Cracking in Pavements Influenced by

Viscoelastic. P roperties of Soil-Cement

K. P. GEORGE, University of Mississippi

To develop a better understanding of shrinkage cracking (which is known

to be time-dependent) of pavement layers, the author believes that the
material should be considered to be viscoelastic rather than elastic . A
step is taken in this direction when a viscoelastic model is proposed for
setting up the stress-strain-time laws for soil-cement. Results of a
series of constant stress creep tests indicate that the Burgers model
shows the greatest promise in mechanical simulation of actual soil-
cement behavior. Rheological parameters that control the soil prop-
erties are evaluated from the test data. The experimental study of
creep has proved beyond doubt that creep and shrinkage are not inde-
pendent phenomena. That is, the soil-cement which exhibits high
shrinkage also shows generally a high creep.
The implications upon the crack spacing and crack width of the hy-
pothesis that the stress-strain relation tends to be time-dependent are
discussed. The fact that shrinkage stresses developed ina viscoelastic
material are on the average about 50 percent smaller than those in a
comparable elastic material is significant insofar as crack frequency is
concerned. Evidence is overwhelming that rapid shrinkage favors large
stresses in the material. The agreement between the observed or re-
ported values of crack width and that computed according to the visco-
elastic theory is excellent.
In predicting the behavior of pavement layers subjected to ambient
conditions, the viscoelastic approach is superior to any known elastic
or empirical laws. The results show convincingly that cracking of ce-
ment base can be controlled by adequate extended curing.

•THROUGH the use of stabilizing agents, locally available road materials are cur-
rently being stabilized for economical highway construction. Of the dozen or so dif-
ferent materials reported in the literature, those most commonly used are cement,
lime, lime-fly ash, and asphalt. Although it is generally believed that under field con-
ditions the behavior of the resulting material tends to be inelastic, all of the available
methods employed to estimate the stresses in the bases and subgrades are based on
the conditions of ideal elastic behavior. There is a great need, therefore, for a sys-
tematic investigation of the stress-strain-time relationships (simply known as creep)
of these materials. Creep is defined as the total time-dependent deformation of the
material due to load. A knowledge of the creep in stabilized soils is particularly valu-
able in the study of the drying shrinkage and the resulting cracking by virtue of the hy-
pothesis that shrinkage and creep are interdependent phenomena.
The principles of viscoelasticity have been successfully used to explain the mechan-
ical behavior of high polymers and much basic work (1) has been accomplished in this
area. Mechanical models used t9describe the deformation-time behavior under stress
consist of combinations of elementary units of springs and dashpots in series or paral-
lel. The relationship between stress cr and strain £, respectively, for the spring

Paper sponsored by Committee on Soi I-Portland Cement Stabilization and presented at the 48th Annua I
Meeting.
47
48

and the dashpot is a = Et: and a = 11(dt:/dt) where E = Young's modulus (lb- in. - 2 ) and
'17 =modulus of viscosity (lb-sec-in. - 2 ). The fundamental models of linear viscoelas-
tic solids are the Maxwell model and the Kelvin model, which are comprised of spring
and dashpot arr anged, respectively, in a series and in par allel. The stress-strain-
time behavior of these models is treated in standard references on viscoelasticity and,
therefore, will not be discussed here.
In recent yea:r,-s, the theory of viscoelasticity has been employed to explain the me -
chanical behavior of asphalt mixtures (2, 3, 4) and soils (5, 6). The present investi-
gation, however' is directed to the study of the viscoelastic-behavior of soil-cement,
and to the application of these concepts in developing a better understanding of crack-
ing in pavements. Although the results and discussions are specially applicable to the
rheological behavior of soil-cement, the concepts presented herein are considered to
have a wider applicability.
A soil-cement base, or any pavement layer for that matter, which tends to contract
becfl.use of shrinkage or ambient temperature, if fully or partially prevented from do-
ing so, will be severely stressed in tension. When the stress exceeds the tensile
strength of the material, the base cracks. Accordingly, George (7) has presented
simplified solutions to the crack-spacing and crack-width problem-: In this study, it
is assumed that the material is perfectly elastic. The implications of a more realistic
hypothesis that the stress-strain relation tends to be time-dependent are discussed in
the present paper. Using information obtained from a series of constant stress creep
tests, the author chooses an appropriate mechanical model for predicting the deforma-
tional behavior of soii-cement. Analytical solutions, show'mg the influence upon crack
spacing and crack width of the rheological parameters of the material, will be pre-
sented. The predicted value of the crack width is compared with that obtained from
field observations. Using the rheological model, the researcher extends the theoret-
ical study to bring out the effect of shrinkage rate upon the ambient stress and in turn
the cracking.
MATERIALS AND METHODS
Materials
The scope of the study made it necessary to restrict the laboratory investigation to
tests on two soils. Table 1 gives compositional data and a brief statement of physical
properties. Grain-size distribution of these soils is illustrated elsewhere (l, Fig. 3,
p. 62). For convenience, each soilisidentified by aoneletter-two digit system, for ex-
ample K03 means soil No. 3 with kaolin as the predominant clay mineral.
Type I portland cement and distilled water were used in preparing test cylinders.
Sample P1·eparation
'£he procedure for preparing the soil and blending the soil-cement mixture has b en
described elsewhere (!!). The test cylinders, 2.8 in. in diameter and 8.4 in. high, were

