You are on page 1of 15

Appendix D

Creep Models Recommended by Design Codes

D.1 General Structure of Creep Design Formulae

This appendix summarizes the main equations and parameters recommended for
creep design calculations by the International Federation for Structural Concrete
(known under the French acronym fib) and by the American Concrete Institute
(ACI).
The International Federation for Structural Concrete (in French “fédération in-
ternationale du béton, fib”) was created in 1998 by merging the Euro-International
Committee for Concrete (Comité Euro-International du Béton, CEB) and the Inter-
national Federation for Prestressing (Fédération internationale de la précontrainte,
FIP). The fib Model Code 2010 is a successor of CEB Model Codes 1990 and 1999.
The creep and shrinkage model incorporated into this code will be referred to as the
fib Model. The final draft published in 2012 substantially differs from the earlier
CEB Model.
The American Concrete Institute (ACI) had a permanent committee TC 209
Creep and Shrinkage in Concrete. The creep model adopted by this commit-
tee in 1971 (ACI Committee 209 1971) was developed in the late 1960’s by
Branson and coworkers (Branson, Meyers and Kripanarayanan 1970, Branson and
Christiason 1971, Branson 1977). In 1976, based on Bažant, Carreira and Walser
(1975), Bažant (as member of ACI TC 209) proposed a model with a much
higher multidecade creep, terminating with a logarithmic asymptote, which was
not adopted. The ACI 1971 Model was reapproved in the 1982 recommendations
(ACI Committee 209 1982), which were expanded by the AAEM method (see Sec-
tion 4.2). The most recent version, labeled as 209R-92, was published in 1992 (ACI
Committee 209 1992) and again reapproved (with the sole dissenting vote of the
first writer) in 2008. After the recent publication of bridge deflection data (Bažant,
Yu, Li, Klein and Křı́stek 2010, Bažant, Hubler and Yu 2011a, Bažant, Hubler and
Yu 2011b, Bažant, Yu and Li 2012a, Bažant, Yu and Li 2012b), a debate on a re-
vision has begun in ACI TC 209 by the time of writing. The ACI Guide for Mod-
eling and Calculating Shrinkage and Creep in Hardened Concrete (ACI Commit-

613
614 D Creep Models Recommended by Design Codes

tee 209 2008) for the first time presents not only the ACI Model, but also the B3
Model, the CEB Model (predecessor of the fib Model), and the Canadian GL2000
Model proposed by Gardner and Lockman (2001) and slightly modified by Gardner
(2004). Model B3 was approved in 1995 as an international RILEM Recommenda-
tion (Bažant and Baweja 1995a), as the sole model recommended by RILEM.
Concrete design codes and recommendations usually classify concrete according
to its strength. The fundamental property is the uniaxial compressive strength mea-
sured on cylinders of diameter 150 mm (6 in.) and height 300 mm (12 in.) at age 28
days. Due to inevitable scatter, the strength does not have a unique value (even for
concrete produced in the same plant from the same mix and the same batch using the
same technological procedure of mixing, casting and curing). So, the strength needs
to be considered as a random variable. The strength value used in design procedures
for assessment of the ultimate load-carrying capacity of a structure is not the mean
value but a certain “safe lower bound”.
In creep design, the structure is analyzed in the service state, and the mean value
of deflection is of most interest. If the elastic modulus and creep properties of con-
crete are estimated from strength, the mean value of strength (rather than the lower
bound) must be used. The codes specify an approximate formula that links the mean
strength to the safe lower bound, called the “characteristic strength” by fib and the
“specified strength” by ACI.
The fib Model Code (as well as the previous CEB Model Codes) uses the charac-
teristic compressive strength, fck , understood as the value below which only 5% of
all possible strength measurements may be expected to fall. The mean compressive
strength, fcm , is estimated as

fcm = fck + 8 MPa (D.1)

