Fracture Mechanics of Concrete and Concrete Structures -

Assessment, Durability, Monitoring and Retrofitting of Concrete Structures- B. H. Oh, et al. (eds)
2010 Korea Concrete Institute, Seoul, ISBN 978-89-5708-181-5

Creep and effective stiffness of early age concrete slabs

H.G. Park & H.J. Hwang
Seoul National University, Seoul, Korea
J.Y. Kim
Kwangwoon University, Seoul, Korea
G.H. Hong
Hoseo University, Chung-Nam Do, Korea
J.H. Im & Y.N. Kim
Housing Eng. Team, Samsung Engineering and Construction, Seoul, Korea

ABSTRACT: Experimental study was performed to investigate the effective stiffness and long-term deflection of
concrete slabs subjected to loading at early age. From the tests of one-way slabs, long-term deflections were meas-
ured. The test variables were the magnitude of loading, load pattern, reinforcement ratio, compression re-bars, and
concrete age. The material properties of early age concrete resulting from cylinder tests were compared with the pre-
dictions by current design code, ACI 209. The long-term deflections of the slab specimens were also compared with
the predictions. The result showed that ACI 209 & 318 overestimated the compressive strength and elastic modulus
of the early age concrete cured at low temperature, and underestimated the long-term deflections of slabs.

1 INTRODUCTION Table 1 presents the test parameters for the speci-

As construction period for multi-story building is
getting shorter, frequently, construction load causes
concrete cracking in early age slabs. The initial dam-
age of the early age concrete accelerates the long-
term deflection and aggravates serviceability prob-
lem due to excessive deflection.
Lee et al. (2008) measured long-term deflections
of slabs loaded at ages of 3, 7, 28 days. The results
showed that the long-term deflection as well as the
initial deflection was significantly increased when
loading was applied to the concrete at early age.
This is because the mechanical properties of the ear-
ly-age slabs were worsened when early loading was
mens. T3B, the control specimen, had the rein-
forcement ratio of 0.52% (D13-@200mm). At age of
3 days after concrete casting, loading of 12kN was
applied, and was maintained for 110 days (Fig. 2(a)).
T3S was the same as T3B except for the magnitude of
load (8kN). In T7B, 12kN was applied at age of 7
days. In T3C, multiple step loading was applied as
shown in Figure 2(b). In the step loading, 4kN was
loaded at 3 day intervals, and only slab weight was
maintained after 12 days. TT3B had doubled rein-
forcement ratio, 1.04% (D13@100mm). TC3B was re-
inforced with compression re-bars (D13@200mm).
150mm
Bearing Plate
(140mm x 800mm)
160mm
P/2 P/2

applied. However, more experimental tests on vari- Section

ous parameters are required to quantify the material Dial Gage G D13

and mechanical properties of early age slabs.

1850mm 400mm 400mm 1850mm

T3B, T3S, T7B, T3C

d= 124mm ~ 129mm
2 TEST SET-UP AND PARAMETERS
800mm
D13@200

Six one-way slabs were tested. Figure 1 shows the TT3B

dimensions and details of the specimens. The net d= 124mm

D13@100
span was 4.5 meters. Two concentrated loads were
TC3B d=20mm
applied at the mid-span. The reinforcement ratio d=129mm
ranged from 0.50% to 1.04%. Concrete mix was de- D13@200

signed for the concrete strength 30MPa at 28 days. Figure 1. Specimens and Test setup.
 D (Test
J = 1.
Table h, T )specimens.
h (1) explicitly
where accounts
=4.0, =0.85for the evolution
for moist curedofnormal
hydration
ce-
Speci Reinforce age at Load Load Effective Steel reactiont isandtheSFconcretes
ment, content. age, Thisandsorption
f c ,28 isisotherm
the 28
TheD13@200
proportionality
-mens -ment
coefficient
loading condition
D(h,T) is called
depth ratio(%) readsstrength of the concrete.
day
moisture
T3B
permeability
T3S D13@200 3day
3day
and it
LC1
is a nonlinear
12kN
LC1 8kN 124 function
124 0.52
0.52
This result indicates that current design code sig-
of the relative
T7B D13@200 7day humidity h and
LC1 temperature
12kN T (Baant
129 0.50 nificantly overestimates the strength of the early age
& Najjar
T3C 1972). The
D13@200 3daymoisture
LC2 mass --- balance
129 requires
0.50 weIn
(h, , ) = G ( , ) 1