TABLE 1
SOIL IDENTIFICATION AND COMPOSITIONAL DATA

K03 M30
Description
(Tishomingo County, Miss .) (Vicksburg, Miss .)

Gradation (:t; finer by wt)

2mm 92 100
0.05 mm 20 92
0.002 llllll 16 23
Liquid limit, <t 31 37
Plastic limit 10 13
Shrinkage limit, <( 20 22
Textural c1assU!cat1on iiandy Silty ciay
Engineering classification A-2-4(0) A-6-9
Predominant clay mineral Kaolinite Montmorillonlte, illite
Optimum moisture, <t 13. 5 17.0
Proctor density, pcf 118.B 110.3
Maximum linear shrinkage on
drying in 55:1 RH, in ./in. 0. 0025 o. 0129
 49

molded according to the ASTM Suggested

Method of Making and Curing Soil-Cement
Compression and Flexure Test Specimen
in the Laboratory.

Apparatus and Test Procedure

The cylinders, after seven days' moist
curing, were tested in a constant axial
creep test in the loading frame (Fig. 1).
Dial gages reading to 0.0001 in. served
to measure the deformation. The sus-
tained stress applied to the specimens
approximated 50 percent of the ultimate
strength.
Although creep in tension was relevant
insofar as cracking in pavements was con-
cerned, some difficulties were encoun-
tered in its experimental determination;
therefore, it was decided to evaluate creep
in compression. To justify this approach,
the results of a 10-year study on concrete
Figure 1. Schematic of loading frame to test for made by the U. S. Bureau of Reclamation
creep. Mechanical advantage of the lever sys- (9) may be cited. The authors concluded
tem 40. that the magnitude of creep in tension and
in compression was of the same order,
and both were similarly affected by vari-
ous factors.

DISCUSSION OF RESULTS
Although the major emphasis in this study was on evaluating the effect of creep on
cracking of pavements, one important result which evolved during the experimental
determination of creep warrants special mention here.

Effect of Drying Shrinkage on Creep

A brief survey of the literature on the creep of concrete revealed two prevalent
points of view concerning the influence of drying shrinkage on creep. One school of
thought holds that creep and shrinkage are additive. Creep is thus calculated as the
difference between the total time deformation of the loaded specimen and the shrinkage
of a similar unloaded specimen stored under the same conditions during the same
period. The other viewpoint is that drying of moisture during a period of sustained
loading influences creep. The creep results shown in Figure 2 pertain to identical
soil-cement test specimens, one exposed in a 55 percent relative humidity (RH) con-
dition (curves labeled a) and the other sealed against moisture gain or loss (curves
labeled b). These results and others not reported here emphasize that creep and
shrinkage are not independent phenomena.
Based on the present study, the following conclusions seem relevant. First, for a
given soil-cement mixture, creep is higher the lower the relative humidity. Second,
as seen from the results of soil M30, a mixture which exhibits high shrinkage generally
also shows a high creep. It is hoped that more detailed results will be presented in
another report (10).
In view of the finding that drying enhances creep, the author decided to perform
creep tests under conditions of nearly 100 percent RH.