For instance, concrete of grade C30 has the characteristic cylinder strength of 30
MPa and the mean strength of 38 MPa. Gardner and Lockman (2001) proposed
another formula,
fcm = 1.1 fck + 5 MPa (D.2)
which gives exactly the same difference between fcm and fck for concrete of grade
C30 but takes into account an increase of the standard deviation for higher grades.
The ACI recommendations (ACI Committee 318 2005) are based on the specified
compressive strength, fc′ , which should satisfy the following two conditions:
• only 1 out of 100 test results is below a minimum value, defined as fc′ − 500 psi
for low-strength concrete with fc′ ≤ 5000 psi, and as 0.9 fc′ for normal-strength
concrete with fc′ > 5000 psi,
• only 1 out of 100 averages of 3 consecutive test results is below fc′ .
Here, a “test result” is understood as the average of values measured on two samples
(cylinders). The required average (mean) compressive strength fcr′ is estimated from
the specified compressive strength fc′ using the formula
D.1 General Structure of Creep Design Formulae 615
(
max ( fc′ + 1.34S, fc′ + 2.33S − 500 psi) if fc′ ≤ 5000 psi
fcr′ = (D.3)
max ( fc′ + 1.34S, 0.9 fc′ + 2.33S) if fc′ ≤ 5000 psi

where S is the standard deviation of tests (for the specific concrete plant, determined
from previous tests). If the standard deviation is not available, the required mean
compressive strength is evaluated as

f ′ + 1000 psi if fc′ < 3000 psi
 c


fcr′ = fc′ + 1200 psi if 3000 psi ≤ fc′ ≤ 5000 psi (D.4)


1.1 fc′ + 700 psi if fc′ ≥ 5000 psi

Note that 3000 psi = 20.7 MPa and 5000 psi = 34.5 MPa.
To facilitate the comparison among various codes and models, we use a unified
notation for the fundamental constants and variables: t = current time, t ′ = age
at loading, t0 = age at the end of curing (onset of drying), fc′ = characteristic or
specified strength, f˜c = mean compressive strength (dependent on age), f¯c = mean
compressive strength at age 28 days, E = elastic modulus (dependent on age), E28 =
conventional modulus, i.e., elastic modulus at age 28 days, φ = creep coefficient,
φ28 = creep coefficient in its alternative definition,1 henv = relative humidity of the
environment,V /S = ratio between the volume V of the concrete member and surface
area S exposed to drying. These symbols replace the original ones used in the code
specifications (e.g., fcm and Ecm for the mean strength and modulus, or RH for the
relative humidity). On the other hand, the original notation is kept for parameters
that are specific to each individual model. In all the equations to follow, time and
age should be substituted in days.
Design code specifications related to creep assessment usually combine the fol-
lowing components:
1. Time evolution of the (mean) elastic modulus, often derived from the time de-
pendence of (mean) compressive strength.
2. Dependence of the creep coefficient on the current time and the age at loading.
3. Definition of the compliance function in terms of the elastic modulus and creep
coefficient.
4. Set of rules for the estimation of model parameters based on the concrete mix
properties, type of curing, size and shape of concrete member, ambient relative
humidity and similar factors.

1 The standard definition considers the creep coefficient as the ratio of the creep strain to the

elastic strain induced by the same stress (as the creep strain) at the actual age of loading; see
equation (3.14). This standard creep coefficient is used by the ACI Model and will be denoted as
φ . However, the fib Model considers the creep coefficient as the ratio of the (actual) creep strain to
the (fictitious) elastic strain that would be caused by the same stress applied at age 28 days. This
alternative type of creep coefficient will be denoted as φ28 . It is used by the fib Model and also by
GL2000. No matter which specific definition of creep coefficient is used, it should be noted that
the very concept of elastic strain is highly ambiguous, as it depends on the duration within which
the load is applied, which can vary from 0.1 second to 1 hour.
616 D Creep Models Recommended by Design Codes

D.2 CEB and fib Model Codes

A design code developed by CEB and FIP was first approved and published in 1991
under the name of “CEB-FIP Model Code: Design Code” and republished in 1993
in CEB Bulletins No. 213 and 214. The creep description embodied in this code
was based on the work of a task group coordinated by Müller and Hilsdorf (1990).
CEB and FIP merged in 1998 into fib, and an updated version of the code appeared
in 1999 in fib Bulletins No. 1–3 and was co-opted in 2002 for Eurocode 2. A new
update is referred to as the “fib Model Code 2010” and its final draft was published
in 2012 (?).
The code considers the total strain in concrete as the sum of the initial strain (in
the sense of the instantaneous elastic strain), creep strain, shrinkage strain and ther-
mal strain. The sectional approach is used, i.e., the variation of internal stresses and
humidity across the cross section is neglected. Equations given below refer to the
mean sectional behavior of a concrete member moist-cured at normal temperatures
for not more than 14 days. They are intended for ordinary structural concrete with
the mean compressive strength between 20 MPa and 130 MPa, subjected to com-
pressive stress not exceeding 40% of the mean strength (at loading) and exposed to
an environment with the mean relative humidity between 40% and 100% and mean
temperature between 5◦ C and 30◦ C. The age at loading should be at least 1 day.

