TIME-DEPENDENT BEHAVIOUR

OF REINFORCED CONCRETE SLABS

R.I. GILBERT
R.F. WARNER

UNICIV REPORT No. R-173 OCTOBER 1977

THE UNIVERSITY OF NEW SOUTH WALES
KENSINGTON NSW AUSTRALIA 2033
ISBN 85841 138 5
 TIME-DEPENDENT BEHAVIOUR
OF

REINFORCED CONCRETE SLABS

by

R.I. GILBERT 1 and R.F. WARNER 2

SUMMARY:
A non-linear, layered finite element model for predicting the time
dependent behaviour of reinforced concrete slabs under sustained
transverse loading, is presented. The effects of biaxial creep and
shrinkage are accounted for using a difference formulation of a non-
linear creep model and an incremental relaxation type solution proced-
ure. Agreement between theoretical calculations and deflection
measurements from an in-service reinforced concrete floor,slab and
laboratory controlled beam tests is shown to be good.

1 Teaching Fellow, School of Civil Engineering, University of New South

Wales, Kensington, N.S .W., Australia.
2 Professor, School of Civil Engineering, University of New South Wales,

Kensington, N.S.W., Australia.

 1.

1. INTRODUCTION

The finite element method has been widely used to study the short-term
behaviour of reinforced concrete slabs. However, the time-dependent
behaviour of slabs under sustained load and the effects of biaxial
creep and shrinkage have received little research attention, either
theoretically or experimentally [6].

In an earlier report, the authors reviewed previous finite element

studies of slabs. A layered finite element model was also used to pre-
dict the instantaneous, non-linear response of reinforced concrete slabs
under transverse loading [3]. The present paper describes an extension
of that work to treat time-dependent effects in reinforced concrete
slabs. Although directed primarily towards long-term service-load be-
haviour, the extended model can be used to predict the time-varying be-
haviour of slabs subjected to any variable history of loading including
both service loads and overloads.

A non-linear discrete element analysis is used in which the finite

elements are subdivided into layers, each of which may or may not con-
tain steel reinforcement. The different stages of instantaneous mater-
ial behaviour, including elastic strain, tensile cracking of the
concrete, tension stiffening of the intact concrete between the cracks
in the tension zone, compressive yielding of the concrete and yielding
of the steel reinforcement, are represented by suitable biaxial stress-
strain relations for each layer of each element. Compatibility of
deformations between the concrete and steel and perfect shear bond bet-
ween the layers is assumed.

A method of temporal discretisation is employed, whereby creep and

shrinkage strains in any chosen time increment are calculated from a
non-linear constitutive relation and then introduced into the slab
analysis by a direct 'relaxation' type procedure [l].

Theoretical calculations are found to agree well with test results

previously obtained from an existing reinforced concrete flat slab
building and from laboratory controlled beam tests.

2. BIAXIAL CONSTITUTIVE RELATIONS

A non-linear creep model for concrete under time-varying axial stress

has been generalised to treat the case of biaxial stress which occurs in
 2,

slabs. The basic difference formulation was originally used in studies

of time-varying behaviour of beams and columns [9, 10], and has more
recently been used in the analysis of long-term behaviour of indetemin-
ate concrete beams and frames [SJ, As the uniaxial creep model has al-
ready been described (10], only details of the extension to two
dimensional stress states will be mentioned here.
A typical, thin concrete slice of a rectangular finite element is
con~ider~rl. At any time t after first loading, the vector of total
concrete strain is assumed to consist of component vectors of instantan-
eous, creep and shrinkage strains:
{E(t)} = {Ei(t)} + {EC(t)} + {Esh(t)} (1)
or Esh(t)
Ex(t) = E~(t) + Ei(t) +
X

Ey(t) E~(t) E~(t) Esh(t)

y
Exy(t) E~y(t) EiyCt) Ei~(t)
In order to model the short-term material non-linearities, the
instantaneous strains are taken to consist of a linear-elastic, fully
recoverable component, {Ee~(t)}, and a non-recoverable plastic component,
{ePt(t)}:

= (2)

Although these instantaneous strain components are not directly

dependent on time, additional increments of 'instantaneous-, strain may
occur after loading, due to time dependent cracking caused by shrinkage.
Full details of the instantaneous stress-strain relations for the various
non-linear material states have been discussed elsewhere [3], and only a
brief description will be included here.
Experimental investigations [4] have shown that the strength of
concrete under biaxial compression is greater than the uniaxial compres-
sive strength, while the strength of concrete under biaxial tension is
independent of the principal stress ratio and equals the uniaxial tensile
strength. Concrete is here assumed to be elastic-plastic in compression
with compressive yielding governed by the von Mises criterion and assoc-
iated flow rule. This is slightly more conservative than the experiment-
al yield surface for concrete in the biaxial stress state [4], but is an
acceptable approximation in under-reinforced slabs under service con-
ditions.
 3.

