J. King Saud Univ., Vol. 3, Eng. Sei. (1), pp.

69-87 (1411H/1991)

Prediction of Drying Shrinkage of Portland Cement Paste:

Influence of Shrinkage Mechanisms

Jamal A. Almudaiheem
Assistant Professor, Civil Engineering Department, College of Engineering,
King Saud University, P.D. Box 800, Riyadh 11421, Saudi Arabia

Abstract. Ultimate drying shrinkage of paste was investigated in terms of the Gibbs-Bangham shrinkage
and the capillary stress mechanisms. Shrinkage measurements and nitrogen sorption isotherms were
obtained on cement pastes of water-to-cement (W/C) ratios of 0.4, 0.6 and 0.75. The contribution to ulti-
mate drying shrinkage from these two mechanisms was obtained from ultimate shrinkage versus increase
in surface free energy curves in the relative humidity (RH) range of 95 to 0 percent. The Gibbs-Bangham
stress mechanism is the major contributor to shrinkage stresses in this RH range. The capillary stress
mechanism is active in the RH range of 95 to 50 percent.
The maximum Gibbs-Bangham stress shrinkage, which occurs at 0 percent RH, is linearly proportional
to total surface area as obtained from volume-thickness (V-t) sorption analysis. Maximum capillary stress
shrinkage occurs at about 50 percent RH. This component was found to be linearly proportional to
cumulative capillary surface area. V-t adsorption analysis was used to estimate the cumulative capillary
surface'area. The equations for predicting maximum Gibbs-Bangham and capillary stress shrinkage were
extended to include effect of RH of drying. They were tested on different shrinkage results of well-hyd-
rated pastes of 0.4 and 0.6 W/C ratios and found to be accurate.

Introduction

Drying shrinkage of concrete is a property of the microstructure of the hydrated

cement paste and the strong interaction between the hydrophilic surface and water
[1,2]. Drying process involves removal of water which in turn creates substantial
stresses. These stresses become significant in high surface area systems such as
cement paste.
Four shrinkage mechanisms have been proposed to explain the drying shrinkage
phenomena of porous materials. They are: change in surface free energy (Gibbs-
Bangham effect) of the solid phase due to sorption of water vapor [3-9], capillary ten-
sion effect due to meniscus formation in the capillary pores which results in equal

69
© 199 (A.H. 141). Production and hosting by Elsevier B.V. on behalf of King Saud University. This is an open access article under
the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
70 Jamal A.AI-Mudaiheem

hydrostatic compression in the solid phase [3,4,9-12], changes in the disjoining pres-
sure in areas of hindered adsorption [3,4,8,13,14]; and movement of interlayer water
[5-8,10,12,14]. The contribution to overall shrinkage from the individual shrinkage
mechanism is unclear. It is well known that opinions differ widely about the role of
the four mechanisms in causing the shrinkage. There seems to be an agreement
between different researchers that most of the shrinkage occurring upon drying at
RH higher than 40 percent is caused by capillary tension mechanism [3-6,8,15].
Whereas, shrinkage is mainly caused by changes in surface energy of the gel particle
upon drying at RH lower than 40 percent [3,4,8,15].
In the present study the feasibility of predicting drying shrinkage of paste in
terms of the predominant two mechanisms (namely the Gibbs-Bangham and Capil-
lary tension) will be investigated. A wide range in pore structure variations will be
obtained by varying the W/C ratio and curing time.

Materials and Methods

A commercial Type I portland cement was used. All mixing and curing opera-
tions were performed at a room temperature of 23 ± 2°C (74 ± 3°F). The water to
cement (W/C) ratios used were 0.4,0.6 andO.75. The pastes were hand mixed for two
minutes then rotated in sealed containers at 10 rpm before casting to minimize
sedimentation and bleeding. The pastes were cast in plexiglass molds after they
reached suitable consistency and rotated for 24 hours. The slabs were demolded and
cured until time of test in air-tight containers filled with lime-saturated water. At the
end of the curing period, thin specimens measuring 2.5 x 10 x 75 mm (0.1 x 0.4 x
3 in.) were cut from the slabs using a diamond saw.