D.2.1 CEB Model

The creep description according to the original CEB Model can be summarized by
the following equations:
1/3
f¯c

E28 = 21.5 GPa × αE × (D.5)
10 MPa
r  
p
E(t) = E28 exp s[1 − 28/t] (D.6)

φRH β f
0.3
t − t′

φ28 (t,t ′ ) = (D.7)
0.1 + t ′0.2 βH + t − t ′
1 φ28 (t,t ′ )
J(t,t ′ ) = + (D.8)
E(t ′ ) E28

Equation (D.5) estimates the conventional modulus of elasticity from the mean
compressive strength at age 28 days. Parameter αE reflects the influence of aggre-
gate type and is set to 1.0 for quartzite aggregates, 1.2 for basalt and dense lime-
stone aggregates, 0.9 for limestone aggregates and 0.7 for sandstone aggregates.
The modulus of elasticity is understood as the unloading modulus in a static uniax-
ial compressive test, after previous loading to 40% of the compressive strength. The
modulus obtained from (D.5) corresponds to a load applied approximately within 1
D.2 CEB and fib Model Codes 617

second, while the modulus used by ACI and given by (D.28) corresponds to a load
applied within about 0.01 day.
Development of the elastic modulus with time (due to aging) is described by
equation (D.6), in which s is a parameter equal to 0.38, 0.25 or 0.20, depending
on the strength class of cement and hardening characteristics (e.g., s = 0.25 for
normally hardening cement of strength class 42.5 or for rapidly hardening cement
of strength class 32.5).
According to (D.7), the creep curves (after subtraction of the initial strain) have
the same shape, but their amplitude depends on the age at loading. Parameters φRH
and β f express the influence of relative environmental humidity, henv , and mean
compressive strength at age 28 days, f¯c , and are given by the following expressions:
"   #  0.2
1 − henv 35 0.7 35
φRH = 1 + 10 p ¯ (D.9)
3
2Ac /u f c f¯c
16.8
βf = p (D.10)
f¯c

The ratio 2Ac /u in (D.9) is the notional size of the member (to be substituted in
mm), evaluated from the area of the cross section, Ac , and the perimeter of the
member in contact with the atmosphere, u. The compressive strength, f¯c , has to be
substituted in MPa. Finally, parameter βH in (D.7) again depends on the strength,
relative environmental humidity and notional member size, and is given by
s
2Ac  35
βH = 1.5 18

1 + (1.2henv) + 250 ¯ (D.11)
u fc

If the right-hand side of (D.11) exceeds 1500 35/ f¯c , then βH is set to this value as
p

a limit value.
The effect of temperature on creep can be incorporated through the factor
 
1500
βT = exp − 5.12 (D.12)
T

where T is the absolute temperature in K. Factor βT multiplies βH in (D.7). It is


equal to 1 for T = 293 K, which corresponds to the reference temperature 20◦C.
In summary, the compliance function of the CEB Model, obtained by combining
equations (D.6)–(D.8), has the form
" r #!
φRH β f
0.3
t − t′

′ 1 s 28 1
J(t,t ) = exp − 1 − +
E28 2 t′ E28 0.1 + t ′0.2 βH βT + t − t ′
(D.13)
618 D Creep Models Recommended by Design Codes

D.2.2 fib Model

The updated version known as the fib Model Code 2010 reuses equations (D.5),
(D.6) and (D.8) of the original CEB Model but equation (D.7) is revised by adopt-
ing the separation of basic and drying creep introduced in models BP (Bažant and
Panula 1978), BPKX (Bažant, Kim and Panula 1991) and B3:

φ28 (t,t ′ ) = φbc (t,t ′ ) + φdc (t,t ′ ) (D.14)

The basic creep is represented by the unbounded logaritmic function2


 !2 
1.8 30
φbc (t,t ′ ) = ¯0.7 ln 1 + 0.035 + ′ (t − t ′ ) (D.15)
fc tadj

and drying creep by the bounded function


γ (t ′
φRH t − t′ adj )

412
φdc (t,t ′ ) = ¯1.4 (D.16)
′ )0.2
fc 0.1 + (tadj βH + t − t ′

with the mean compressive strength f¯c substituted in MPa.