A maximum normal stress criterion is assumed to govern tensile

cracking of the concrete. Cracking occurs in a direction perpendicular
to the major principal stress, cr 1 , when cr 1 reaches the uniaxial tensile
strength. After cracking, the total tensile force is transferred across
each crack by the tensile steel but between the cracks the tensile con-
crete carries stress locally, mainly in the direction of the steel bars,
due to the bond between the concrete and the reinforcement. This
'tension stiffening' effect contributes significantly to the bending
stiffness of slabz in the post-crack.lug range.

Various methods for treating the tension stiffening effect in slabs

have been examined elsewhere [3], including the use of average unloading
stress-strain relations for concrete after cracking. A simple,alternat-
ive procedure was developed, which relies on an adjustment of the stiff-
ness of the tensile steel in regions of cracked concrete. This method
has been found to give accurate results in economic solution times, and
has been adopted here. After cracking, the concrete is thus assumed not
to carry stress perpendicular to the crack direction, but an additional
stress in the steel accounts for the total internal tensile force carried
by the concrete between the cracks,conveniently lumped at the level of
the steel and oriented in the direction of the bars.

The reinforcing steel is assumed to be elastic-plastic in both

tension and compression, with stiffness only in the bar directions.
When the total strain in the direction of the reinforcement exceeds the
yield strain, the steel in this direction is assumed to have yielded.
If unloading of the steel occurs after yielding, the unloading branch of
the stress-strain curve has a slope equal to the elastic modulus of
steel.

The concrete creep strains are assumed to consist of three compon-

ents: a fully hardening and non-recoverable 'Dischinger' component,
{Ed(t)}; a non-hardening but fully recoverable viscoelastic component,
{Ev(t)}; and a component which takes into account non-linear effects at
high stress levels, {En(t)}:

+ + (3)

An increment {6Ec(tn)} in the creep strain vector, which occurs

during the small time increment 6t between the times tn and tn +
1
,

depends on the principal stresses, fopCtn)}, existing at the beginning

of the time increment. The vector of elastic, instantaneous, stress-
 4.

related strains in the principal stress directions attn is:

1 Ve (tn)
(4)
EcCtn) EcCtn)

The elastic modulus and the creep Poisson's ratio of concrete attn,
i.e. EcCtn) and VcCtn), are considered constant throughout the analysis.
The increments in the two linear components of creep are calculated
from the elastic instantaneous strains in the principal stress direct-
ions and from the creep properties of the concrete as follows:

{AEp(tn)} (5)

{AE~(tn)} (6)

The terms Ty and¢~ are constant creep parameters associated with

the viscoelastic component, and A¢d(tn) is an increment in a Dischinger-
type creep function ¢d(tn). Numerical values for all these quantities
are obtained from constant stress creep tests carried out in the low-
stress linear range where the total creep consists of only two compon-
ents:
(7)

Upon completion of all creep, i.e. at time infinity; we have:

(8)

Two creep functions ¢(t) and ¢d(t) are derived from the creep terms
Ec(t) and Ed(t) by dividing by the linear elastic component of instant-
aneous strain, Ee.I!,

(9)

EC(t) - EV(t) (10)

e;el
At time infinity, the final values are
e;C
(11)
..,* -
"' - e;el
*

,1,d _ E~
..,* - e;e.t
(12)
 s.

The constant parameter¢! is likewise defined:

e;V
V * (13)
¢* = eR.
e:
so that the following relations apply:

¢~ = a.d cp* (14)

ct>r = c1 ctd) • ¢* (15)

An exponential relation is used as an approximate representation

of viscoelastic creep under constant stress:

(16)

The creep function cpd(t) can now be evaluated from experimental

data:

(17)

The non-linear component of creep strain is assumed to be zero

whenever a principal stress is less than a prescribed critical value,
Oc· When a principal stress exceeds Oc, a power function of stress is
used to evaluate the non-linear creep in the principal stress directions:

{6e:~Ctn)} = [{6e:~Ctn)} + {6e:~(tn)}J . f(cr) (18)

and for 01 :;; ac f(cr) = 0 (19)

oc < (Jl f(cr) = a.TI ~a,- ac] n

(20)
ao - ac
o 1 is any principal stress greater than crc; a0 is the uniaxial compres-
sive strength of concrete; ¾ and n are open parameters to be determin-
ed from available test data.