concrete cured at low temperature. 1
+
that theD13@100
TT3B variation3dayin time LC1
of the 12kN
water mass
124 per1.04
unit a crecent
s study,
1 c Kim s et al.10(1998)
(g
used
modified
c )h
1 c
volume
TC3B of concrete
D13@200 (waterLC1
3day content w) be129/20
12kN equal to1.00the Arrhenius model to predict
thee time-dependent com-(4)
divergence of the moisture flux J pressive strength affected by curing temperature.
10(g )h
In
1 c c 1
Load the present study, the modified Arrhenius model was
K1 (the )e
c , stime-dependent
8,12kN= J
used to calculate compressive

w
(2)
DL
t
strength of the concrete cylinders.

The water content w 7) can be expressedConcrete

as the Age
sum where the first term E (gel
1 isotherm)
1 represents (2)
the
Age at Loading(3, t = t =
physically bound
exp (adsorbed)
water and
t the second
of the evaporable water
(a) One we (capillary
step loading (LC1) water, water e
R T + 273 T + 273
term (capillary isotherm) represents the capillary
r
vapor, and adsorbed water) and the non-evaporable
Load
water. This
1 + texpression is valid only for low content
2

(chemically bound) water wn (Mills 1966, t =

of SF.
em t
a + tcoefficient G1 represents the amount of
The
e
2
(3)
Pantazopoulo
8kN
& Mills 1995). It is reasonable to water per
e

unit1 + volume held in the gel pores at 100%

4kN
assume that the evaporable water is a function of
2
t
DL
relative
S = S 1humidity, and
t it can be expressed (Norling
e (4)
relative 3humidity,
6 9 12
h, degree of hydration, c, and u

Mjornell
ur
b +
1997) as
t e
2

degree of silica(b)fume reaction, s, i.e. we=we(h,c,s)

Concrete Age
S=
S k tu r em (5)
= age-dependent
2. Loading history. sorption/desorption isotherm
Multiple step loading (LC2)
1 + kr tem
Figure
(Norling Mjonell 1997). Under this assumption and c c+ ks s
G ( c s ) = k vg (5)
by The
substituting Equationare1 categorized
into Equationinto2 two one 1
,
c vg s
slab specimens where T =curing temperature, Tr =reference curing
obtains
groups, according to the date of concrete casting and where kcvg andkJ/mol
temperature(20),
ksvg areatEmaterial parameters. From the
=apparent activation en-
effective slab depth.
we Group1, w TC3B,wthe effective depth maximum
ergy(= 33.5
amount of water per unit volume that can
20, 62.9 kJ/mol at 0),
For h + ( DT7B,h
T3C, e & + e & + w& R =universal
fill all pores gas constant(=8.314 J/mol-K), =an
(both capillary pores and gel pores), one

wash129t mm, while
) =
h forGroup2, (3) factor, te = equivalent age at the ref-
c
c
T3B,s T3S,
s
the effective depth was 124 mm. Group2 was cast 9
n TT3B,
can
age calculate
conversion K 1 as one obtains
erence temperature, tem = modified equivalent age at
days after Group 1 was cast. the reference temperature, and Sur =limiting
where
The we/h isofthe
magnitude loadslope
wasof the sorption/desorption
12kN or 8kN, which was strength. a =0.6855 and b =1.0655 gat

c hcuring
low
10
c


isotherm (also called moisture
2~2.45 times the slabs self-weight, the capacity). The
typical max-
w
temperature. kr =rate s + s G e
c constant s
0.188 0.22
1
at standard tempera-
1

governing equation (Equation 3) must be completed at age t . (6)