A stainless steel clamp was mechanically attached to each specimen end to facili-
tate shrinkage measurements. The clamp has a spherical slot at its center for accurate
sitting on the steel spheres attached to each measuring end in the dial gauge. The dial
gauge was constructed by Dr. L.J. Parrott (from Cement and Concrete Association,
London) to Dr. W. Hansen [16]. The reproductibility of the dial gauge reading was
within ± 5 microstrain. Drifts in readings were compensated for by using an invar
stainless steel bar. The specimens were dried in desiccators conditioned at 90, 75, 60,
50, 30, 11 and 0 percent RH. The RH levels were obtained using aqueous solution of
sulfuric acid. The zero percent RH corresponds to the vapor pressure above concen-
trated sulfuric acid. The RH was maintained constant in each desiccator by stirring
the solution continuously during the entire drying period. The concentration of the
sulfuric acid solution in each desiccator was monitored with time to ensure that the
RH is constant. The desiccators were kept in an environmental chamber where the
temperature was maintained at 23 ± IOC (74 ± IOF).
Specimens used for pore structure measurements were solvent replaced with
methanol for at least 3 weeks. The methanol was changed several times during this
 Prediction of Drying Shrinkage of Portland Cement Paste 71

period. Solvent replacement is a widely accepted and recommended technique for

pore structure measurements using Nitrogen Sorption Isotherm or Merury Intrusion
Porosimetry. The solvent replacement technique was used in order to protect the
original pore structure from extensive pore collapse on drying [17]. This technique
was used with the entire specimens. Therefore, if there were any effect from the
interaction between the methanol with the C-S-H surfaces, as claimed by some
researchers [18], the effect would be relative and the findings will hold true. This is
supported by Parrott [19]. About 50 to 100 mg of each specimen was crushed into
pieces of 2.5 mm thickness and dried at 0 percent RH for at least one week before
pore structure measurement using nitrogen sorption.

The dynamic flow method was used to obtain the nitrogen sorption isotherms*.
The carrier gas was helium. Desorption and adsorption curves were obtained in the
relative vapor pressure (P/Po) region of about 0.98 to about 0.05. For V-t analysis the
reference t curve by Cranston and Inkley [20] was used.

Shrinkage and pore structure results of different pastes of 0.4 and 0.6 W/C ratios
cured for 165 days were used as well as in the present study to validate the proposed
approach.

Theoretical Background

Gibbs-Bangham Shrinkage Stress

The changes in surface free energy with relative humidity are described by the
Gibbs free energy equation:
10-7 x RT fP/PO V a P
~'Y = - - dln- (1)
V P/Po Ss Po
where ~'Y is the calculated increase in surface free energy in N/m, R is the gas
constant (dyne.cm/mol.K), T is the absolute temperature in degrees Kelvin, V is the
molar volume of adsorbed liquid (cm 3/mole), V a is the volume of adsorbed liquid
(cm 3/g dry), Ssis the specific surface area (m 2/g. dry), and P/Po is the relative vapor
pressure.
Bangham and his co-workers [21,22] proposed the following empirical equation
which relates the expansion (E exp ) of highly porous material (charcoal) and the
changes in surface free energy (~'Y):

(2)

where Ais a material constant. Amberg and Maclntoch [23] confirmed the linear
relationship between the expansion of Vycor glass and the decrease in surface free
energy upon adsorption of water.

* Quantasorb, Quantachrome Co., six aerial way, Svosset, New York

72 lamal A.Al-Mudaiheem

Bangham and Maggs [24] have related Ato the elastic modulus of a high surface
area solid (Es) by the following expression:
Ss Qs
A = -- (3)
Es
where Qs is the solid density. Different expressions have been proposed by Hiller
[25] and Yates [26]. They differ only from the expression by Bangham and Maggs by
a proportionality factor. Thus, in general:

(4)

where K is constant.

Capillary Shrinkage Stresses

Menisci are formed in capillary pores upon drying. The menisci create hydrosta-
tic tension in the remaining water which is transferred to the solid pore walls as
hydrostatic' compressive stresses (Pca ). These stresses increase with decreasing pore
size as predicted by the Laplace equ:tion:
2 1'<0
Peap = - r - (5)

where r is the radius of curvature of the meniscus assuming a hemispherical

surface, and 1'<0 is the water density.
The relationship between the radius of curvature and vapor pressure (P/Po) can
be obtained from the Kelvin equation where:
P 2 1'<0 V
In - (6)
Po R T r
combining equations ~ and 6 it can be shown that
P
Pcap = K t In-
Po (7)

where Kt is a material constant.