With the material parameters ctd• ctn, n and Tv, together with the
creep function ¢(t), evaluated from test data, the increment in the
vector of creep strains developed in a typical concrete layer during a
small time increment in the principal stress directions can be calcul-
ated:
(21)

The increment in the creep strains in the global coordinate direct-

ions is obtained from the following strain transformation:
(22)
 6.

= [:: :: _:: l (23)

-2cs 2cs c 2 -s~

In Eq. 23 c = cos e; s = sin 0; and 0 is the angle between the
maximum principal stress direction and the global coordinate axis.

It is here assumed that the curve of shrinkage strain versus time

is affine with the total creep-time curve for constant stress. The
shrinkage strain developed during the small time increment at in each
global coordinate direction is expressed as

Ae:Sh(t) _ sh li~(t0 ) (24)

l..l n - e:* . ~*

e:;h is the value of shrinkage strain at time infinity.

3. FINITE ELEMENT FORMULATION

A layered 1 compatible 1 sixteen degree of freedom rectangular plate bend-
ing element has been adopted 1 with the layering details as proposed by
Wanchoo and May [8].
The prescribed displacement field for the element is expressed as
{w} = [A] • {a} (25)

where [A] is a matrix of polynomial terms in x and y and {a} is the

vector of undetermined displacement coefficients. The generalised
nodal deformations, consisting of displacements, slopes and twists, can
be expressed as

Qi= w = A . {a} = [G] . {a} (26)

aw/ax clA/ax
aw;ay clA/cly
a2 w/asay a2 A/axay
The generalised element displacements are given by
{q} = [C*] . {a} (27)

[C*] is the 16 x 16 'connectivity' matrix obtained by substituting the

local coordinates of the four element nodes into the 4 x 16 [G] matrix
of Eq. 26.
The vector of curvatures can be written as
 7.

{K} = [BJ . {a}

[BJ =
[ ,,32 1,,,
/ay2
2a 2;axay
l . [A]
(28)

(29)

By substituting Eq. 27 into Eq. 28, the vector of curvatures can

be expressed in terms of the generalised displacements:
{K} = [BJ . [C*J- 1 • {q} (30)
For the typical layered element of Fig. 1, the vector of instantan-
eous strains in the biaxial stress state, {£i}, at a depth z from the
top surface of the slab is given by

(31)
{£~} is the vector of instantaneous strains at the top reference surface.

By equating the sum of the in-plane forces of each layer to zero,

the assumption of zero membrane forces is built into the basic element
formulation:
m
l: llhk ([Ck] {K}) = 0 (32)
k=!

In Eq. 32, llhk = hk hk_ 1 ; zk = (h]< + hk_ 1 )/2; and [Ck] repres-
ents the appropriate stress-instantaneous strain relation for the k-th
layer and depends on the current material state and composition of the
layer.

By rearranging and simplifying Eq. 33, {£~} is expressed as

follows:
fF] . {K} (33)

(34)

The fFJ matrix can readily be formulated from the current constit-
utive relations for each layer and their respective thicknesses and
positions within the element.

The strain-displacement relationship for the k-th layer is obtained

by substituting Eqs. 30 and 33 into Eq. 31:

{£t} = [BtJ {q} (35)

zk
where [Bt] =h . [[ F] [BJ [C*]-1 - h . [BJ [C*r 1]

= [F*] [BJ [C*r 1 (36)

 8.

It will be noted that these equations are quite general and are
valid through all ranges of structural response within the limitations
of small deflection plate theory. The various material behaviour states
are conveniently accounted for by using the appropriate constitutive
relation for [Ck].
Equilibrium of the external loads {Qk} and the internal stresses
is expressed by the volume integral

{Qk} = fv LB~JT. {crk} . dV (37)

in which {crk} = [Ck] . {Et} (38)
and {Et} is the instantaneous component of total strain in the k-th
layer.
By substituting Eqs. 35 and 38 into Eq. 37, we obtain the stiffness
equation:

The stiffness matrix of the k-th layer is [Kk].