In formulae (D.15)–(D.16), tadj is the adjusted age at loading, which reflects the
effect of elevated or reduced temperatures and of the cement type on the maturity
of concrete. For cement strength classes 32.5 R and 42.5 N, the adjusted age tadj′ is

equal to the temperature-adjusted age tT , defined by the Arrhenius-type rate equa-
tion
dtT′
 
QT QT
= exp − (D.17)
dt RT0 RT
in which QT /R = 4000 K, T0 = 293 K is the reference temperature, and T is the con-
crete temperature (mean value over the section), substituted in K. For other cement
strength classes, the temperature-adjusted age tT′ is transformed into the adjusted
′ according to the formula
age tadj
 α
′ 9
tadj = tT′ 1+ (D.18)
2 + (tT′ )1.2

with exponent α = −1 for strength class 32.5 N, α = 0 (leading to tadj′ = t ′ ) for


T
strength classes 32.5 R and 42.5 N, and α = 1 for strength classes 42.5 R, 52.5 N
′ ≥ 0.5 days.
and 52.5 R. The code stipulates that tadj

2 The decomposition into basic and drying creep and the asymptotic logarithmic character of the

basic creep compliance bring the fib model closer to models B3 and B4. These features were
not present in the first draft of the fib code, published in 2010, which was criticized in Bažant,
Yu, Hubler, Křı́stek and Bittnar (2011d). They were incorporated into the final draft, published in
2012.
D.3 ACI Model 619

Note that the load duration t − t ′ in (D.15)–(D.16) is based on the actual physical
time, not on the adjusted one. The influence of the member size on drying creep is
in (D.16) reflected by parameter
s
2Ac 35
βH = 1.5 + 250 ¯ (D.19)
u fc

Same as for the CEB model, the ratio 2Ac /u is the notional size of the memberp (to
be substituted in mm), and if the right-hand side of (D.19) exceeds 1500 35/ f¯c ,
then βH is set to this value as a limit value. The influence of environmental humidity
is taken into account by parameter
1 − henv
φRH = 10 p
3
(D.20)
2Ac /u

and the exponent in (D.16) is given by


1
γ (tadj

)= q (D.21)

2.3 + 3.5/ tadj

′ substituted in days, as usual.


with tadj
′ takes into account the effect of tem-
The temperature-adjusted age at loading tadj
perature prior to loading. The effect of temperature on the development of creep can
be incorporated in the same way as for the CEB model, i.e., using factor βT given
by (D.12), which multiplies parameter βH in (D.16). In addition to that, the basic
creep coefficient φbc should be multiplied by φT and the drying creep coefficient φdc
should be multiplied by φT1.2 , with

φT = exp[0.015(T − T0 )] (D.22)

where T0 = 293 K is the reference temperature. The code also specified a transient
creep coefficient, to be used if the increase of temperature occurs while the structural
member is under load.

D.3 ACI Model

The empirical model developed by Branson and Christiason (1971) was incorpo-
rated into the recommendations of the American Concrete Institute (ACI Commit-
tee 209 1971) and reapproved in later versions (ACI Committee 209 1982, ACI
Committee 209 1992, ACI Committee 209 2008). The question of a new update is
currently under discussion. The version presented here corresponds to the guide No.
209.2R-08, published by ACI Committee 209 in 2008.
The ACI Model uses the following equations:
620 D Creep Models Recommended by Design Codes

t ¯
f˜c (t) = fc (D.23)
a + bt
q
E(t) = 0.043 ρ 3 f˜c (t) (D.24)
(t − t ′ )ψ
φ (t,t ′ ) = φu (t ′ ) (D.25)
d + (t − t ′)ψ
1 + φ (t,t ′)
J(t,t ′ ) = (D.26)
E(t ′ )

Times t and t ′ have to be substituted in days. Parameters a and b in (D.23) depend on