No fundamental relationship has been developed between capillary stresses and

shrinkage (Eeap ) of Portland cement paste. This may be attributed to the complixity
of the pore system. In general, one would expect that
Eeap = f (Ps' Pg, E s) Peap (8)
where f (Ps ' Pg, E s) is a function of the pore structure, pore geometry, and elastic
modulus of the solid. From equations 7 and 8 it follows that:
 Prediction of Drying Shrinkage of Portland Cement Paste 73

p
ecap = K l f (Ps' Pg, EJ In Po (9)

Thus, a relationship between In: and capillary stress shrinkage may be empir-
ically obtained. 0

Results and Discussion

Ultimate Drying Shrinkage of Paste Versus Relative Humidity of Drying, as Influ-

enced by WIC Ratio and Curing Time

Figure 1 illustrates the development of drying shrinkage over drying time at 90,
75,60, 50,30, 11 and 0 percent RH for the 0.6 W/C ratio paste cured 30 days. Each
point on the shrinkage curves is the average of two companion specimens. The vari-
ations in shrinkage between companion specimens were within 0.02 percent. The
curves are similar to the other paste systems investigated. The shape of the shrinkage
curves is in general agreement with those reported by other investigators [27-29].

Relative humidity of drying NG

1 .2 o 0 .,.
e . 11 .,.
X 30 .,.
Q 50·'.
1 .0 + 60 .,.
A 75·'.
* 90·'.
~ 0.8
~

=
~

=
'C
i
~
0.6
=
.~
&.
Q
0.4

0.2

o. oL-..._ _f!lS::!:i::~;:c:.......-..~.u.&J..---'-""".a..u.L.l~...I-...I.""""'~
1
10
Drying time, min.

Fig. 1. Development of drying shrinkage over time curves for thin paste specimens of 0.6 Wle ratio hyd-
rated 30 days.
74 Jamal A.AI-Mudaiheem

The results illustrate that ultimate drying shrinkage is not reached within the
drying period of 275 days. However, these values were assumed to be ultimate since
the additional shrinkage extrapolated out to several years is insignificant. Ultimate
drying shrinkage values for the paste systems investigated are shown in Table 1.

Table 1. Ultimate drying shrinkage (%) versus RH of drying for the paste systems investigated

Length of
W/C Percent of relative humidity
curing
ratio (days) 90 75 60 50 30 11 0

0.4 3 0.07 0.17 0.24 0.28 0.34 0.40 0.79

30 0.06 0.16 0.25 0.34 0.41 0.49 0.91
240 0.04 0.13 0.25 0.32 0.40 0.46 1.00

0.6 7 0.07 0.17 0.22 0.26 0.33 0.39 0.80

30 0.10 0.20 0.30 0.40 0.48 0.52 1.04
240 0.07 0.18 0.29 0.41 0.47 0.55 1.10

0.75 9 0.05 0.17 0.25 0.30 0.38 0.51 0.89

30 0.08 0.22 0.30 0.39 0.47 0.54 1.08
240 0.05 0.20 0.32 0.47 0.56 0.61 1.04

Contribution to Ultimate Shrinkage from the Two Stress Active Shrinkage

Mechanisms
Two stress active shrinkage mechanisms may be interpreted from the curves of
ultimate drying shrinkage versus calculated increase in surface free energy shown in
Fig. 2. The major shrinkage contributor illustrated by the dotted lines is due to the
Gibbs- Bangham stress mechanism since it follows equation (2). This mechanism is
active over the entire RH range between 95 and 0 percent. The systems shown are
pastes of 0.4, 0.6, and 0.75 WIC ratio cured for 30 days. The curves are typical for the
systems investigated, and are in good agreement with those reported by Amberg and
MacIntoch [23] on porous Vycor glass.
The sorption isotherms and the V-t curves are shown in Figs 3a, 3b, 3c and Figs.•
3d, 3e, 3f, respectively. The V-t curves were constructed from nitrogen sorption
isotherms and the reference t curve of Cranston and Inkley [20]. The increase in sur-
face free energy versus RH was calculated from equation (1). The relationship
between volume of liquid adsorbed normalized for surface area (Va/S) versus In PIP0
was estimated from V-t adsorption curves corrected for capillary condensation vol-
ume. It was found that, the relationship between (Va/S) versus In PIP0 is similar for
the paste systems investigated. Therefore, the calculated increase in surface free
energy versus RH is similar for these systems. The results are shown in Table 2.
 Prediction of Drying Shrinkage of Portland Cement Paste 75