0 1

imum construction load for actual slabs. The magni- ture ) = S =compressive strength
K ( c20.
by appropriate
tude of load were boundary
between andtheinitial conditions.
cracking load and the 1 s
,

As shown g the

in Figure 3(a),
h
modified Arrhenius
c c
yield load of the slabs. For T3B at age of 3evaporable
The relation between the amount of
10

model predicted well ethe test results

1
days, the of the concrete
1

water andload
cracking relative
and the humidity
yield loadis called
were adsorption
1.6 kN and cylinders for both Group 1 and Group s2.
isotherm
17.8 if measured with increasing relativity
kN, respectively. The material
Figure parameters
3(b) compares vg and kmodulus
thek elastic
c
vg and of g1 can
the
humidity
Deflectionsdesorption
and were measured isotherm in thecenter
at the slabs opposite
for be calibrated by fitting experimental
concrete cylinders with the prediction by ACI 209 data relevant to
case. Neglecting their difference (Xi
110 days(Fig. 1). Temperatures were also measured.et al. 1994), in free (evaporable)
(2008). water content
In the prediction, the test in concrete
results for theat
the following, sorption isotherm will
After curing of the concrete, the specimens were be used with various ages (Di Luzio & Cusatis 2009b).
compressive strengths was used to evaluate the elas-
reference
tested to ambient
at the both sorption and desorption
temperature as low asconditions.
5. tic modulus of the early age concrete. Nevertheless,
By the way, if the hysteresis of the moisture 2.2 Temperature
current design code evolution
overestimated the test results.
isotherm would be taken into account, two different This result indicates that in the concrete cured at low
3relation,
TEST evaporable
RESULTS water vs relative humidity, must Note that, at early
temperature, age, sinceof the
the decrease chemical
elastic modulus reactions
was
be used according to the sign of the variation of the associated with cement hydration
more pronounced than the decrease of compressive and SF reaction
relativity
3.1 Cylinderhumidity.
test The shape of the sorption are exothermic, the temperature field is not uniform
strength.
isotherm for HPC is influenced by many parameters, for non-adiabatic systems even if the environmental
100 mm 200
especially thosemm that influence
concrete extent were
cylinders and rate of for
cured the temperature is constant. Heat conduction can be
3chemical
days. After reactions and, in turn, determine water
3 days, they were cured in a pore described
3.2 in concrete, at least for temperature not
Slab test
structure
tank. and3(a)
Figure poreshows
size distribution (water-to-cement
the variations of the com- exceeding 100C
Figure 4 shows (Baant &deflections
the long-term Kaplan of1996), by
the slab
ratio, cement chemical composition, SF content,
pressive strength of the early age concrete. The re- Fouriers law,
specimens. In which
(Figs. reads
4(a)-(e)), the effects of design
curing time and method, temperature, mix additives,
sults showed that the strengths of the concrete at
parameters (the magnitude of load, concrete age at
etc.). In the literature various formulations can be
ages less than 10 days were as low as 30% of the q = loading,
initial T (7)
loading type, reinforcement ratio, and
found to describe
predictions by currentthedesign
sorption
codeisotherm
ACI 209. of normal compression re-bars) are shown by comparing the
concrete
f (t ) = (Xi etf al. 1994). However, in the present
t (1) where
test is the
q of
results thecontrol
heat specimen
flux, T T3Bis the absolute
or T3S with
paper the+ semi-empirical expression proposed by
c c ,28
t
Norling Mjornell (1997) is adopted because it temperature,
other and is the heat conductivity; in this
specimens.