the type of cement and type of curing. For moist-cured concrete and cement of type
I (ordinary Portland cement), their recommended values are a = 4 and b = 0.85. In
(D.24), ρ denotes the mass density of concrete in kg/m3 , the strength f˜c should be
substituted in MPa and the resulting modulus E is also in MPa. Combining (D.23)
with (D.24), we can describe the evolution of the elastic modulus by
r
t
E(t) = E28 (D.27)
a + bt
where q
E28 = 0.043 ρ 3 f¯c (D.28)
is the elastic modulus at age 28 days. For normal weight concrete, formula (D.28)
can be replaced by (3.6). If the modulus is measured directly, it is preferable to use
in (D.27) the actually measured (mean) value instead of the estimate (D.28). It could
be somewhat disturbing that if the typical parameters a = 4 and b = 0.85 are used,
the fraction in (D.27) is not exactly equal to 1 for t = 28 days. This is caused by a
truncation error. To get the exact coincidence between E(28) and E28 , one should
use b = 6/7 instead of b = 0.85, and equation (D.27) then reads
r
7t
E(t) = E28 (D.29)
28 + 6t
The creep coefficient specified in (D.25) has the standard meaning according
to the definition (3.14). The recommended values of the parameters in (D.25) are
d = 10 and ψ = 0.6. The ultimate creep coefficient for standard conditions is φu =
2.35. The standard conditions in the sense of the code are described by a number of
parameters specifying the concrete composition and curing, member geometry and
environment, and loading. For instance, it is assumed that the load is applied at the
end of curing, at age 7 days for moist-cured concrete or 1 to 3 days for steam-cured
concrete, and that the ambient relative humidity is henv = 40% and the volume-
surface ratio is V /S = 38 mm. The volume-to-surface ratio of the concrete member
V /S is equivalent to a half of the notional size 2Ac /u used by the fib Model Code if
the ends of prismatic specimens or members are sealed and thus not counted in S.
For instance, for an infinite slab, V /S is one half of the notional size.
D.4 GL2000 Model 621

For other than standard conditions, φu is corrected by the product of six factors
that depend on t ′ , henv , V /S, slump, ratio of fine aggregate to total aggregate by
weight, and air content:

φu (t ′ ) = 2.35 γ1 γ2 γ3 γ4 γ5 γ6 t ′−m (D.30)

where

1.25 for moist curing
γ1 = (D.31)
1.13 for steam curing
γ2 = 1.27 − 0.67 henv (D.32)
2 
γ3 = 1 + 1.13 e−0.0213V/S (D.33)
3
γ4 = 0.82 + 0.00264 ssl (D.34)
γ5 = 0.88 + 0.0024 a f /at (D.35)
γ6 = max(1, 0.46 + 0.09 αair) (D.36)

0.118 for moist curing
m= (D.37)
0.094 for steam curing

In these equations, t ′ is the age at loading in days, henv is the ambient relative hu-
midity, V /S is the volume-surface ratio in mm, ssl is the slump in mm, a f is the
mass fraction of fine aggregate, at is the mass fraction of total aggregate and αair is
the air content in percent.
In summary, the compliance function of the ACI Model, obtained by combining
equations (D.25)–(D.27) and (D.30) and substituting the recommended values of
parameters d = 10 and ψ = 0.6, has the form

2.35γ (t − t ′ )0.6
r  
′ 1 a
J(t,t ) = b+ ′ 1+ ′ m (D.38)
E28 t (t ) 10 + (t − t ′ )0.6

where γ denotes the product of factors γ1 to γ6 , given by (D.31)–(D.36).

D.4 GL2000 Model

The model proposed by Gardner and Lockman (2001) and denoted as the GL2000
Model is a major modification of the earlier Atlanta97 Model (or GZ Model) of
Gardner and Zhao (1993). The modification co-opts significant aspects of the 1978
BP Model (Bažant and Panula 1978), particularly the mathematical form of shrink-
age dependence on the drying time and thickness (or the volume/surface ratio),
and the additive separation of drying creep from basic creep. Minor adjustments
are incorporated in the final version (Gardner 2004), reproduced in the ACI Guide
209.2R-08 (ACI Committee 209 2008). The model is applicable to concretes with
characteristic strength up to 82 MPa that do not experience self-desiccation.
622 D Creep Models Recommended by Design Codes

The GL2000 Model in the form described in the ACI Guide 209.2R-08 can be
summarized by the following equations:
 p 
f˜c (t) = f¯c exp s(1 − 28/t) (D.39)
q
E(t) = 3.5 + 4.3 f˜c (t) (D.40)
" s s #
2(t − t ′ )0.3 7(t − t ′ ) t − t′
φ28 (t,t ) = Φ

+ + ch
(t − t ′ )0.3 + 14 t ′ (t − t ′ + 7) t − t ′ + 0.12(V/S)2
(D.41)
1 φ28 (t,t ′ )
J(t,t ′ ) = + (D.42)
E(t ′ ) E(28)