Influence of Creep on Shrinkage-Cracking

Shrinkage Cracking of Pavement Layers-Elastic Solution-Cracksinpavementlayers
may be due to the contraction caused by two factors : changes in moisture content and
50

ambient temperature. Since the contraction resulting from drying is more important
in s tabilized s oils than that caused by temperature, emphasi s in this study 's on th
cracking caused by drying shrinkage. Tension stresses can be set up in a slab as a
r esult of linear shrinkage. If the slab is fr ee to move (no fric tion between the slab
and the underlying layer), stresses will not result. If friction exists between two ad-
jacent layers, however, restraint results in the top slab from the friction forces. In
long slabs the friction forces may be sufficient to cause over-stressing of the material
which produces a crack. Based on these assumptions, George (7) has derived expres-
sions for crack spacing and crack width. Respectively, these equations are
2 Ou
Lmax -- (1)
1-"Y
and
oT E:c L -~ (2)
4 Et
where
Lmax slab length at which tensile stresses become critical, ft;
Ou ultimate te ns ile strength, lb-fC 2 ;
µ. coefficient of sliding friction;
y unit weight of material, lb-ft- 3 ;
OT total crack width, ft;
E:c total s hrinkage, in./in. ; and
Et modulus of elasticity of soil-cement in tension, lb-ft- 2 •
From Eq. 1, it appears that for a specific slab placement, the crack spacing is a
direct function of the tensile strength of the material; whe r eas from Eq. 2, it can be
deduced that the crack width is primar -
ily a function of the linear shrinkage.
Some approximate calculations for
- M30, 107o cemem crack spacing and crack width were
0 .8 - - - - K03, 6'o/o made with the soil properties chosen
as follows: Soil K03 with 6 percent ce-
ment is calculated to have a tensile
strength of 33 lb-in.- 2 (60 percent of
the laboratory strength), a unit weight
of 119 lb-ft- 3 , a shrinkage of 0.0010
.
"'
in./in. (8), and a modulus in tension of
66,000 lb-in. - 2• The coefficient of sub-
c
~ 0.41 - - - - -..;- . i - - - - - - 1 - - - - - - + --l grade resistance was assumed to be 3.0
"' (11) . Calculations s how that Lmax =
27 ft, and OT = 0 .238 in . When the com -
(bl
puted quantities were compared with
those observed in the field, ·there was
reasonable agreement between the pre-
dicted and observed values of crack
.o-----....0.- ---- (bl

spacing. So far as crack width was con-

40 80 120
cerned, however, the agreement was
extremely poor. For example, the mea-
sured crack width ranged from 0.02 in.
Figure 2, (a) Creep of test specimens exposed in a to 0.10 in., which bracketed the average
55 percent RH condition; creep is calculated as the value of 0.02 in. reported by Marshall
difference between the tota I time deformation of the (12).
loaded specimen and the shrinkage of a similar un- - The disparity in results can be at-
loaded specimen. (b)Creepof test specimens sealed tributed to the fact that in deriving Eqs.
against moisture gain or loss-soil: silty clay (M30) 1 and 2, it is tacitly assumed that
at 10 percent and sand (K03) at 6 percent cement; cement- treated soil is perfectly elastic .
stress level: 50 percent of ultimate, 165 and 195 This assumption may not be valid. The
psi, respectively. following discussion, therefore, is an
 51

iL-, (o) attempt to rationalize these expres-

sions , based on a realistic hypothesis
~~ that the material is viscoelastic.
Rheological Model for Soil-Cement-
As stated p r eviously, the theor y of
viscoelasticity is concerned with the
relation between strain as a function
of time and stress as a function of
time. To illustrate this behavior con-
veniently, several mathematical mod-
els have been proposed. A model that
is to represent accurately the me-
(c)
" '
Time, t chanical properties of the material
should contain certain elements that
1. (ampule Em from in 5 lonlaneo~
elo5tlc;; deformalian
simulate the major characteristics
f~limote
2. Ek
05 i;hown al ler1
3 . Compvt.11 'Im from slope in (q)
(elastic, inelastic and/or time-depen-
region ol curve
4. f vgluole T(ond lhl4 Ilk) from
dent strain) observed in various me-
Time, '·t chanical tests. Besides the Maxwell
lit ro one point in (p) region
of curve,