Proceedings of FraMCoS-7, May 23-28, 2010

 40 I cr J = D ( h , T ) h (6)
Ie = 2
Ig
35 M cr I cr
1 1 The
I g proportionality
coefficient D(h,T)
30
Ma
S treng th (MPa)

25 moisture permeability and it is a nonlinea

20
ACI 209
of the relative humidity h and temperature
where I g is the moment of inertia of the gross sec-
15
Group 1
& Najjar 1972). The moisture mass balanc
tion, and I cr is the moment of inertia of the cracked
10
Group 2
that the variation in time of the water mas
section. M cr is the cracking moment of the slab de-
5
Group 1 eq.
volume of concrete (water content w) be eq
fined with f c ( t ) (concrete strength at age t).
0
Group 2 eq.
divergence of the moisture flux J
Using the effective moment of inertia, the curvature
0 10 20 30 40 50 60
distribution along the slab length was calculated
d ay
w = J Then, by integrating
from the momentdistribution.
(a) Time-dependent compressive strength t
the curvature distribution, the inelastic short-term
30000
deflection of the slab specimens calculated.
25000 Thedeflections
In this test, only the water content w can be expressed a
for superimposed
E las tic Mo dulus (MP a)

of the evaporable water

loading were measured. Therefore, the deflection we (capillary wa
caused by the selfvapor,
weightand
of adsorbed water) and the non-e
20000
the slab was calculated,
15000 (chemically bound) water
and was added to the measured short-term deflection wn (Mil
Group 1
to estimate the total short-term deflection of the is reas
Pantazopoulo & Mills 1995). It
10000
Group 2 slabs. assume that the evaporable water is a fu
5000
The long-termrelative humidity,
deflections , degree of
of the hspecimens were hydration
0
ACI 209
degree of silica fume reaction,
calculated by multiplying the short-term deflection s, i.e. we=w
0 5 10 15 20 25 30 35 40 = age-dependent sorption/desorption
with the creep coefficients specified by ACI-318
S tre ng th (MP a)
(2008). In ACI-318,(Norling Mjonell
the creep 1997).
coefficient Under this
is defined as assum
(b) Relationship between strength and elastic modulus by substituting Equation 1 into Equati
Figure 3. Test results for compressive strength and elastic mod-
obtains
ulus of concrete cylinders. = (6)
1 + 50 '
w h
As shown in (Figs. 4(a)-(e)), the long-term de- e + ( D h) = we w
&c + e &s + w
flections pattern was similar regardless of the de- The long-term deflection h as
h tis calculated
sign parameters. However, all the test parameters
c s
were able to reduce long-term deflections. The T = ( t ,t0 ) D
TC3C showed the least result in the long-term de-
where we/h is the slope of the(7) sorption/
flection as well as in the short-term deflection.
isotherm (also called moisture capac
Figure 4(c) shows the effect of loading type. In
governingdeflection.
where D is the short-term equation (Equation 3) must be
this figure, T3C was compared with T3S that had
by appropriate
Figure 4 compares the predictedboundary
long-termanddeflec-
initial conditi
the same maximum load 8kN. Though the super-
The relation between the amount of e
tions and the corresponding test results.
As shown in thewater andACI-318
figures, relative underestimated
humidity is called
imposed load was completely removed, deflection
recovery was not significant due to concrete crack
isotherm
the test results though if measured
the material with com-
test results, increasing
pressive strength humidity
and elastic andmodulus
desorption
were isotherm
used for in th
damage developed by the initial loading. For this this estimation. case. Neglecting their difference (Xi et al.
reason, T3C showed considerable long-term de-
flection.
the following, sorption isotherm will be
reference to both sorption and desorption c
5 CONCLUSIONS By the way, if the hysteresis of the
4 EVALUATION OF SLAB DEFLECTION
isotherm would be taken into account, two
The primary findingrelation,
of evaporable
the presentwater studyvscanrelative
be humi
The long-term deflections of the slab specimens
be used according to the sign of the varia
summarized as follows.
were predicted by using current design code. As
1) The compressive relativity
strengthhumidity.
of early age The shapeat of the
concrete
shown in Figure 3, since the compressive strength
isotherm for HPC is influenced byofmany p
ambient temperature 5 was as low as 30%
the predictionsespecially
by currentthose thatcode.
design influence extent and
The modi-
and elastic modulus of early age concrete are not ac- fied Arrheniuschemical
curately predicted by current design code, the results
reactions and,
model considering in turn,
the effect of determ
of cylinder tests were used in the predictions.
structure
curing temperature welland pore sizethe
predicted distribution
compres- (water-
sive strength. ratio, cement chemical composition, SF
In the present study, Bischoffs equation (Bischoff
& Scanlon 2007) was used to calculate the effective
2) Current design curing
codestime and method, the
overestimated temperature,
elastic mix
modulus of theetc.).
earlyInage
the concrete
literaturebyvarious
20% even formulatio
moment of inertia( I e ) of the slabs. In the Bischoffs
equation, the effective moment of inertia( I e ) is de-
found to
though the concrete describeresulting
strengths the sorption
from theisotherm
test were used concrete (Xi et al. 1994). However, in th
in the estimation.
fined as 3) Although the paper the semi-empirical
compressive strength andexpression
elastic pro
Norling
modulus resulting fromMjornell (1997)
the cylinder testsis were
adopted b

