee1

E"
!::../;

!::../;
strain in concrete after a time t, under
sustained load
strain in concrete upon application of load
stress decrement in concrete due to transfer of
force from concrete to steel, during time t and
stress decrement in concrete during an
IZOO __ P 'fOOD kN
infinitesimal time interval i.
or
~ Eco _ Effective
00
a: Modulus of Reinforced
Eq. (2) demonstrates the well-known phenomena that
0
3 6
BOO a!1
._ Section
.
columns under sustained load exhibit a continuous, but
"' Ecr - Effective Moclllus decreasing, transfer of force from concrete to steel.
"'0 of Plain Concrete
6 00 Section Since the stress in concrete reduces with time, this re-
sults in both immediate and creep recoveries. The creep
00
recovery portion, at a given moment, is assumed to oc-
zoo cur at the same rate as the creep. This is in accordance
with the method of superposition postulated by
1.5 z.o 25 3.0
McHenry. 3 Hence, the total creep recovery during time
0.5 '0
t under sustained load is obtained by multiplying the
CREEP COEFI'lCfENT, ~ strain decrement, in an infinitesimal time interval i, !::../;/
( 1nvn=0.0394 in 1
2
1mrl = l5.S.l'din 1 t kN=0.225 kip J
En by the creep factor increment d¢; during that time
interval and summing the effect over the entire creep
recovery range 0 ->c/> 1•
Fig. ]-Comparison of effective modulus for a plain The concrete stress fm after a time t, is given by
and reinforced section of an axially loaded member

(3)
where

instantaneous modulus of elasticity of the con- The term under the integral sign may be approximated
crete by 1::../;cf>,/2£., and making this substitution in Eq. (2)
reduced modulus of elasticity of the concrete yields
creep coefficient (also called creep factor) de-
fined as

creep strain
cf>l
initial strain
In approximating the term under the integral sign, an
average stress of (\12)1::.../; is assumed and multiplied by
In this paper, a simple procedure for predicting the the creep recovery range cf>, i.e., creep recovery strain
effects of creep in reinforced concrete is devised. The = a!::..J;c/> 1 where a = Y2. This value for a is only an ap-
approach is based on the study of the behavior of rein- proximation because the stress reduction is not strictly
forced concrete members under sustained load, with the linear. But it is also not far off from being linear, es-
resulting increase in deformations and stress redistri- pecially after an extended time under sustained load. A
bution between concrete and steel. The solution incor- mor:e accurate, but also more rigorous, approach would
porates an effective modulus of elasticity for a member be to divide the time scale into a number of intervals
with reinforcement. A comparison between the moduli assuming a linear stress reduction in each interval, and
of elasticity for plain concrete and reinforced concrete then add the effects of creep recovery for all intervals.
is shown for the case of an axially loaded member (see Such an approach may be found in Samra4 but has the
Example 1 in Fig. 1). In a reinforced concrete member, disadvantage of requiring extensive creep data, which
the final strain in concrete, after a period of sustained makes it cumbersome for most cases of practical de-
load t, may be expressed in the following manner sign. A study of the effect of using different values of
78 ACI Structural Journal I January-February 1989
 in concrete and steel then may be computed from Eq.
~ Concrete Stress
(3) for the concrete, and by multiplying the steel strain
-----6 Steel Stress determined in Step 3 by its modulus of elasticity for the
7.0
steel. Thus in concrete: f,, = f,; - !::..j;, and in steel: fs,
6.0 = E". E,.
Equilibrium is then checked by verifying that Ps + P,
5.0 = P, where Ps and P, represent the internal forces in
140
- ___ ..A- ___ ___ .~:r-- -- --"" concrete and steel (obtained by multiplying the stresses
4.0 120
0
in concrete and steel by the corresponding areas, re-
"'
"' ~
0

,; 100 spectively), and P is the axial compressive force on the

~
~
30
80 column. If this condition is not satisfied, a new stress
~ 2.0 60 decrement is assumed, as in Step 2, and Steps 2 and 3
8 40 ~ are repeated until equilibrium is guaranteed. The solu-
1.0
20
tion involves an iterative procedure and may require
several trials. Procedures for assessing the convergence
05 1.0 1.5 20 25 3.0 of the solution may be used but will not be discussed
Creep C011ffiOent.
here. Obviously, the use of a simple computer program
greatly enhances the solution. Example I helps clarify
( J MPo : /45.0 Psi J
the procedure.