fluid and the Kelvin solid, there are

numerous ways of systematically
building up more complicated models.
Figure 3. Strain-time relationship for the linear and Of these composite models, the linear
the Burgers models: (a) assumed load-time relation- model and the Burgers model have
ship; (b) the linear model; (c) strain-time relationship been chosen for further study and
for the linear model; (d) the Burgers model; and (e) verification. To illustrate their suit-
strain-time relationship for the Burgers model. ability to predict the behavior of soil-
cement, typical results of tests on a
particular mixture are also presented.
The linear model is obtained by adding a Maxwell model to a Hookean model in par-
allel, as shown in Figure 3b. The behavior of this system simulates a material with
a continuous skeletal structure. The coupled parallel Maxwell unit serves to transfer
the load to the elastic elements in the model following the relaxation or lag time dic-
tated by the Newtonian model. The general strain-time curve, for a constant stress
applied at t = 0, is given in Figure 3c. When this cur ve is c ompared to that for a typ-
ical soil-cement mixture (Fig. 4, data points) the drawback of t his model is apparent
in that the removal of cr after a ce r tain period of time, say t = t 1, will caus e the m odel
to regain the lost strain . This is governed by the relaxation time. The linear model ,
therefore , i s abandoned and will not be fur ther di scussed her e .
The Burgers model combines in a series the Kelvin and !VlPxwell models as shown
in Figure 3d. In the phenomenological theory of linear isothermal viscoelasticity the
constitutive law for the Burgers model is

71m Ek + '17m Em + 71k Em dcr (3a)

O" +
Em Ek dt

where
£ = strain, in./in.;
er stress, lb-in. - 2 ;
t time, sec;
E modulus of elasticity, lb-in. - 2 ;
T7 modulus of viscosity, lb-sec-in.- 2•
Subindices refer to model elements in Figure 3d.
This equation can be written in the normalized form:

(3b)
52

in which
'l'lm Ek + 11m Em + 7Jk Em
P1 = Em Ek

The solution of Eq. 3a for strain f: for a constant stress cr acting between t = 0( =t 0 )
and t = t 1 is

((t) cr [ E~ + 'I'!~ + Elk ( 1 -exp (-t/r))] (4)

for to s: ts: t1. The value

~eased
T =: is called the relaxation time. If the stress cr is re-
at t = t 1, the total strain would be governed as follows:

(5)

The first term on the right-ha.."ld side of Eq. 5 represents the elastic deformation from
the Hookean element in the Maxwell model. The permanent deformation, or set, is
given by t11e second term on the right-hand side. The Kelvin model (third term) plays
its part in the aftereffects of load or stress survival as shown in the stress-strain-time
diagram shown in Figure 3e.
The iew characteristic results of soil-cement, which furnished strong clues as to
the suitability of the Burgers model in the investigation, are that (a) the instantaneous
recovery in the specimens (Fig. 4) is nearly the same as the instantaneous elastic de-
formation, and (b) the specimens subjected to various amounts of creep exhibited the
same amoun.ts of elastic recovery. It is hypot."1esized, therefore, that Burgers model
shows the greatest promise in mechanical simulation of actual soil-cement behavior.
Ex erimental Verification of Rheolo ·cal Model-The viscoelastic constants (Em,
Ek, 71m, 'l'lk for the four-element model are most easily approximated from the re-
sults of a creep test described in Figure 3e.
To determine the ability of the Burgers model to reflect the characteristics of soil-
cement mixture, and to determine the usefulness of the equations developed for the
model, rheological constants estimated from the creep test are applied to Eqs. 4 and 5.
As shown in Figure 4, the agreement between the experimental and predicted data is

Soll: K03 6 per cent cement

Strv:ss Level: ~percent of ultimate, 200p$i
Solid Line-Theory, Equations 4 ondS
Broken Line - Data Points

c
·~
II)

'.!
.. 0.04
Soll Parameters:
Erns0.89•I06 lb-in~•
E,.= l.70•10 1 lb-ln~ 2
'1.n= 2.70•10• lb-irf1 -mln
ry1 = 4.59•10' lb-in·•-m
-r• 2100nm

40 BO 120 160
Time, hours

Figure 4. Comparison of creep test data with that predicted by the Burgers model.
 53