The strength evolution factor in (D.39) has the same form as in the fib Model
Code, but the relation between the elastic modulus and the strength is different. The
recommended values of the strength development parameter s are 0.335 for cement
of type I (ordinary Portland cement), 0.4 for type II (modified cement) and 0.13 for
type III (rapid-hardening cement). Equation (D.40) is valid in the SI units, with E
in GPa and f˜c in MPa. Combining (D.39) and (D.40), we can express the evolution
of elastic modulus more directly as
r  
p
E(t) = 3.5 + (E28 − 3.5) exp s(1 − 28/t) (D.43)

where q
E28 = E(28) = 3.5 + 4.3 f¯c (D.44)
is the conventional modulus.
In expression (D.41) for the creep coefficient, the volume-surface ratio V /S needs
to be substituted in mm. Parameter

ch = 2.5(1 − 1.086h2env) (D.45)

depends on the ambient humidity and vanishes for henv = 0.96, which is the value
approximately corresponding to the relative pore humidity under sealed conditions.
Therefore, the last term in the brackets in (D.41) (which contains ch ) corresponds
to drying creep, and the first two terms correspond to basic creep. The model also
takes into account the effect of drying before loading. If the member is loaded at
the same time as it is exposed to drying, parameter Φ is equal to 1 (and thus can be
omitted from (D.41)). However, if the first loading occurs at age t1 larger than the
age t0 at the onset of drying, the correction factor is evaluated as
s r
t1 − t0
Φ = 1− (D.46)
t1 − t0 + 0.12(V/S)2
D.5 JSCE Model 623

Note that Φ depends on the age at first loading, t1 , which does not need to coincide
with t ′ in the compliance function. Of course, in a standard creep test the load is
applied at once and t1 = t ′ . However, if the material response is evaluated at vari-
able stress according to the principle of superposition, t ′ in the integral stress-strain
equation sweeps the interval from t1 to t but the factor Φ to be used in (D.41) is
evaluated from t1 and thus remains constant.
In the original version of GL2000 (Gardner and Lockman 2001), the constant
multiplying (V /S)2 in (D.41) and (D.46) was 0.15 instead of 0.12, and the evolution
of strength was described by

t 0.75
f˜c (t) = f¯c (D.47)
a + bt 0.75
instead of (D.39).
In summary, the compliance function of the GL2000 Model in its final version,
obtained by combining equations (D.41)–(D.43), has the form
1
J(t,t ′ ) = " r #! + (D.48)
s 28
3.5 + (E28 − 3.5) exp 1−
2 t′
" s s #
Φ 2(t − t ′ )0.3 7(t − t ′ ) t − t′
+ + + ch
E28 (t − t ′ )0.3 + 14 t ′ (t − t ′ + 7) t − t ′ + 0.12(V/S)2

D.5 JSCE Model

The creep model recommended by the Japan Society of Civil Engineers (JSCE) as a
part of the Standard Specifications for Concrete Structures (Standard Specifications
for Concrete Structures – 2007 ”Design” 2010) is applicable to concrete with wa-
ter/cement ratio between 0.4 and 0.65 and strength up to 55 MPa (or up to 70 MPa
if the water/cement ratio is reduced to increase the strength), loaded by stresses not
exceeding 40% of the strength, at ambient relative humidities between 45% and
80%, and for volume/surface ratios between 100 mm and 300 mm.
The JSCE guidelines specify the additional strain due to creep separately from
the elastic strain. The corresponding compliance function can be written as
1 h  i
J(t,t ′ ) = ′
+ ∆ J(t ′ ) 1 − exp −0.09(t − t ′ )0.6 (D.49)
E(t )

where E(t ′ ) is the elastic modulus and

∆ J(t ′ ) = 1.5(c + w)2(w/c)2.4 (lnt ′ )−0.67 × 10−9/MPa + (D.50)


 −2.2
V
+450(c + w)1.4(w/c)4.2 ln (1 − henv)0.36 t0−0.3 × 10−9/MPa
10 S
624 D Creep Models Recommended by Design Codes

is the final increase of compliance due to creep. The first term in (D.50) represents
the contribution of basic creep and the second is the additional compliance due to
drying creep. The cement and water contents c and w should be substituted in kg/m3,
volume V in mm3 , surface in contact with outside air S in mm2 , and times t, t ′ and
t0 in days.
To take into account the effect of temperature different from 20◦ C, the actual
ages t, t ′ and t0 should be replaced by temperature-adjusted ages computed in the
same way as according to the fib Model; see (D.17).
For high-strength concrete with compressive strength exceeding 55 MPa, a dif-
ferent formula for the compliance function is recommended:

1 4w(1 − henv) + 350


J(t,t ′ ) = + ln(1 + t − t ′) (D.51)
E(t ′ ) 12 + fc (t ′ )

where fc (t ′ ) is the compressive strength at the age of loading, substituted in MPa. It


is interesting that this compliance function is logarithmic, in contrast to the bounded
compliance function for normal-strength concrete specified in (D.49).
The JSCE code also approves creep predictions based on the B3 Model, ACI
Model or CEB Model.