Proceedings of FraMCoS-7, May 23-28, 2010

J D (h,the
=used, T )hACI 318 creep model still underesti-
(1) explicitly
80 accounts for the evolution of hydration
mated the long-term deflections of the slabs reaction
70
and SF content. This sorption isotherm
The proportionality
which were subjectedcoefficient
to loading atD(h,T) is called
early age. reads
60

deflections (mm)
moisture permeability and it is a nonlinear function 50
of the
80 relative humidity h and temperature T (Baant
40

& Najjar
70 1972). The moisture mass balance requires we h c s ) =( , G1 ( c , s )1 1
T3B +
that60the variation in time of the water mass per unit 30

,

(g )h
TC3B
deflections (mm)

volume
50 of concrete (water content w) be equal to the 20
e
10
c ACI 318 (T3B) (4)
1 c
divergence
40 of the moisture flux J 10
(g
ACI)h318 (TC3B)

30
T3B 0
0 10 20 K30( c40 s50
,
10
1c
) e 60 70 80 90
c 1
100 110 120
20 = J
T3S age (days)
Slab
(2)
1
w

10 t
ACI 318 (T3B)
(e) Effect of compression re-bars
ACI 318 (T3S)

The0 water
0 where4.the
Figure first term
Long-term (gel isotherm)
deflections represents
of slab specimens; effectsthe
of
10 20content can60be 70expressed
30 40 w 50 as the
110 sum
of the evaporable water Slabw
age
80 90 100
(capillary
(days) water,
120
water physically bound (adsorbed) water and the second
design parameters.

vapor, and(a)adsorbed water)

e
and the non-evaporable term (capillary isotherm) represents the capillary
(chemically
Effect of
bound)
magnitude
water w
of load
(Mills 1966, water. This expression is valid only for low content
ACKNOWLEDGEMENT
Pantazopoulo
80
n
& Mills 1995). It is reasonable to of SF. The coefficient G1 represents the amount of
assume
70 that the evaporable water is a function of water per unit volume held in the gel pores at 100%
This research was financially supported by the
relative humidity, h, degree of hydration, c, and
60 relative humidity, and it can be expressed (Norling
Housing Eng. Team of Samsung Engineering &
Mjornell 1997) as
deflections (mm)

degree
50 of silica fume reaction, s, i.e. we=we(h,c,s) Construction, and the authors are grateful to the au-
= 40age-dependent sorption/desorption isotherm thorities for their support.
(Norling Mjonell 1997). Under this assumption
T3B and c c+ ks s
G ( c s ) = k vg (5)
30
by 20substituting Equation 1 into Equation 2 one
T7B 1 c vg s
,