Ninety-five percent RH was the starting point for estimating the increase in surface
free energy with decrease in RH due to the uncertainty in obtaining (Va/S) at t values
greater than 1.6 nm. Further, it is assumed that zero drying shrinkage has occurred
between 100 and 95 percent RH. A zero percent RH in the calculations corresponds
to a value of -7.5 in In PIPo.
1. 4 ---....---...--......,r------.,r---~--_,--__'T--....,
o 0.40 W/C ratio paste

1. 2 * 0.60 W IC ratio paste

o 0.75 W/C ratio paste

1.0

0.2

o. 0 rd!::::.:.:..::.:.:..l._ _..J-_ _..L-.-_......J._ _ _ _

~ ~ _ - . ._ _- - J

o 50 100 150 200 250 300 350 400

Cal. increase in surface free energy, N/m x 10.3

Fig. 2. Drying shrinkage versus increase in surface free energy curves obtained for thin paste specimens
of 0.4, 0.6 and 0.75 W IC ratio hydrated 30 days.

Table 2. Calculated increase in surface free energy for incremental RH steps

RH D.y
3
0/0 (X 10· N/m)

90-95 10.1 10.1

75-90 26.4 36.5
60-75 24.8 61.3
50-60 17.7 90.0
30-50 43.2 122.2
11-30 66.8 189.0
0-11 152.6 341.6
76 lamal A.AI-Mudaiheem

0·24--......---.--....----.--.......
0.4 WIC ratio paste (a) 0.4 W IC ratio paste
0-20 0_ Adsorption isotherm (d)
6 _ Desorption isotherm
0-16

0-12

0.6 W IC ratio paste 0.6 W/C ratio paste

... 0·20
~

'C
(b) (e)
Cl)
;;)-
e
CJ
0·16
M
z
g 0·12
'C
~
~
...c o.o8
fIJ
~
e 0·04
e=
>
O.p
0·24
0.75 W/C ratio paste \.0.75 W/C ratio paste
0·20 (f )

0·16

0·12

0·08

o·O~_--I _ _. J . . - - - " - - _....._ ..

0.0 0.2 0.4 0.6 0.8 1.0 o 4 8 12 16 20
Relative pressure, PIP0 Thickness. t, nm x 101

Fig. 3. Nitrogen isotherms and V-t curves of 0.4, 0.6 and 0.75 WIC ratio pastes hydrated for 30 days. Sol-
vent replaced in methanol for 3 weeks before drying at 0% RH.
 Prediction of Drying Shrinkage of Portland Cement Paste 77

In Fig. 2 capillary stress shrinkage is the component superimposed onto the

Gibbs-Bangham stress shrinkage in the RH range between 95 and about 50 percent.
This is supported by nitrogen V-t curves in Figs 3d, 3e, and 3f which show capillary
condensation in the t value range of about 0.68 to 1.1 nm. This corresponds to an RH
range of about 50 to 85 percent RH. The capillary stress shrinkage at about 50 per-
cent RH is about 50 to 60 percent of the Gibbs-Bangham stress shrinkage. Further,
the capillary condensation volume below about 50 percent RH is insignificant. Con-
sequently, capillary stress shrinkage is insignificant below 50 percent RH.
The results in Fig. 2 suggest that capillary stress shrinkage is entirely irreversible
since it increases gradually from 95 percent to its maximum value near 50 percent RH
for each of the three systems investigated. The capillary stresses, however, are not
cumulative. The irreversible nature of capillary stress shrinkages agrees well with
results by other investigators [3,10,30,31].
Figs 4a, 4b, and 4c show capillary stress shrinkage versus RH of drying curves of
the 0.4, 0.6 and 0.75 W/C ratio pastes and the influence curing time. These curves
were constructed from ultimate shrinkage versus increase in surface free energy
curves such as those shown in Fig. 2. For the 0.6, and 0.75 W/C ratio pastes capillary
stress shrinkage at 50 percent RH increases with curing time. It is unaffected by cur-
ing time in the 0.4 W/C ratio paste. Figs 4a, 4b, and 4c show that the capillary stress
shrinkage increases with increasing W/C ratio.
Shrinkage stresses below 50 percent RH may be entirely due to Gibbs-Bangham
stress. However, the additional shrinkage between 50 and 11 percent RH is only a
fraction of the Gibbs-Bangham stress shrinkage predict.ed by the dotted lines in Fig. 2.
This transition range occurs due to a drop in the total stresses caused by the disap-
pearance of capillary stresses. Below 11 percent RH the shrinkage versus increase in
surface free energy relationship will again follow that described by the dotted line.
The occurrence of the transition range suggests that, a major part of first drying
shrinkage is inelastic decrease of interlayer spaces and that true Gibbs-Bangham
shrinkage is only a fraction of total shrinkage. Further, shrinkage in this range is
nearly linear with increase in surface free energy. Table 3 illustrates that the addi-
tional shrinkage between 50 and 11 percent RH is approximately 52 percent of the
original Gibbs-Bangham stress shrinkage in this RH range as predicted by the dotted
line.
78 Jamal A.AI-Mudaiheem