D.6 Comparison of Compliance Functions

For illustration, the graphs of compliance functions will be plotted for the four mod-
els described in this appendix and for the B3 Model. The purpose of the figures is
merely to show the shape of the compliance graphs and their main features; a sys-
tematic comparison and evaluation is not attempted here. In addition to creep in an
environment of 70% relative humidity, basic creep will be considered as a special
case.
Parameters of individual models are determined or estimated for a concrete with
the same composition and under the same conditions as in Example 3.1. From the
mean strength f¯c = 45.4 MPa, the conventional modulus is estimated according
to the ACI formula (3.6) as E28 = 31.9 GPa, while the GL formula (D.44) gives
E28 = 32.5 GPa. Assuming limestone aggregates (αE = 0.9), the fib formula (3.7)
gives 32.0 GPa. Since these values are quite close, we set E28 = 32 GPa for all the
models.
The values of parameters of Model B3 are taken from Example 3.1: q1 = 18.81,
q2 = 126.9, q3 = 0.7494, q4 = 7.692, q5 = 327.0, all in 10−6 /MPa; τsh = 1121 days
∞ = 701.1 × 10−6.
and εsh
For the CEB Model, we consider s = 0.25 (normally hardening cement of
strength class 42.5), parameter β f = 2.49 follows from (D.10), and parameters
φRH = 1.46 and βH = 376 follow from (D.9) and (D.11) with henv = 0.7 and
2Ac /u = 100 mm. For basic creep, φRH is set to 1, and parameter βH evaluated
D.6 Comparison of Compliance Functions 625
p
from (D.11) would exceed its maximum allowed value, 1500 35/ f¯c = 1317, so it
is set to that value.
For the fib Model, parameters φRH = 0.6463 and βH = 370 follow from (D.20)
and (D.19) with f¯c = 45.4 MPa, henv = 0.7 and 2Ac /u = 100 mm. The temperature-
′ is considered as equal to the actual age, t ′ .
adjusted age tadj

(a) (b)
compliance, J(t,t ′ ) [10−6 /MPa]

compliance, J(t,t ′ ) [10−6 /MPa]


140 140
120 GL2000 120 GL2000
B3 B3
100 fib 100 fib
80 CEB 80 CEB
ACI ACI
60 JSCE 60 JSCE
40 40
20 t ′ = 7 days 20 t ′ = 7 days
0 0
0.01 1 100 10000 0.01 1 100 10000
load duration, t − t ′ [day] load duration, t − t ′ [day]
compliance, J(t,t ′ ) [10−6 /MPa]

compliance, J(t,t ′ ) [10−6 /MPa]


140 140
120 GL2000 120 GL2000
B3 B3
100 fib 100 fib
80 CEB 80 CEB
ACI ACI
60 JSCE 60 JSCE
40 40
20 t ′ = 28 days 20 t ′ = 28 days
0 0
0.01 1 100 10000 0.01 1 100 10000
load duration, t − t ′ [day] load duration, t − t ′ [day]
compliance, J(t,t ′ ) [10−6 /MPa]

compliance, J(t,t ′ ) [10−6 /MPa]

140 140
120 GL2000 120 GL2000
B3 B3
100 fib 100 fib
80 CEB 80 CEB
ACI ACI
60 JSCE 60 JSCE
40 40
20 t ′ = 365 days 20 t ′ = 365 days
0 0
0.01 1 100 10000 0.01 1 100 10000
load duration, t − t ′ [day] load duration, t − t ′ [day]

Fig. D.1 Compliance functions predicted by various creep models: (a) basic creep, (b) creep at
ambient humidity henv = 70%