The stiffness of each layer is calculated and summed to obtain the
stiffness of the element and the individual element stiffnesses are ap-
propriately combined to obtain the global structural stiffness matrix.

When the stiffness of an element changes, due to a change in mater-

ial. state, the constitutive relation for the affected layer is approp-
riately modified. This results in a change of the stiffness of all
layers within the element due to the reformulation of the '[F] matrix as
defined by Eq. 34. The modified layer stiffnesses can be ·readily found
and summed to obtain the updated element stiffness which adequately re-
flects the current material state.
The choice of the top surface of the slab as a reference surface
(Fig. 1) is a convenient expedient in the element stiffness formulation
and allows the use of elements of different thicknesses to be handled
without complicating the analysis.

4. NON-LINEAR SOLUTION PROCEDURE

An incremental, piece-wise linear calculation procedure is used to model
the non-linear and inelastic response which is characteristic of rein-
forced concrete behaviour under both instantaneous and sustained load.
 9.

4.1 Instantaneous Response

The most important contribution to the non-linear instantaneous behav-

iour of reinforced concrete slabs is the tensile cracking of the concrete.
Cracking is treated analytically by increasing the external load in small
increments and progressively changing the material properties of the lay-
ers as they become cracked. All other material non-linearities are
treated in a similar manner.

ThP elastic structural stiffnor;r; matrix is first assembled and the

linear elastic displacements produced by the full instantaneous load are
determined. The state of stress in each layer is examined and the load
is scaled down to the value which produces first cracking in the most
highly stressed tensile layer. When cracking occurs, the concrete
stress normal to the crack drops to zero, thus creating an unbalanced
stress within the element. The stress-strain relation of the cracked
layer is modified and the stress loss is transformed into an equivalent
set of nodal forces, called transfer forces, {Qtr}, using the principal
of virtual work:

(40)

where {otr} is the vector of transferred stresses in the global coordin-

ate directions. The stiffness matrix is reformulated using the modified
constitutive relation and equilibrium is restored within the model by
applying the transfer forces to the structure using the 'initial stress'
technique, as suggested by Zienkiewicz et al [12]. Additional displace-
ments and associated stresses and strains are produced and the new state
of stress in each layer is examined for further changes in material
state. This iterative procedure continues until no further changes in
material behaviour occur and convergence to the non-linear state of the
slab at the current load level is obtained.

This process is repeated at each load level, as the load is in-

creased in small increments from first cracking to the full instantan-
eous load. If the initial sustained load is less than the full instant-
aneous load, the slab is appropriately unloaded to the sustained load
level and the time dependent solution procedure commences.

4.2 Time Dependent Response

Time dependent behaviour is calculated at a finite number of instants on

the time scale, with the external load level only allowed to vary at
 10.

these selected time instants.

During a typical time increment ~t, between the times tn and tn+i,
the increments in creep strain due to the stresses existing attn and
the shrinkage strain increments are calculated for each layer of each
element. These 'creep' increments in strain are initially converted
into 'relaxation' decrements in stress. The stress decrements are those
which are needed at time tn+i to restore the strains to their original
values at time tn. Horizontal equilibrium, temporarily lost due to the
application of the relaxation stresses, is restored by transforming
these stresses into nodal forces which, when applied, produce the addit-
ional displacements and stresses and strains which result from the time
effects in the interval ~t. It will be noted that a relaxation type
constitutive law could also be used directly in the first stage of this
calculation.

The resulting state of stress in each layer at tn+i is then examin-

ed for changes in local material behaviour. Checks are made to determ-
ine whether additional cracking has occurred, whether existing cracks
have closed, whether the steel has yielded and so on. Any changes in
material state within any layer are treated exactly as was described for
instantaneous response. With equilibrium and strain compatibility en-
sured throughout, the time dependent behaviour of the slab at the end of
the time increment, tn+i• is obtained when the appropriate constitutive
relation has been established for each element layer. Any changes in
the external load level at the next time instant are analysed and the
time dependent solution procedure for the next time increment commences.

The great advantage of this 'initial stress' relaxation approach is

that the usual time consuming search techniques required to establish
equilibrium and strain compatibility are eliminated, When the magnit-
udes of the creep and shrinkage strains in each layer have been determ-
ined and the material yield criteria are satisfied in all layers, the
time dependent state of the slab is obtained directly. This solution
procedure is general and does not rely on any one particular model of
the creep and shrinkage characteristics of the concrete.