0.30
(a) 0.4 W/C ratio paste
0·25 a CURED 3 DAYS
x CURED 30 DAVS
A CURE 0 240 DAYS
0.20

0·15

0·10

0·05

0·00
0·30
(b) 0.6 WIC ratio paste
0·25 -- D CURED 7 DAYS

e~
oS
~
0.20
eo
~
=
·C 0.15
..c
Cl}

.....
Cl}
Cl}
~

0·10
..
Cl}

~
·aeo 0.05
U
0·00
0·30
(c) 0.75 W/C ratio paste
0·25 -- D CURED 9 DAYS

0.20

0·15

0.10

0·05

0·00
0 20 40 60 80 100
Relative humidity, %

Fig. 4. Capillary stress shrinkage versus relative humidity curves for 0.4, 0.6 and 0.75 WIC ratio Port-
land cement pastes.
 Prediction of Drying Shrinkage of Portland Cement Paste 79

Table 3. Shrinkage between SO and 11 percent RH as fraction of Gibbs-Bangham stress shrinkage

W/C Cured, days Shrinkage ratio (percent)

0.4 3 48
30 48
240 51
0.6 7 64
30 50
240 40
0.75 9 72
30 51
240 40

Effect of Nitrogen Surface Area and RH of Drying on Gibbs-Bangham and Capillary

Stress Shrinkage

Maximum Gibbs-Bangham stress shrinkage versus total surface area and capil-
lary stress shrinkage versus capillary surface area are shown in Figs 5 and 6 respec-
tively for the pastes investigated. Shrinkage values were obtained from ultimate
shrinkage versus increase in surface free energy curves whereas surface areas were
estimated from nitrogen V-t curves. The total surface areas were obtained from the
slopes of the initial straight line portions of the V-t curves. The capillary surface areas
were calculated from the V-t adsorption curves. Assuming cylindrical pore shape,
the pore diameter range in which condensation occurred upon adsorption was about
4 to 20 nm. The capillary surface areas for pores larger than about 20 nm was esti-
mated from the slope of the V-t adsorption curve at a t-value of 2 nm. This t-value
was used in order to avoid effects of micropore filling. Incorporating the capillary
surface area of pores larger than 20 nm is justified since a substantial amount of capil-
lary stress shrinkage occurs between 90 and 95% RH as seen in Figs 4a, 4b, 4c. This
corresponds to capillary pore diameter between about 20 and 40 nm respectively.
Table 4 illustrates that micropores are present in all the paste investigated since the
total surface area values are substantially larger than those calculated for the capil-
laries. Further, the micropores are larger in width than 1.6 nm since no downward
deviation in the V-t curves occurs at t-values smaller than 0.8 nm. Linear regression
of the results in Figs 5 and 6 yields the following two equations:

EOB (o%RH) = 0.655 + 0.00524S tot (10)

and,
Ecap (50 % ) = 0.01524 + 0.00251S cap (11)
80 Jamal A.AI-Mudaiheem

Table 4. Nitrogen surface areas of the pastes studies

W/Cratio Curing length (days) Stot m 2/g dry Seap m 2/g dry

0.40 7 30 17
30 34 17
240 45 22

0.60 7 49 34
30 60 39
240 84 54

0.75 9 35 26
30 72 49
24Q 96 62

2.•:0 -, ......- - -..... -----r----,..---....,

%
0 RH

1.6

1.2

0.0'----..1.------'---.......- - - " - - - -....

o 20 40 60 80 100

Total surface area, m2/g dry

Fig. 5. Gibbs-Bangham stress shrinkage atO % RH versus total surface area

 Prediction of Drying Shrinkage of Portland Cement Paste 81

o .20 - - . . . - -.....- ....- ......~-.....- .......--...----.

