A Study of Polysulfide Sealants for

Joints in Bridges
JOHN P. COOK
Assistant Professor of Civil Engineering, Rensselaer Polytechnic Institute

The polysulfide sealants used in expansion joints in bridges are commonly

assumed to be perfectly elastic materials. This paper shows that these
sealants are acutally viscoelastic and exhibit the interrelated phenomena
of creep and stress relaxation. Curves are included for modulus of elas-
ticity, creep, and stress relaxation, with shape factor, Shore hardness
and temperature as parameters. Sample solutions are shown for the
stresses in the sealants under various loading conditions, including tension
plus shear and compression plus shear. A stress relaxation equation is
derived using a curve-fitting technique and the stress relaxation relation-
ship is verified by standard laboratory methods and by the method of photo-
elasticity.

•THE ENGINEER or scientist today, in dealing with the properties and behavior of
solid materials, generally has at his command theories which, if not perfect, are at
least acceptable and consistent. The engineer who worked with rubber-like materials
had until recently no such assurance. It remained for the chemist and physicist cor-
rectly to postulate certain properties of highly extensible molecules before consistent
theories were developed.
Although much of the significant work in this field has been done in recent years,
the problem dates back almost 100 years and shows three distinct roots:
1. The general problem of flow in solid materials, beginning with the work of J. C.
Maxwell;
2. Research in the field of rubber-like materials by such investigators as Treloar,
Kuhn, Guth, Rivlin, and Tobolsky, dating back roughly 25 years; and
3. Specific works in the field of joint sealants, notably that by Tons (.!).
Recent papers have shown a very intimate relation between the first two of these
roots in the effort to explain the behavior of various polymers, including the polysulfide
rubbers. Outside of Tons' work, very little effort has been made to apply this earlier
work to the specific problem of expansion joints.
A study of this history brings two cases into focus, the perfectly elastic sealant and
the material which flows with time. Since the polysulfide sealant can be formulated
with a wide range of properties, both of these cases should be considered.
A joint sealant may fail mechanically in any one of the several ways shown in
Figure 1:
1. The adhesive failure is a loss of bond between the sealant and the joint wall
caused by a tensile load. It may start as a small localized failure and then peel rather
rapidly under the action of stress.
2. The cohesive failure is a tearing of the sealant material, also under tension.
3. The extrusion failure occurs under the combined action of compression and
traffic. The material under compression is extruded above the roadway surface and
then folded and flattened under the action of traffic.

Paper sponsored by Committee on Sealants and Fillers for Joints and Cracks in Pavements
and presented at the 43rd Annual Meeting.
11
12

--- (a) (b)

/.J/li'EC T/tJ/1/ tJF Ttf'AFF/C

( c) (d)

(e)

Figure 1. Types of joint failures: (a) adhesion, (b) cohesion, (c) spalling, (d) ex-
trusion, and (e) intrusion.

4. The spalling failure is not, strictly speaking, a sealant failure, but its effect is
just as destructive. In this case, the concrete adjacent to the joint interface spalls,
generally under the action of heavy traffic, and the salt solution leaks past the sealant.
The spalling failure is generally localized in the primary traffic lane but it breaks the
continuity of the joint and quite often leads to a general peel-type adhesive failure.
5. The intrusion failure occurs under the combined action of extension and traffic.
The sealant under tension necks down, forming a pocket, which fills with dirt. As the
joint closes, this dirt is trapped within the sealer mass. This weak spot usually leads
to a cohesive failure during the next extension cycle.
These failures are described in the literature. However, the present work demon-
strates the extent of stress relaxation in the sealant, so two more failures attributable
in large part to flow and stress relaxation are shown in Figure 2.
In each of the two cases shown in Figure 2, the sealant is loaded in normal fashion
(either tension or compression) and then held at constant deformation until some degree
of stress relaxation has taken place. On the next return cycle it begins deforming from
this new shape and does not return to the original rectangle. In each case, a failure
is imminent with additional cycling.
All these potential failures are accelerated by the effects of aging and weathering.
However, the manufacturers of the various sealant materials are constantly striving to
improve these products in their resistance to weather, salts, acids and solvents, so
the effects of weathering will recieve no treatment here.
Each of the failure conditions mentioned presents a separate problem and the solu-
tions may vary quite markedly, depending on the field conditions. The attempt being
 13