obtains ACI 318 (T3B) REFERENCES

10
ACI 318 (T7B) where k vg and ksvg are material parameters. From the
c
0
w 0 h10 20 30 40 50w 60 70 w80 90 100 110 120 maximum
ACI 209R-92.amount of waterofper
2008. Prediction unitShrinkage,
Creep, volume and
thatTem-
can
e + ( D h) = Slabeage
(days) e (3)&+ & + w& fillperature
all poresEffects
(bothincapillary
Concrete pores and gel
Structures. ACIpores), one
Committee
h t h c
s n
can209.
calculate as one obtains
K1Concrete
(b) Effect of concretec age at initial
s loading American Institute.
ACI 318-08. 2008. Building Code Requirements for Structural
80 T3S T3C Concrete. ACI Committee 318. American Concrete Insti-
where
70
we/h is the slope of the sorption/desorption
ACI 318 (T3S) ACI 318 (T3C) tute.

10 g

1 c

h
c

isotherm (also called moisture capacity). The Bischoff, P.H.w0 & .188 s + 0A.
s s G1Effective
c .222007.
0Scanlon, 1 e

Moment of Iner-
60
governing equation (Equation 3) must be completed tia for Calculating Deflections of Concrete Members Con- (6)
deflections (mm)

K1(taining
, ) =Steel Reinforcement and Fiber-Reinforced Polymer
by 50appropriate boundary and initial conditions. c s 10 g

h 104(8): 68-75.

The relation between the amount of evaporable 1 c c

40 Reinforcement. ACI Structural Journal.
e
Kim, J.K. & Moon, Y.H. & Eo, S.H. 11998. Compressive
water
30 and relative humidity is called adsorption
strength development of concrete with different curing time
isotherm
20 if measured with increasing relativity Thetemperature.
and material parameters
Cem Concr Res. and ks1761-1773.
kcvg28(12): vg and g1 can
humidity
10 and desorption isotherm in the opposite be calibrated
Lee, by fitting
J.I. & Scanlon, experimental
A. & Scanlon, M.A. 2008. dataEffect
relevant to
of Early
case.0 Neglecting their difference (Xi et al. 1994), in freerestraint
Age (evaporable)
Loading on water content in concrete at
time-Dependent deflection and Shrinkage
the following,
0 10 20 sorption
30 40 50isotherm
60 70 80
Slab age (days)
will90be100used
110 with
120 various ages (Di Luzio
tios. ACI SP. 246: 149-165. & Cusatis 2009b).
Cracking of Slabs with Low Reinforcement Ra-
reference to both sorption and desorption conditions.
By the way,(c)ifEffect the ofhysteresis of the moisture
loading type
2.2 Temperature evolution
isotherm
80 would be taken into account, two different
relation,
70 evaporable water vs relative humidity, must Note that, at early age, since the chemical reactions
be 60used according to the sign of the variation of the associated with cement hydration and SF reaction
deflections (mm)

relativity
50
humidity. The shape of the sorption are exothermic, the temperature field is not uniform
isotherm
40
for HPC is influenced by many parameters, for non-adiabatic systems even if the environmental
especially those that influence extent T3B and rate of the temperature is constant. Heat conduction can be
chemical
30
reactions and, in turn, determine
TT3B
pore described in concrete, at least for temperature not
structure
20
and pore size distribution (water-to-cement
ACI 318 (T3B) exceeding 100C (Baant & Kaplan 1996), by
ratio,
10 cement chemical composition, SF content,
ACI 318 (TT3B) Fouriers law, which reads
curing0 time and method, temperature, mix additives,
etc.). 0In 10the20literature various formulations can be
30 40 50 60 70 80 90 100 110 120
Slab age (days) q = T (7)
found to describe the sorption isotherm of normal
concrete (Xi et al. 1994).
(d) Effect However, ratio
of reinforcement in the present where q is the heat flux, T is the absolute
paper the semi-empirical expression proposed by temperature, and is the heat conductivity; in this
Norling Mjornell (1997) is adopted because it

Proceedings of FraMCoS-7, May 23-28, 2010