50 % RH

o .16

o .12

o. oO.~ ......_ ......__.-..-._......._ ....._.-..

o 10 20 30 40 50 60 70 80
Capillary surface area, m2/g dry

Fig. 6. Capillary stress shrinkage at SO % RH versus capDlary surface area

Where EGB(O%RH) is the Gibbs-Bangham stress shrinkage at zero percent RH,

and Ecap (500/0 RH) is the capillary stress shrinkage at 50 percent RH. Stot and Scap are
the total and capillary surface areas in m2/g dry respectively. The surface areas were
obtained from nitrogen sorption analysis. The correlation coefficients for equations
(10) and (11) are 0.742 and 0.901 respectively.
The linear relationship between Gibbs-Bangham stress shrinkage and total sur-
face area is in good agreement with theory (equation 4). The positive intercept with
shrinkage at zero surface area is another indication that a major fraction of first dry-
ing shrinkage is due to a decrease in interlayer spacing. This is also consistent with the
results in Fig. 6.
Equations (10) and (11) can be modified to include the effect of RH. Based on
the results shown in Table 2 for the increase in surface free energy versus decrease
in RH a linear relationship is obtained in the RH range of 95 to 20 percent. It is
described by the following equation:
6"1 = 174.5 - 1.85RH
where (12)
200/0 ~ RH ~ 950/0
82 lamal A.AI-Mudaiheem

A correlation coefficient of 0.996 was obtained from the regression line in this
RH range. The RH range below 20 percent has little practical importance, therefore,
it is excluded in this study. The follwing equation is obtained by combining equations
10 and 12.
174.5 -1.85RH
EGB(RH) = (0.264 + 0.00211Stot) ( 137.5) (13)

This equation is used to calculate the shrinkage component due to Gibbs-Ban-

gham stress in the RH range of 95 to 50 percent. The shrinkage component due to
capillary stresses in the RH range of 95 to 50 percent can be conveniently obtained
from the following empirical relationship:

In RH
100
Ecap (RH) = (0.015 + 0.0025 Scap) ( lm 0.5 ) (14)

where

In the RH range of 95 to 50 percent total drying shrinkage, Et is obtained by

superimposing the two shrinkage components. Thus:

(15)

Between 50 and 11 percent RH the additional shrinkage, which is Gibbs-Ban-

gham stress induced, is obtained from the following equation:

(16)

where Etr is the additional shrinkage in the transition range (i.e. 50 to 11 percent RH).
Thus, total shrinkage (Et) between 50 to 11 percent RH is given by the following
expression:

(17)

Testing of the Proposed Equations for Predicting Drying Shrinkage of Paste in the RH
Range of 95 to 20 percent

Excellent agreement between measured and predicted ultimate shrinkage of a

0.4 W/C ratio paste was obtained at any RH between 95 and 20 percent. This is shown
in Fig. 7. For the 0.6 W/C ratio paste good agreement was obtained as seen in Fig. 8.
For this paste, the predicted shrinkage values are slightly below those measured with
a maximum deviation of about 20 percent at 75 percent RH. This deviation might be
explained by the other two mechanisms. But, unfortunately, these two mechanisms
can not be quantified by any mean.
 Prediction of Drying Shrinkage of Portland Cement Paste 83

Thus, the empirical equations developed in this study can be used to predict the
ultimate drying shrinkage of pastes of different W/C ratios in the RH range between
95 and 20 percent. The total estimated surface area and cumulative capillary surface
area values for the two paste systems used are shown in Table 5.

1 . 2 ....----.....,------....-----..,..-----~-------.

0.4 W/C ratio paste

1.0 measured
predicted

0.8

0.6

0.4

0.2

o. 0 '-- ....A- ~ ....._ --'-_ __ _ . i l_ _ _

o 20 40 60 80 100
Relative humidity, 0/0

Fig. 7. Predicted and measured curves ofdrying shrinkage versus RH of Portland cement paste hydrated
for 165 days
84 lamal A.AI-Mudaiheem

1.2-----------......- ----T-----.-...,-----...
0.6 W/C ratio paste

1 .0 _. _. - . measured

\ predicted

0.8 \

0.6
"x"

0.2

o.OL..- -.... ---'- ~ --'-_ _____........