For the ACI Model, we consider parameters a = 4, b = 6/7 and m = 0.118 (moist-
cured concrete and cement of type I). If we knew the mass density ρ , we could es-
timate the conventional modulus from (D.28). For instance, for ρ = 2400 kg/m3 we
would get 34.1 GPa, which is not that far from the value of 31.9 GPa obtained from
(3.6). Since the exact mass density is not specified, E28 = 32 GPa will be used, same
626 D Creep Models Recommended by Design Codes

as for the other models. It remains to determine parameters γi , i = 1, 2, . . . 6, and their


product, γ . We set γ1 = 1.25 (moist curing), and for henv = 0.7 and V /S = 100 mm
we get γ2 = 0.801 and γ3 = 0.756. Since the slump, fraction of fine aggregate and
air content are not known, parameters γ4 , γ5 and γ6 are set to 1 (default value). The
resulting parameter γ is thus γ = 1.25 × 0.801 × 0.756 = 0.757. For basic creep, we
take γ2 = 0.627 corresponding to henv = 0.96, and the resulting value of γ is 0.593.
It is somewhat disturbing that parameter γ and thus also the compliance function is
affected by the size of the member even in the absence of drying (through γ3 , which
depends on V /S). Note that the effect of the member size is incorrectly introduced
by vertical scaling of the compliance function rather than by its horizontal shift in
the log-scale.
For the GL2000 Model, V /S is set to 100 (mm) and, for henv = 0.7, formula
(D.45) gives ch = 1.17. According to (D.46), parameter Φ depends on the age at
first loading, t1 , and on t0 = 7 (days) and V /S = 100 (mm). In the present example
we consider loading at ages t1 = 7, 28 and 365 days, and the corresponding values
of Φ are 1.0, 0.932 and 0.722. In the case of basic creep, ch is set to zero and Φ
is considered as 1, because drying in fact never starts and thus it does not precede
loading.
For the JSCE Model, the compliance increase due to creep is determined from the
cement content c = 450 kg/m3, water content w = 170 kg/m3 , volume/surface ratio
V /S = 100 mm and onset of drying at t0 = 7 days. Formula (D.50) gives ∆ J(7) =
35.7 × 10−6 /MPa, ∆ J(28) = 24.9 × 10−6 /MPa and ∆ J(365) = 17.0 × 10−6/MPa
for basic creep, and ∆ J(7) = 39.2 × 10−6 /MPa, ∆ J(28) = 28.4 × 10−6 /MPa and
∆ J(365) = 20.5 × 10−6/MPa for creep at drying. The evolution of elastic modulus
is estimated using the ACI formula (D.27).
Fig. D.1 displays the compliance curves for the six models considered here. The
left column corresponds to basic creep and the right one to drying creep, in each
case for three different ages at loading (from top to bottom, t ′ = 7, 28 and 365 days).
One drawback of the code formulae according to CEB, ACI and JSCE, which is best
seen in Fig. D.1b, is that they lead to bounded creep curves. Long-time experiments
indicate that the creep curves approach straight lines in the semi-logarithmic scale,
which means that, for long load durations, the compliance grows as a logarithmic
function and thus is unbounded. This feature is directly incorporated into Models
B3 and B4 and into the new fib Model (see the solid curves in Fig. D.1), and is also
reasonably well captured by the GL2000 Model Code 2010, which uses a bounded
compliance function but the finite limit would be approached for load durations
by several orders of magnitude larger than any durations of practical interest. To
illustrate that, the compliance functions for age at loading 28 days (including the
effect of drying creep) are replotted in Fig. D.2 for hypothetical load durations up
to 109 days.
Note that the slope of the line approached by the new fib Model is substantially
lower than for the B3 Model. This is true in general, because the coefficient multi-
plying the logarithmic term is q4 + 0.1q3 for the B3 Model and 1.8/(E28 f¯c0.7 ) for the
fib Model. In the present example, these coefficients amount to 7.77 × 10−6/MPa
and 3.89 × 10−6/MPa, resp. In general, q4 = 20.3(a/c)−0.7 × 10−6 /MPa, which is
D.6 Comparison of Compliance Functions 627

for aggregate/cement ratios between 2 and 6 roughly in the range between 6 and
12 × 10−6/MPa, while the coefficient for the fib Model is evaluated as

1MPa 0.7 1 1MPa 1.03̄


   
1.8
× = × 185 × 10−6/MPa (D.52)
E28 f¯c αE f¯c

which is for limestone aggregates (αE = 0.9) and for strengths between 20 and 45
MPa roughly in the range between 3 and 8 × 10−6/MPa.

compliance, J(t,t ′ ) [10−6 /MPa]


200
GL2000
B3
150 fib
CEB
100 ACI
JSCE
50
t ′ = 28 days
0
100 102 104 106 108
load duration, t − t′ [day]

Fig. D.2 Compliance functions at age 28 days and ambient humidity henv = 70% predicted by
various creep models and plotted up to extremely long load durations

You might also like