Fig. 3- Variation of concrete and steel stress with creep

coefficient (Example 1-Axial/y loaded member)
Members subjected to combined thrust
and bending
This case is slightly more complex than the previous
00 kN 80 kN Section Properties
I
one. Two separate iterations are required, one on the
8 : 300mm, H : 250rnm

A'§.;._ _l__~---,~"'-----'----_~"1>/#• CO'Icrete Cover= 65mm

stress decrement !::..j;, and the second on the position of
the neutral axis C" and two equilibrium conditions are
A 5 1 Midspan J:942mm1.
~--~ rn + 3m + 3m + 3m --1- checked, namely, force equilibrium and moment equi-
As (Middle Support J=l 140mnf
librium. The procedure of the solution is as follows:
Materials Fe! : 21 MP0 . F y :42QMP0
1. Determine the stress and strain distributions in the
section, in concrete and steel, by considering the effect
--- ---
110 .0
on the uncracked or cracked section of a short-term
,o-- -- 0.4
load. (In this paper, only the cracked case for a mem-
E 10 o ber under combined thrust and bending is considered,
e / ~·6•\
2:
/
but the approach is equally applicable to the uncracked
If J1'
~9.0 'b'',~/ 0.35 .~
section.) See Fig. 2.
:i ~;/ 2. Assume the stress decrement !::..j; in the extreme
...
y/
z
g
""2:8 .0 p' concrete fiber in compression. From Eq. (6), with the
~
/ q,"'?

...Q // (J-#'' "'