o 20 40 60 80 100
Relative humidity, 0/0

Fig. 8. Predicted and measured drying shrinkage versus RH curves of Portland cement paste hydrated
for 165 days

Table 5. Surface area values for the 0.4 and 0.6 W/C ratio pastes
W/C Length ofcuring (days) Total surface area Capillary surface area
(m2/gdry) (m2/gdry)

0.40 165 53 27
0.60 165 83 50

Conclusions

The Gibbs-Bangham mechanism was found to be the major contributor to shrin-

kage stresses of cement paste in the entire RH range of 95 to 0 percent. The capillary
stress mechanism was found to be active in the RH range of 95 to 50 percent. Its con-
tribution to overall shrinkage increases with W/C ratio and is maximum at about 50
percent RH. The capillary stress shrinkage at 50 percent RH is about 50 to 60 percent
of Gibbs-Bangham stress shrinkage.
 Prediction of Drying Shrinkage of Portland Cement Paste 85

The contribution of the two mechanisms to ultimate drying shrinkage versus RH

can be obtained from shrinkage versus increase in surface free energy curves.
It was found that maximum Gibbs-Bangham stress shrinkage, which occurs at 0
percent RH, can be accurately predicted from an empirical relationship with total
surface area as obtained from nitrogen V-t curves. The maximum capillary stress
shrinkage, which occurs at 50 percent RH, can be accurately predicted from another
empirical equation which incorporates the cumulative capillary surface area. The
capillary surface area values were calculated from V-t adsorption curves. The shrin-
kage equations were extended to include effect of RH of drying in the RH range of
95 to 20 percent.
Shrinkage stresses below about 50 percent RH are mainly due to the Gibbs-Ban-
gham mechanism. Drying shrinkage between about 50 and 11 percent RH is approx-
imately 52 percent of the original Gibbs-Bangham shrinkage.
The proposed equations for predicting ultimate drying shrinkage of cement
pastes were tested on shrinkage results of two different W/C ratio pasts. Overall
agreement with shrinkage results was found to be good in the RH range tested. The
differences between actual and predicted curves could be explained by the other two
mechanisms if there were any means that they can be quantified.

References

[1] Verbeck, G. "Pore Structure." Significance of Tests and Properties of Concrete and Concrete Mak-
ing Materials. American Society for Testing and Materials, Philadelphia, STP 169B, (1978), 262-
271.
[2] Powers, T.C. and Brownyard, T.L. ACI Journal Proceedings, 43 (1947),101,249,469, 549,669,
845,933.
[3] Powers, T.C. "Mechanisms of Shrinkage and Reversible Creep of Hardened Cement Paste." Proc.
Intern. Con! Struct. Cone. London (1965), 319-334.
[4] Powers, T.C. "The Thermodynamics of Volurrle Change and Creep." Mater. Struct., 1, No. 6
(1968),487-507.
[5] Feldman, R.F. "Sorption and Length Change Scanning Isotherms of Methanol and Water of Hyd-
rated Portland Cement." Proc. 5th Intl. Symp. Chem. Cem., Tokyo, Part III (1968), 53-66.
[6] Feldman, R.F. and Sereda, P.J. "Sorption of'¥ater on Compacts of Bottle-Hydrated Cement, 2.
Thermodynamic Considerations and Theory of'Volume Change." Journal of Applied Chemistry,
14 (1964), 93-104.
[7] Helmuth, R.A. "Dimensional Changes and Water Adsorption of Hydrated Portland Cement and
Tricalcium Silicate." M.Sc. Thesis. Chicago Institute of Technology, (1965).
[8] Wittman, R.H. "Interaction of Hardened Cement Paste and Water." Journal ofAmerican Ceramic
Society, 56, No. 8 (1973), pp. 409-415.
[9] Parrott, L.J. and Young, J .F. "Shrinkage and Swelling of Two Hydrated Alite Pastes." Int. Symp.
on Fundamental Res. on Creep and Shrinkage of Concrete, by Martinus Nijhoff Publishers (1982),
35-48.
[10] Feldman, R.F. and Sereda, P.J. "A Model for Hydrated Portland Cement and its Practical Implica-
tions." Journal of Engineering, 53, No. 8,9 (1970), 53-59.
86 lamal A.AI-Mudaiheem