~ldool~~ Sh~-Time Rol!tion lo sot!... Equotio!i 6

(.! _______ • /
0.2

§
~. -K03
.,
.- .- ~
--- ---
'i ,.,.. -
l
f
z
0.10
I I

'
/
_,,.~
/

b
/
(/) '
k ' =- - M07
: /

I/
v. ~

•.;-/
:.,,
40 80 120 160
Time, hours

Figure 5. Time-rate of shrinkage of soi I-cement mixes when air dried at 55 percent RH and 72 ± 4° f-
cement content in both soi Is is 6 percent (~).

excellent. The model, therefore, permits a reasonable prediction of such data from
those of the creep test.
Effect of Creep on Crack Spacing-The effect of creep on crack spacing seems to be
only indirect, mainly insofar as the creep affects the tensile strength of the material.
Eq. 1 reveals that crack spacing is strongly dependent on tensile strength. Although
no study has yet been reported concerning the effect of creep on the ultimate tensile
strength of soil-cement, Neville (13), in discussing the results of his creep study on
concrete, reported that the influence of creep on the ultimate strength of a simply sup-
ported beam subjected to a sustained load was not significant. He also stated that the
creep of plain concrete did not per se affect the strength, although under very high
stresses (85 to 90 percent of the rapid ultimate load) creep hastens the approach of the
limiting strain at which failure takes place. So far as soil-cement is concerned,
similar arguments may lead to the conclusion that, when subjected to constant stress
creep, it could fail prematurely because the maximum strain exceeded the limiting
strain. This result, however, is not considered significant. The important influence
on crack spacing could be in the internal stresses developed during shrinkage. Con-
sidering the material is viscoelastic, the tension stress developed as a result of the
time-dependent shrinkage needs to be evaluated.
As indicated in the previous section, the behavior of soil-cement can be satisfac-
torily represented by that of the Burgers model. Assuming that the material under-
goes a constant shrinkage rate, that is, (d 2 ( /dt2) = 0, one can write Eq. 3b as follows:
(6)

As shown in Figure 5 the stress developed in the material because of the imposed strain
can be computed by solving the differential equation with suitable boundary conditions.
The solution for the stress in terms of the rheological parameters is

1 - 1- 1 -1
-
~ d (-~)
E ]
CT (t:") Em 1
[ Em +
~ ~
exp (-at) +
T
exp (7)

where
2

Em)
+ - -
'Tim
54

T7k
T Ek, often called the retardation or delay time.

The ratio t: /t 1 signifies the shrinkage rate as shown in Figure 5.

With the appropriate rheological constants substituted in Eq. 7, the stress developed
in two soils (K03 and M30), as a function of time for two values of shrinkage rates, is
calculated and the result is shown in Figure 6. The stress developed in the viscoelas-
tic material is always smaller than that in a comparable elastic material, and the dif-
ference in the behavior of the materials will be more apparent as the shrinkage rate
decreases. In other words, the material exhibits stress relaxation; and the chances
are that a pavement base of viscoelastic material, such as soil-cement, would exhibit
even fewer cracks. Furthermore, Figure 6 reveals that the stress ratios are nearly
identical in the two soils (at t 1 = 10,000 min), although the same ti-value results in
distinct shrinkage rates. This finding simply indicates that the stress ratio is indepen-
dent of the shrinkage rate, but strongly dependent on the soil properties, especially
Tim/Em ratio.
Shrinkage Rate and Cracking- Eq. 7 reveals the internal stress developed in the ma-
terial whe n a constant rate of s train caused by shrinkage or temperature change is
assumed. It may be asserted that cracking takes place when,

1- - Ta
1 .1 1 ]
=
£i Tim
(t) Em [ Em + ~ - T{J
exp(-at) + r;::r exp (-{Jt) (8)