known creep coefficient ¢>" calculate the apparent
~ / / 0.3 0
modulus of elasticity, and hence calculate the final
"' 7.0
0
/
./ concrete strain and final concrete stress from Eq. (5)
// and (3), respectively.
/'
6.0 0.236 3. Assume the depth of the neutral axis and check the
o. 5 1.0 1.5 2.0 2.0 30
equilibrium of forces; i.e., P, + Ps, + Ps2 = P, where
CREEP COEFFICIENT~
Ps 1 and Ps2 represent the forces in the compression and
( 1mm : 15. S.trfin~ 1m = 3.28ft, 1 MPo =145.0 psi, t kN=0.225kipJ
2
tension steel, respectively. Repeat the position of the
neutral axis until force equilibrium is achieved.
Fig. 4-Deflection versus creep coefficient for a con- 4. Calculate, for the stress distribution and position
tinuous beam (comparison of procedures) of neutral axis determined in Steps 2 and 3, the internal
moment in the section. Check if moment equilibrium is
2. After a time t of sustained load, and due to creep satisfied, i.e., Ms, + Ms 2 + M, = M, where M 51 , M, 2 ,
effects, concrete stress is reduced while the steel stress and M, represent moments of resistance of the
is increased. The stress reduction !::..j; is assumed. Then, compression steel, tension steel, and concrete, respec-
knowing the creep coefficient after time t for the given tively. If equilibrium is not achieved, revise the as-
concrete mix and substituting it into Eq. (6), the appar- sumption of the stress decrement in Step 2 and repeat
ent modulus is determined. Steps 2 through 4 until both force equilibrium and mo-
3. From Eq. (5), the final strain in concrete is calcu- ment equilibrium are correct. Example 2 (see Fig. 4 )
lated. Because of strain compatibility, this strain is applies this procedure to the case of a beam in which P,
equal to the steel strain, i.e., E_" = E,,. The final stresses + P" + P, 2 in Step 3 is set equal to zero.
80 ACI Structural Journal I January-February 1989
 DISCUSSION of the critical section in the column. In the past, the
It is difficult to compare the approach suggested in effective modulus method had been applied to the
this paper with design code procedures for the study of analysis of long columns 10- 12 , but there is a considerable
the effects of creep on reinforced concrete elements. In difficulty in determining the appropriate value of the
the Building Code Requirements for Reinforced Con- sustained modulus since it ignores any changes in stress
crete ACI 318M-83, 1 for instance, the creep coefficient history and shift in the neutral axis. A modified effec-
is not a parameter in the scheme suggested for calculat- tive modulus, as suggested herein, can handle such
ing long-term deformations. However, by estimating changes and, thus, is a better approach.
the period under load for a particular concrete mix to
attain a given ultimate creep coefficient, data may be CONCLUSIONS
generated that could be used to compare the two ap- The proposed procedure for the computation of the
proaches. Another complication arises because the pro- serviceability effects of creep on reinforced concrete
cedure in the ACI Building Code does not separate elements is simple, versatile, and lends itself easily to
creep deformations from shrinkage deformations, computer programming. The introduction of a so-
whereas this paper treats creep deformations individ- called apparent modulus for the reinforced section
ually and does not address the problem of shrinkage greatly simplifies the solution by readily relating con-
directly. To compare results, a breakdown of the crete stresses and strains while considering changes in
expression in the Code between creep deformations and stress and position of the neutral axis. The approach is
shrinkage deformations is needed. This is done in ac- valuable in the study of axially or eccentrically loaded
cordance with the approximate equations provided by columns, which have not been given enough attention
Branson. 6 •7 The results of this comparison and of an- in the design codes and which may be particularly sig-
other comparison between the approach suggested in nificant when considering buckling of columns. The
this paper and that of the British Standards Institution accuracy of the procedure in predicting deflections and
(BS 8110), 8 are shown in Fig. 4. stresses in concrete and steel is fairly good.
A study of the results presented in Fig. 4 indicates a
fairly good correlation between the approach of this
CONVERSION FACTORS
paper and that of ACI 318M-83 or British Standards lmm 0.0394 in.
Institution BS 8110, for the case of a two-span contin- lm 3.28 ft
uous beam with concentrated loads. Such a correlation I mm' 15.5 x 10 • in.'
may be inadequate for a general assessment of the ac- I MPa 145.0 psi
lkN 0.225 kip-force
curacy of the method suggested here. However, given
the complexity of the problem of calculating long-term
REFERENCES
deflections and the innumerable cases that could be I. Ri.isch, Hubert; Jungwirth, Dieter; and Hilsdorf, Hubert K.,
proposed for comparison, the results of Fig. 4 show a Creep and Shrinkage: Their Effect on the Behavior of Concrete
positive correlation, indicating that the method may be Structures, Springer-Verlag, New York, 1983, 284 pp.
accurate enough for practical design p_urposes. Note 2. ACI Committee 318, "Building Code Requirements for Rein-
also that the accuracy of the method depends on both forced Concrete (ACI 3!8M-83)," American Concrete Institute, De-
troit, 1983, Ill pp.
the accuracy of the creep model and the accuracy of the
3. McHenry, Douglas, "A New Aspect of Creep in Concrete and
integration procedure adopted for calculating the de- Its Application to Design," Proceedings, ASTM, V. 43, 1943, pp.
flections along the member from the curvatures at dis- 1069-1084.
crete locations. It is suggested that the Newmark nu- 4. Samra, R., "Time-Dependent Deformations of Reinforced
merical procedure 9 be used for such purposes. This Concrete Columns," PhD dissertation, University of Illinois, Ur-
bana, Apr. 1982, 166 pp.
procedure is approximate but leads to near exact de-
5. Park, Robert, and Paulay, Thomas, Reinforced Concrete Struc-
flections when the diagram of curvatures is made up of tures, John Wiley & Sons, New York, 1975, pp. 457-461.
segments that are bounded by straight lines or by arcs 6. Branson, Dan E., "Compression Steel Effect on Long-Time
of parabolas. By taking more arbitrary divisions in the Deflections," ACI JouRNAL, Proceedings V. 68, No. 8, Aug. 1971,
length of a member, more accurate results are ob- pp. 555-559.
7. Branson, Dan E., Deformation of Concrete Structures," Mc-
tained. For most practical problems, no more than five Graw-Hill Book Co., New York, 1977, pp. 190-196.
or six segments are necessary. 8. "Structural Use of Concrete," (BS 8110), British Standards In-
Example 2 in Fig. 4 represents a case of five integra- stitution, London, 1985, Parts I and 2, 169 pp.
tion points per span. The primary advantage of the 9. Newmark, N. M., "Numerical Procedure for Computing De-
method is in studying the serviceability of columns with flections, Moments, and Buckling Loads," Transactions, ASCE, V.
108, 1943, pp. 1161-1233.
or without a bending moment, a problem not well dis- 10. Broms, Bengt, and Yiest, I. M., "Ultimate Strength of Hinged
cussed in the design codes. An even greater advantage Columns," Transactions, ASCE, V. 126, Part 2, 1961, pp. 309-339.
of the method is in extending it to study the buckling II. Pfrang, E. 0., "A Study of the Influence of Creep on the Be-
of columns under sustained load, particularly slender havior and Capacity of Reinforced Concrete Columns," Technical
columns. In this connection, the purpose is to calculate Report No. 4, Department of Civil Engineering, University of Dela-
ware, Newark, October 1964, 58 pp.
the creep buckling load of long columns on the basis of 12. Green, R., "Behavior of Unrestrained Reinforced Concrete
strain distribution in concrete under a sustained load Columns Under Sustained Load," PhD dissertation, University of
and the continuous change in the moment of resistance Texas, Austin, Jan. 1966, 246 pp.