[11] Young, l.F.; Berger, R.L. and Bentur, A. "Shrinkage of Tricalcium Silicate Pastes: Superposition
of Several Mechanisms." Cementa, 75, No. 3 (1972), 391-398.
[12] Feldman, R.F. and Sereda, P.l. "A Model for Hydrated Portland Cement as Deduced from Sorp-
tion Length Changes and Mechanical Properties." Mater. and Struct., 1, No. 1 (1968),509-520.
[13] Bazant, Z.P. "Thermodynamics of Hindered Adsorption and its Implications for Hardened
Cement and Concrete." Cem. Cone. Res., 2, No. 1 (1972), 1-16.
[14] Wittman, F.H. "The Structure of Hardened Cement Paste-A Basic for a Better Understanding of
the Materials Properties." Cem. Cone. Assoc., London, England (1976), 96-117.
[15] Ishai, O. "The Time-Department Deformational Behaviour of Cement Paste, Mortar and Con-
crete." Proc. Cont Structure of Concrete and Its Behaviour Under Load, Cement and Concrete
Association, London (1968), 345-364.
[16] Hansen, W. "Investigation of Shrinkage Mechanisms in Hydrated and Carbonated Tricalcium Sili-
cate System." Ph.D. Thesis, University of Illinois, (1983).
[17] Litvan, G.G. "Variability of the Nitrogen Surface Area of Hydrated Cement Paste." Cem. Cone.
Res. 6 (1976), 139-144.
[18] Day, R.L. "Relations Between Methanol and Portland Cement Paste." Cement and Concrete
Research, 11, No. 3 (1981),341-349.
[19] Parrott, L. "Measurement and Modeling of Porosity in Drying Cement Paste, Symp. Proc. Mate-
rials Research Society." Microstructural Development During Hydration ofCement, 85 (1987), 91-
104.
[20] Cranston, R.W. and Inkley, F.A. "The Determination of Pore Structures from Nitrogen Adsorp-
tion Isotherms." Advances in Catalysis, 9 (1957), 143-154.
[21] Bangham, D.H. and Fakhoury, N. "The Swelling of Charcoal. I Preliminary Experiments with
Water Vapor, Carbon Dioxide, Ammonia and Sulphur Dioxide." Proc. of the Royal Society, Lon-
don, Series A, 135 (1930), 81-89.
[22] Bangham, D.H.; Fakhoury, N. and Mohamed, A.F. "The Swelling of Charcoal, 11. Some Factors
Controlling the Expansion Caused by Water, Benzene and Pyridine Vapors." Proc. Royal Society,
London, Series A (1932), 138-162.
[23] Amberg, C.H. and Macintosh, R. "A Study of Adsorption Hysteresis by Means of Length Changes
of a Rod of Porous Glass." Canadian Journal of Chemistry, 30 (1952), 1012-1032.
[24] Bangham, D.H. and Maggs, F.A. "The Strength and Elastic Constant of Coal in Relation to their
Ultrafine Structure." Proceedings of a Conference on the Ultrafine Structure of Coals and Cokes,
British Coal Utilization Research Association Committee, London (1943),118-130.
[25] Hiller, K.H. "Strength Reduction of Length Changes in Porous Glass Caused by Water Vapor
Adsorption." Journal ofApplied Physics, 35, No. 5, (1964), 1622-1628.
[26] Yates, D.l.C. "Molecular Specificity in Physical Adsorption." Advances in Catalysis. 12 (1960),
265-311.
[27] Hwang, C.L. "Drying Shrinkage and Microstructure of Hydrated Cement Paste." Ph. D Thesis,
University of Illinois (1983).
[28] Hobbs, D.W. and Mears, A.R. "The Influence of Specimen Geometry upon Weight Change and
Shrinkage of Air-Dried Mortar Specimens." Magazine of Concrete Research, 23, No. 75, 76 (1971),
89-98.
[29] L'Hermite, R.A. and Mamillan, M. "Further Results of Shrinkage and Creep Test." Inter. Confer.
on the Struct. of Cone., Cem. and Cone. Assoc., London (1965), 423-433.
[30] Helmuth, R.A. and Turk, D .H. "The Reversible and Irreversible Drying Shrinkage of Hardened
Portland Cement and Tricalcium Silicate Paste." Journal PCA Res. Dev. Lab. 9, No. 2 (1967),8-21.
[31] Parrott, L.l.; Hansen, W. and Berger, R.L. "Effect of Drying and Rewetting Upon the Pore Struc-
ture of Hydrate Alite Paste." Cem. Cone. Res., 10, No. 5 (1980), 647-655.