where au ultimate tensile strength of soil-cement, lb-in. - 2 •

Examining Eq. 8, it appears that the
risk of cracking is determined by an in-
teraction of the shrinkage rate, the ten-
sile strength, the deformational proper-
ties of the soil-cement, and time. The
important result, however, is that the
tensile stress developed because of re-
straint will be a direct function of the
rate of shrinkage (t'). In other words, if
the load-deformation relationship of the
material is assumed to be viscoelastic,
rapid shrinkage should favor large
stresses. So far as strength of the ma-
terial is concerned, it maybe conjectured
that brittle mater ials could exhibit slightly
increased rupture stress with the rate of
0. 0 0 .4 0.8
loading (17). This increase, however,
is considered insignificant compared to
Dimensionless Time, 1/t, the former. The stress developed in re-
Cfv• " ..!.. l~. I - JT ex p L,._t). - Jr expL,Bt)~
1 lation to rate of shrinkage is significant
C5E 1 LEm 11 -a \ a.- ,6' \ J in the cracking of soil-cement base, since
the shrinkage rate of kaolinite soil-cement
Figure 6. Time-dependent stress in viscoelastic is much larger than that of montmoril-
soi Is due to time-dependent shrinkage. Ratio of the lonite. For example, in Figure 5, the
stress computed considering the material is visco - kaolinte and montmorillonite soil-cements
elastic to that considering elastic is designated attain the same maximum shrinkage in
stress ratio, Soi I parameters from Figure 4. Param- 50 and 250 hours, respectively. In other
eters of soi I M30 at 10 percent cement : Em = 0.30 words, the tension stresses and the ten-
x 106 ; Ek= 0.33 x 106 ; Tim = 1.0 x 109 ; 1/k = l.0 x dency to crack will be much greater in
109 • Time taken to attain the maximum shrinkage kaolinite soil- cement than in montmoril -
is designated t1 • lonite soil-cement.
 55

90
To substantiate this hypothesis, the
crack patterns of two models, soil-
cement mixtures of K03 and M07, are
obtained. The model was a 24 in. by 24
'"c in. by 1 in. deep soil-cement slab molded
"I
0
70 in a flat wooden container 27 in. by 27 in.
by 6 in., inside dimensions. A short de-
... scription and detailed results on model
testing may be found elsewhere (18). Sig-
nificantly, the accumulated crackTength
~ 50
\ and crack propagation of K03 are greater

:;
e
"'- than that of its counterpart M07. Appli-
cation of this result to field situations
would suggest that kaolinite soil-cement
pavement bases warrant special attention
30 and curing during the first few days in
00 0 .2 0.4 order to minimize serious cracking.
6
Shrinkage- Rate, E,, to- in./ in. per min,
Eq. 8 implies that the ultimate strain
at rupture depends on the whole strain
Figure 7. Limiting strain (strain at breaking) de- history of the material before failure.
creases with shrinkage rate according to Eq. 8. When shrinkage occurs slowly, the strain
Soil parameters from Figure 4. capacity of soil-cement is greater than
when rapid shrinkage takes place. This
is exemplified by inserting the experi-
mentally determined rheological constants of the soil-cement in the criterion of crack
formation, Eq. 8, and calculating the strain at failure at different rates of shrinkage.
Such data (Fig. 7) again stress the importance of proper, extended curing of soil-cement
to minimize undesirable cracking.
Effect of Creep on Crack Width-The crack width (2) is influenced by two opposing
factors: the tendency of the soil-cement to shrink, compensated to some extent by the
elongation of the material. Accordingly, the equation for the crack width was derived:

(2)

The first term on the right-hand side, which accounts for the contraction caused by
shrinkage, is not affected by creep. The second term, however, needs revision in
view of the fact that the tensile stress resulting from the restraint below is time-
dependent. The effect of shrinkage on tensile stress seems to be only indirect, mainly
insofar as the shrinkage affects the slab displacement. The key factor that influences
the tensile stress, however, is the coefficient of the subgrade resistance µ. which, ac-
cording to the tests by the Bureau of Public Roads (11), tends to increase with the slab
displacement. Test results also showed that the maximum horizontal resisting force
that could be developed was for an 0.10-in. displacement. Making the simplifying as-
sumption that the subgrade resistance is linearly time-dependent and based on the hy-
pothesis that the space variation of stress (7) is linear (crx = µ. y x), an expression for
the stress is -
t
O-(t, x) = µ(t) yx a (x) ti" (9)

The stress function can be represented in two-dimensional space as shown in Figure

8. The problem thus reduces to evaluating the strain and thereby the deformation of
a section of pavement, length L, subjected to the prescribed stress distribution.
First, the strain caused by the time-dependent stress can be computed by using the
hereditary integral (14). The hereditary integral, applicable to a stress distribution
where a stress a 0 is suddenly applied at t = 0, but that a then varies as an arbitrary
function cr(t) is
56