5. NUMERICAL RESULTS

5.1 Example 1
No well documented, laboratory controlled long-term tests of two-way
 11.

reinforced concrete slabs have been found in the technical literature.

However, field measurements on three in-service reinforced concrete
slabs were presented by Taylor [7]. A flat slab floor in an open air
carpark, constructed in 1965 and monitored for a period of 3½ years,
has been selected for analysis.

Fig. 2 shows a plan of a typical internal panel of the slab, with

the finite element mesh used in the analysis. The slab was 241 mm thick
with 2440 x 2440 x 152 mm drop panels; the reinforcement layout and
elastic material properties assumed in the analysis are as detailed by
Taylor [7]. Transverse loading did not occur until 56 days after pour-
ing; the total sustained load was taken to be the self weight of the
slab, 5,75 kN/m 2 . Full short-term live loading, 2.87 kN/m 2 , was first
applied approximately 70 days after first loading. This' short-term load
±nfluences the long-term behaviour of the slab by causing additional
cracking and hence an overall reduction of the slab stiffness. Due to
the irreversible nature of cracking, the reduction of the slab stiffness
remains even after the live load is removed.

Unfortunately, the creep and shrinkage characteristics of the con-

crete were not recorded but a relatively high shrinkage level was antic-
ipated due to ineffective curing. Based on knowledge of the local con-
cretes at that time, the following time dependent material properties
were assumed:

~* = 2.5; ad= 0.7; ~ = 20; n = 4; E!h = 0.0006 with 30% of the

total shrinkage assumed to have occurred in the 56 days prior to first
loading.

Theoretical and experimental time-deflection curves at the midpanel

and the midpoint of the column centrelines are compared in Fig. 3. Des-
pite the lack of accurate experimental data, the theoretical time-
deflection curves are in good agreement with the actual slab response.
A closer fit to the experimental curves could obviously be obtained by
a suitable modification to the assumed shape of the constant stress
creep curve and the final values of creep and shrinkage strains.

The ratio of long-term to short-term deflection at the midpanel was

5.0, which is considerably higher than the usual code multipliers. Under
initial instantaneous load, the slab was essentially uncracked. However,
considerable cracking occurred with time due to the gradual increase in
tensile stresses caused by shrinkage. TI1e loss of stiffness due to time
 12.

dependent cracking, coupled with the direct effects of creep and shrink-
age, results in the high long-term to short-term deflection ratio.

5. 2 Example 2

To test further the accuracy of the proposed model, beams tested by

Washa and Fluck [11] have been analysed. Three beams, all 127 mm deep
and 305 mm wide with varying quantities of compression reinforcement,
were tested over a pP-rioct of 2½ years. Theoretical and experimental
time-deflection curves at midspan are compared in Fig. 4. The theoret-
ical curves were obtained using the creep and shrinkage data from the
companion loaded and unloaded test specimens. Again, agreement with
test data is seen to be good.

6. CONCLUSIONS

Computer simulation provides a rapid and inexpensive means for investig-

ating the effects of creep and shrinkage on the time-dependent behaviour
of reinforced concrete slabs. To provide spot checks on the accuracy of
the computer model, supplementary experimental evidence is required.
Unfortunately, well-documented experimental results of long-term slab
behaviour are scarce and much work is required in this area.

Within the limits of available experimental results, the non-linear

finite element model presented here accurately predicts both long-term
and short-term behaviour of rectangular reinforced concrete slab systems
under service loads. The idealised constitutive relations have repre-
sented well the various stages of material behaviour, while the element
layering technique has made possible a close study of the state of stress
as well as the orientation and height of cracks throughout the slab. The
direct relaxation procedure is a particularly convenient method for de-
termining strains and stresses at preselected time instants throughout
the load history.
While good agreement with available test data has been obtained, it
should be noted that the computer demands are relatively high. Economic
considerations therefore limit the use of the model as an analytic tool
for the practising engineer. However, as a research tool, the model
provides an excellent opportunity to study the long-term behaviour of
reinforced concrete slabs.

The slab simulation finite element program is at present being used

 13.

in a parametric study of the relative importance of the various factors

that influence long-term slab behaviour. The effects of creep, shrink-
age, load level and reinforcement quantities on long-term slab deflect-
ions, the redistribution of internal actions, the extent and orient-
ation of cracking and the ultimate strength of a variety of slab types
are being considered. The results of this study will be presented in a
second supplementary paper, with a view to the development of rational
recommendations for the calculation of service load behaviour of rein-
forced concrete slabs.