(a)
f .. f --- ---

Figure 2.
- --- (b)
Change in sealant shape due to flow: (a) viscous tension-compression effect,
and (b) viscous compression-tension effect.

made here is not to provide field solutions for all these problems but to evaluate the
polysulfide sealant material; in the course of this investigation, the solutions of some
of these cases present themselves.

PURPOSE OF STUDY
The purpose of this study was to consider the polysulfide sealant first as a perfectly
elastic material and then as a flowing solid and to determine what stresses exist under
load in a bridge joint. Inherent in this search, of course, was the determination of
modulus and, what is more important, of the proper relationship between stress, modu-
lus and strain, with time, temperature, shape factor and extensibility as the necessary
parameters. Experimental work in this study included those tests which were neces-
sary to establish the parametric relationships.
Appendix B of this paper includes sample solutions for the stresses in the sealant
as follows: (a) considering the sealant as a perfectly elastic material--tension (or
compression) alone, tension plus shear, and tension plus end rotation of the structure
as caused by midspan deflection; and (b) considering the sealant as a flowing solid--
tension (or compression) as caused by an imposed strain progTam corresponding to
bridge movement.

THEORY

Sealant as a Perfectly Elastic Material

Many formulas have been presented in the attempt to show the proper relationship
between stress, modulus and strain. Some have used the classical approach through
the theory of rubber elasticity, and others have used an empirical approach. Treloar
(2) has summed up the work of previous investigators and, working with an acceptable
1nolecular model, has developed the kinetic theory expression in terms of extension
ratios and the familiar shear modulus. The approach, which is made on a statistical
basis using the entropy method, is explainable in terms of a mechanical model and
lends itself to extension to the case in which the flow properties of the material are
considered.
The internal work of deformation within a joint sealant according to Treloar' s
Kinetic Theory is expressed as:

( 1)
14

in which G is shear modulus and >.. is ratio of extended length to the original dimension,
in each of three directions. The external work done by the applied forces is

dW (2)

Differentiating the internal work expression and equating it to the external work ex-
pression yields a set of simultaneous equations:

(3)

These are the general stress-strain relations which hold within the range of the Gaussian
probability function on which Treloar' s statistical derivation is based. Practically
speaking, these formulas are valid up to approximately 200 percent extension in the
sealant.
These equations simplify when applied to the joint sealant problem because the joint
extends across two or three lanes of traffic and, hence, has a length in this direction
of 30 to 40 ft as compared to cross-sectional dimensions of approximately 1 in. With
no loss of generality the extension ratio in this transverse direction can always be con-
sidered as unity.
Since the sealant undergoes no change in volume the following relationship must hold
true:

( 4)

Therefore, for a sealant under simple elongation the extension ratios are >..1 = >.., >.. 2 =
1/>.., and A3 = 1.
Since the only external force acting on the sealant is in the tensile (>..1) direction,
me srress reauces to:

i <IW 17
(5)
== dA vJ
Sealant as a Flowing Solid
In macroscopic terms, the viscoelastic behavior of polymers is usually separated
into three components, instantaneous elasticity, delayed elasticity and viscous flow.
Billmeyer (~ associates these properties with molecular structure as follows:
1. Instantaneous elasticity--stretching of the primary valence bonds and straighten-
ing of the bond angles in the main polymer chain (a reversible action);
2. Delayed elasticity--reversible uncoiling of the polymer chains and orienting them
in the direction of the stress; and
3. Viscous flow-irreversible slipping of the chains past one another.
These actions are, of course, stated in completely general terms and the interrela-
tion between them is a function of other parameters such as temperature.
To the laboratory observer, the essential difference between the elastic and the
viscoelastic body is that the latter shows the two interrelated phenomena of creep and
stress relaxation. Both of these phenomena are capable of explanation in terms of
molecular structure. The mechanism of creep, which is demonstrated later, becomes
apparent when a creep curve is shown together with the model movements which repre-
sent the behavior.
 15
The method of analogous mechanical models has been selected for use in this pres-
ent work because it defines the behavior of the material in a manner which can be
logically explained in terms of the molecular structure of the polymer. The method is
said to be an approximation but Billmeyer (3) and many others feel that because of its
sound basis in molecular theory, it offers the best solution currently available.
The model method consists of the identification of molecular behavior with certain
types of response to an applied stress. These responses are then represented by simple
mechanical models which can be explained in mathematical terms. An ideally elastic
material is represented by a spring, which obeys Hooke's law, that is, stress pro-
portional to strain (Fig. 3). A completely viscous response is represented by the hy-
draulic dash pot (Fig. 4). For this element, stress is proportional to strain rate. The
third basic element, the friction unit, shows no movement until stress reaches some
limiting value represented as a friction force (Fig. 5).
Certain basic combinations of these units occur so often that these combinations are
considered almost as elements. The series combination of a spring and a dashpot is
known as a Maxwell model (Fig. 6) and represents a material which shows an instan-
taneous elasticity and a straight-line creep under load and some amount of flow or
permanent set on removal of the load. The parallel combination of a spring and a
dashpot is known as a Kelvin (or Voigt) unit (Fig. 7), which represents a material dis-
playing a damped or delayed elasticity under load but no permanent set on removal of
the load.

Figure 3. Model for ideally elastic material .

!--3]---
Figure 4. Model for viscous flow .

v< l7o /VO /'?OVEHE/VT

9;;;;» n,L '"'""* ..J.,., . , , ,:

<::r> 'To YIE'-.0
r;r-

Figure 5. Friction model .

Figure 6. Maxwell model.

16

Figure 7. Kelvin model .

Figure 8. General creep curve .

To describe a material with a model system we must know which basic elements
are present and what combinations exist. A creep curve (strain vs time at constant
load) for the material in question will generally establish the presence of springs and
dashpots in a system. The friction element can be identified by plotting various load-
increment curves of elongation vs time.
The creep curve for the polysulfide rubbers, shown in Figure 8, duplicates the
curve shown in many texts as the generalized curve for all rubbery polymers. Com-
parison of this curve with the creep curves of the Maxwell and Kelvin units immediately
suggests an additive combination of the Maxwell and Kelvin units with a four-element
model to describe material behavior. The presence of a yield point (friction unit) is
r111Prl nnt hv PrPPn tPsts ::it vPrv low stress levels. (One such laboratorv snecimen
was still d~formi~g after 6 mo~ths at only 1. 3 psi). · ·
For clarity the creep curve is reproduced in Figure 9 to a distorted scale and the
::ictions of the model elements associated with each part of the curve are shown below
the curve. This curve really represents a family of curves with different magnitudes
corresponding to different Shore hardness values and shape factors. The stress levels
used are quite low (5 to 10 psi). On application of load, there is an instantaneous de-
formation Eo (a spring element) and then a curved section which is delayed elasticity
(a Kelvin unit), a straight-line flow with time (a dashpot) up to the unload point. On
removal of the load, these actions appear in reverse: an elastic snapback(= E0 ), a
delayed recovery and some permanent set.
The four-element model is the simplest model to describe the material behavior,
but many authors, including Billmeyer (3), Treloar (2) and Alfrey (4), state that this
is an oversimplification and should be used with caution, because itapplies only to a
material with a single stress relaxation time. The polymers, in general, exhibit
stress relaxation but the details of the stress relaxation process depend on the multi-
plicity of ways in which the molecules can regain their most probable configurations.
Consequently, there are so many modes of relaxation that the spectrum of relaxation
times can be approximated by a distribution function. If there were but a single re-
laxation process, the rate of stress decay would be a simple exponential function.
However, Stern and Tobolsky (5) in a specialized study of polysulfides have established
that this particular polymer differs from the rest of the rubbery polymer group in that
it does exhibit a single stress relaxation time. This contribution, together with the
creep curve, should establish the four-element model as an adequate approximation to
 17
UNLO AD