E: (t) a(O)J(t) + f J(t - t')

dO' ' dt'
dt '
(lOa)
0

Where J (t) is known as the creep compliance of the material; for the Burgers model

t P1 qi - q2 ( ) P2 ( )
J (t) = - + 2 1 -exp ( - t/r) + - exp ( - t/r) (11)
ql q q2
1

in which Pu Pi, q 1, and q 2 have the same interpretation as in Eq. 3b. Eq. lOa shows
how the strain at any given time depends on all that has happened before. This is quite
different from what happens in an elastic material, whose strain depends at any time
solely on the stress acting at that time only. Through integration by parts, Eq. lOa
may be brought into another, often more useful, form:

r(
t
I) dJ (t - t I) dt I
E: (t) O' (t ) J (0) + ~ O' t d (t - t I) (lOb)
0

Second, the total elongation can be obtained as follows:

1./2
62 = 2 I 0
E(t, x) dx (12)

_, T /.L .L / \

Substituting for a(t), j (0), and ':f'ct--t~/ in Eq. lOb and the resulting equation in turn
in Eq. 12, and performing the required integration and simplification, the following ex-
pression for 5;i in the range t < t 1, is obtained:

- µyL2 { [-1 + _1 + _ t
- 4 Em Ek 217m
J_!_t 1
+ __!__ tTl [exp (- !_T) - 1]}
Ek
(13)

For t > t 1 the expression becomes

tl )~ } (L)
2
- ~
yL {- 1 + - 1 + - 1 [ t - -t1 ] + - 1 -T [ exp ( - -t ) - exp ( - -
t -- 4
4 Em Ek 'l'/m 2 Ek t 1 T T

To emphasize the significance of the elongation (6 2), it might be well to note that the
crack width decreases with the increase in 6 2 •

o\ Figure 8. Stress variation in two-dimensional

~e~~ space. Stress linearly varies: (i) from crack-
/ ' ' ed ge to center of slab and (ii ) from t = 0 to
' •U, x•2)
t = t 1 • Subgrade resistance and in turn the
x
stress reaches maximum in time tr

,..,.«t•t;,11=f>
/
/
 57

62\IE
-
02E
E~ Emt) -I +Em- -'~exp --1
- .. It..__,_.•_
E• 277'" t, E, t,
t t jJ
i;

Figure 9. Reduction ratio plotted against

time . Ratio of the elongation for a visco-
e las tic materia I to that for an e las tic is
designated reduction ratio .

Dimensionless Time, t

In order to investigate the influence of the viscous properties of the material on

crack width, Eq. 13 is evaluated in soils K03 and M30, respectively, at two values of
shrinkage rates, and the resulting data in dimensionless form are shown in Figure 9.
In these computations, the subgrade resistance and the linear shrinkage were assumed
to attain their respective maximum in the same time interval (tJ. The fact that the
reduction ratio (o2VE/li2E), elongation of viscoelastic material to that of elastic, in-
creases with time signifies the importance of viscous properties of soil-cement in re-
ducing crack width.
The material properties presented earlier in the report in conjunction with the data
in Figure 9 (reduction ratio = 3) were again used to predict the crack width. The agree-
ment between the actual and the predicted data (0.06 in. compared to 0.08 in.) is excel-
lent, showing that the rheological model reflects the characteristics of the material to
a marked degree.
When the plots for the two soils K03 and M30 are examined, the soil parameters
between soils do not seem to have significant influence on the reduction ratio. Insofar
as the relative significance of the parameters of a mix is concerned, the modulus of
viscosity TJm exerts the most influence on crack width in that a smaller TJm contributes
to narrower cracks.
By comparing the plots for different values of shrinkage rates in Figure 9, one may
infer that by decreasing the shrinkage rate the crack width could be controlled. In
other words, prolonged curing of the base should favor narrower cracks.