ACI Structural Journal I January-February 1989 81

 APPENDIX-NUMERICAL EXAMPLE 1 5. Assume stress decrement in concrete !:,.f
A 300 x 300 mm' column has a total height of 2.5 m and is sub- Assume !:,.f = 3.288 MPa
jected to an axial load P = 1000 kN. Reinforcement is uniformly 6. Calculate the apparent long-term modulus £,.,
distributed in the section with a total area of 6000 mm'. If J:' = 21
MPa, j, = 420 \1Pa, and the creep coefficient </>, = 1.5, determine £,
the stresses and strains in concrete and steel before and after creep, £'"= ( ')!:,.J:
as well as the axial shortening of the column. (I + </>,) - I + 2_<1>· • };,

Solution 21,500
12,680 MPa (1.84 X 10" psi)
A. Before creep 3.288)
I. Calculate the modular ratio n (2 "5)- (1.7 5) ( 7.153

£, 2 X 10' MPa (29 X 10" psi) 7. Calculate the final strains in concrete and steel
(ACI 318M-83, Section 8.5.2)
£, 4700 .JJ: (AC! 318M-83, Section 8.5.1) f = b: = 7.153 = 5 64 X 10'
£, 4700 -}2T = 21,500 MPa (3.12 X 10" psi) " £,., 12,680 .
2 X 10'
n £,/£, = - - = 9.3
21,500 E, = f, = 5.64 X 10·'

2. Calculate the initial concrete and steel stresses using the trans- 8. Calculate the final stresses in concrete and steel
formed area approach. Concrete stress is given by
f, = };, - !:,.f = 7.153 - 3.288 = 3.865 MPa (560.4 psi)
p
Ag + (n - !)A, .t:, f, · £, = (5.64 X 10 ') (2 X 10') = 112.8 MPa (16,356
psi)
concrete stress,J:, = (1000) (10')/(300 x 300) + (8.3) (6000) = 7.153
9. Calculate the final forces in concrete and steel
MPa (1037 psi)
steel stress, f.= n·j" = (9.3)(7.153)= 66.52 MPa (9645 psi)
P,, = f, · (A, - A') = (3.865) [(9 x 10') - 6000] (10 ')
324.7 kN (73.06 kips)
3. Calculate the inital forces in concrete and steel concrete force,
P.; = J:, · A, = J:, (A, - A,) P, = f, · A, = (112.8) (6000) (10') = 676.8 kN (152.28 kip)
= (7 .153) (9 X I 0' - 6000) (I 0 ')
= 399.1 kN (89.80 kip) 10. Check force equilibrium

4. Check force equilibrium P,, + P, = 324.7 + 676.8 = 1001.5 kN (225.4 kips) OK

P" = P, = 600.9 + 399.1 1000 kN (225 kip) OK
II. Calculate the axial shortening
B. After creep
Use an iterative procedure. The results of the final cycle are /:,., = f" · L = (5.64 X 10 ') (2.5) (10 ') 1.41 mm (0.0556
shown in the following. in.)

82 ACI Structural Journal I January-February 1989