7. REFERENCES

[l J BRESLER, B., and SELNA, L.: "Analysis of Time-Dependent Behaviour

of Reinforced Concrete Structures", Paper No. 5, Symposium on Creep
of Concrete, ACI Publication SP-9, 1964.

L2] FERRY BORGES, J., and ARANTES E. OLIVEIRA, E.R.: "Non-linear

Analysis of Reinforced Concrete Structures", Publications,
International Association for Bridge and Structural Engineering,
Zurich, Vol. 23, 1963.

[3J GILBERT, R.I., and WARNER, R.F.: "Non-linear Analysis of Reinforced

Concrete Slabs with Tension Stiffening", UNICIV Report No. R-167,
University of New South Wales, Kensington, Australia, January, 1977.

[4] KUPFER, H., HILSDORF, H.K., and RUSCH, H.: "Behaviour of Concrete
under Biaxial Stresses", Journal ACI, Proc. , Vol. 66, No. 8, 1969.

[SJ LAI, K.L., and WARNER, R.F.: "Non-linear Behaviour of Indeterminate

Concrete Structures under Sustained Loading", UNICIV Report No. R-149,
University of New South Wales, Kensington, Australia, November, 1975.

[6] SCANLON, A.: "Time dependent Deflections of Reinforced Concrete

Slabs", Ph.D. Thesis, University of Alberta, Edmonton, Alberta,
Canada, 1971.

[7] TAYLOR, P.J.: "The Initial and Long-Term Deflections of Reinforced

Concrete Flat Slabs and Plates", M.E. Thesis, University of New
South Wales, Kensington, Australia, 1971.

[8] WANCHOO, M.K., and MAY, G.W.: "Cracking Analysis of Reinforced

Concrete Plates", Journal of the Structural Division, ASCE, Vol.1O1,
No. STl, 1975.
 14.

[9] WARNER, R.F.: "Non-linear Creep in Concrete Columns", Final

Report, Madrid Creep Symposium 1970, International Association
for Bridge and Structural Engineering.

[10] WARNER, R.F., and LAMBERT, J.H.: "Moment-Curvature-Time-Relations

for Reinforced Concrete Beams", Publications, International
Association for Bridge and Structural Engineering, Zurich, Vol. 34,
1974.

r11J WASHA, G.W., and FLUCK, P.G.: "Effect of Compressive Reinforce-

ment on the Plastic Flow of Reinforced Concrete Beams", Journal
ACI, Proc., Vol. 49, 1952.
,12] ZIENKIEWICS, 0.C., VALLIAPPAN, S., and KING, I.P.: "Elasto-Plastic
Solutions of Engineering Problems, Initial Stress, Finite Element
Approach", International Journal for Numerical Method in Engineer-
ing, Vol. 1, 1969.
 top raference---
surfoce
X

layczr
layer

Fig. 1. Layered Finite Element Model

 t
I
,-----, I
r----,
I I I I
I
I II I
I
201 I I
I
I I I
L----.J - - .J

E
-w-J------9-l-----~,o-- ~ I{)

r .... -..~-,
I I r.._..~--.-.,
I
: •
I
I
I
l '
•
I
I
I
:
L----...1
I
l t..-----'
I

8610 mm.

Fig.2. Typical Internal Panel of Flat Slab

Tested by Taylor.[71.
 Age from Stripping ( years)
0 1 2 3
0

-E
E
--- 5
...,._. __
t1xpczrimental
C th"ore:ticol
0
.....
u
.-...cu
GJ 10

0
'ii 1!5
C ~
.........
C,
0.
......... ........
......... .....
·-
"C
~
20
-
25-----~~------------------'
ii/;' .I
J
ii" i
. I

0 ',11 2 3
-E 0

E
- experimental
0
C

.....
!5 - -- ...,__ thaoratlcal
uQJ
....cu 10
10
C
QI --
- C

---
- 15
Q.I
L..

C
QJ
u
C 20
E
::,
0
u
25

Fig. 3 Time-Deflection Curves Structure 3

Flat Slob tested by Taylor. (7).
 160
.-,
E A SC-- 0'
-E
C
-- - - - - - --
.2 120
....
0

-
0
GJ
GJ
- __ ...

C
g_
c,n
-0
·-2".
experimental
40
theorflticol

100 200 300 400 900

Time (days)

Fig. 4. Time-Deflection Curves for Beams Tested by

Washa and Fluck, (11).