RESULTS AND CONCLUSIONS

In this paper, limitations imposed by purely elastic analysis of the cracking prob-
lem are reviewed. In order to take the time-dependent material properties into ac-
count, the Burgers model has been proposed. To validate the predictability of the
model and to approximate the rheological parameters, a series of constant stress creep
tests was performed in the laboratory. The theoretical study, making use of the Burgers
model, resulted in the following observations:
1. Creep and shrinkage are not independent phenomena; for any mixture, creep is
higher the lower the relative humidity.
2. The Burgers model shows the greatest promise in mechanical simulation of
actual soil-cement behavior.
58

3. Stress computed according to viscoelastic theory is on the average 50 percent

smaller than that computed by the elastic theory.
4. Insofar as the rate o.f shl'ink.age is concerned, rapid shrinkage favors large
stresses.
5. The crack width computed according to the viscoelastic the ory is within practi-
cal limits and is in excellent agreement with that observed in the field.
This paper concludes that in predicting the behavior of pavement layers subjected
to ambient conditions, the viscoelastic approach is superior to any known elastic or
empirical laws. The results, although largely qualitative, suggest the possibility of
controlling the crack intensity by enhancing the viscous properties of the material.
The findings of this study are conclusive in showing that cracking of cement base can
be minimized by adequate extended curing.

ACKNOWLEDGMENTS
This report is part of the research conducted in connection with Project 5666 of the
Engineering Experiment Station, University, of Mississippi, under the sponsorship of
the Mississippi State Highway Department and the U.S. Department of Transportation,
Bureau of Public Roads. This support is gratefully acknowledged.
The opinions, findings, and conclusions expressed in this publication are those of
the author and not necessarily those of the State or the Bureau of Public Roads.
Frank C. Yao, Graduat~ Assistant in Soil Mechanics, assisted in the testing and
analysis of data.

REFERENCES
1. Alfrey, T ., Jr. Mechanical Behavior of High Polymers. lnterscience Publishers,
New York, 1948.
2. Secor, K. E., and Iv!onismith, C. L. Viscoelastic Properties of . 4..sphalt
. Con-
crete. HRB Proc., Vol. 41, p. 299-320, 1962
3. Pister, K. S., and Monismith, C. L. Analysis of Viscoelastic Flexible Pave -
ments. HRB Bull. 269, p. 1-15, 1960.
4. Monismith, C. L. Temperature Induced Stresses and Deformations in Asphalt
Concrete. Proc., AAPT, 1965.
5. Singh, A., and Mitchell, J. K. General Stress -Strain- Time Function for Soils.
Jour. Soil Mech. and Found. Div., ASCE, Vol. 94, No. SMl, p. 21-46, 1968.
6. Mitchell, J. K., Campanella, R. G., and Singh, A. Soil Creep as a Rate Pro-
cess. Jour. Soil Mech. and Found. Div., ASCE, Vol. 94, No. SMl, p. 231-254,
1968.
7. George, K. P. Cracking in Cement-Treated Bases and Means for Minimizing It.
Highway Research Record 255, p. 59-71, 1968.
8. George , K . P. Shrinkage Characteristics of Soil~Cement Mixtures. Highway
Research Record 255, p. 42- 58, 1968.
9. U. S. Bureau of Reclamation. A 10- Year Study of Creep Properties of Concrete.
Concrete Laboratory Report No. SP-38, Denver, Colo., 1953.
10. Yao, F. C. Creep on Cracking of Soil-Cement Base. Univ. of Mississippi,
Library, Oxford, 1969.
11. Kelly, E. F. Application of the Results of Research to the Structural Design of
Concrete Pavements. Public Roads, Vol. 20, 1939.
12. Marshall, T. J. Some Properties of Soil Treated With Portland Cement. Sym-
posium on Soil Stabilization, Council of Scientific and Industrial Research,
Australia, 1954.
13. Neville, A. M. Theories of Creep in Concrete. Proc., ACI, Vol. 52, p. 47-60,
1955.
14. Fliigge, W. Viscoelasticity. Blasidell Publishing Co., Waltham, Mass., 1967.
15. Abdel-Hady, M., and Herrin, M. Characteristics of Soil-Asphalt as a Rate
Process. Jour. of Highway Div., ASCE, Vol. 92, No. HWl, p. 49-69, 1966.
 59

16. Hansen, T. C. On Rheology of Hardened Concrete. Proc., Swedish Cement

and Concrete Research Inst., Vol. 37, Stockholm, 1962.
17. Goldstein, M. N., Misumsky, V. A., and Lapidus, L. S. The Theory of Proba-
bility and Statistics in Relation to Rheology of Soils. Proc., Inter. Conf. on
Soil Mech. and Found. Eng., p. 123, Paris, 1961.
18. George, K. P. Mechanics of Fracture and Cracks in Soil-Cement Pavement
Base. Offered for publication in Proc., ASCE, 1969.