L!J

(b) l (d)

Figure 9. Model movements: (a) spring G, deformed, (b) Kelvin unit and Th deforming,
(c) ~3 still deforming, (d) spring G, recovered, (e) Kelvin unit recovering, and (f) ~3
remains deformed.

describe the action of the viscoelastic

joint sealant under load. Strictly speaking,
in a material with a true single relaxation
time, no creep recovery would be expected
at the termination of a stress relaxation
experiment. Actually, a small amount of
creep recovery was noticed but was con-
sidered not sufficient to invalidate the ap-
proximation.
The basic four-element model is ex-
cellent for demonstrating the three char-
acteristics of polymer behavior, i.e.,
instantaneous elasticity, delay elasticity,
and flow. This model is conveniently
Figure lO. Equivalent model (Model B) . solved for strain when an imposed stress
on the system is specified. However, for
the more practical case of solving for
stress under an imposed strain program,
it is more convenient to substitute an
equivalent model than to solve the basic model. The equivalent model shown in Figure
10 is related to the basic model through a set of equivalency equations included in Ap-
pendix B with sample problem solutions.
The basic equation for stress taken from the model is
18

in which
B strain rate,
viscous constants taken from the creep curve,
elastic constants taken from the creep curve,
initial displacement,
relaxation times (defined as the time necessary for the stress in the
material to deca y to 1/ e times the initial stress), and
t = time.
A c omplete development of the stress equation has been given previously (~.

EXPERIMENTAL WORK
The materials used in this work were all two-component cold-poured polysulfide
rubbers . All materials were commercial grade, in quart or gallon containers, and
we r e donated by the manufacturers. Polysulfide joint sealants are marketed with values
of Shore hardness ranging from 5 to 50. Since New York State currently specifies a
Shore range from 10 to 15 for bridges and pavement joints, only those materials in and
around this range were selected for test. The sealants are available in both gray and
black and both types were used without distinction. The sealants were grouped into
three categories by hardness: the soft range (Shore hardness 3 to 8), the medium
range (8 to 16), and the hard range (16 to 22). One set of modulus tests used a material
with a hardness of 30. All hardness measurements were Shore A.
In all, eight tests were conducted. Two of these were to confirm the assumptions
made in Tons' paper, i.e., the constant volume assumption and the parabolic neckdown
a ssumption. The other tests were of cycling tension and compression, creep, modulus
of elasticity, stress relaxation, photoelastic correlation, and peel strength.
Flexible sealants for joints have many applications in industry which necessitate a
wide range of sizes and shapes for the sealant. Since the present work was concerned
with t he expans ion joi nts in bridges , only those shapes which might be a pplicable to this
specific use were included. In all, eight size s were included in each test, ½,- and 1-in.
1 1
wirlP. i:;r,P.<'i mP.ns in j oini' rl e!)th s of / ~, 1: 1 / ~ and 2 in. Each test was performed in each
of the Shore hardness ranges and, in most cases, was repeated with materials from at
least two manufacturers in each range. Thus, the creep test, for instance, would in-
volve the testing of 16 specimens in each hardness range or 48 specimens in all.

Description of Experiments
The constant volume assumption was checked in a triaxial compression tester in the
Rensselaer Soil Mechanics Laboratory. The apparatus is standard in soils work and
will not be described in great detail here. It consists of a hollow plastic cylinder filled
with water into which the specimen is placed. Loading is accomplished by a 1-in. di-
ameter plunger extending through the top cover plate. A graduated tube to show differ-
ences in water level is used as a reading device.
The parabolic neckdown assumption was checked by using an Ames dial gage as a
depth indicator. A sealant specimen mounted on concrete blocks was placed in the
cycling machine and extended . The Ames gage was mounted in a wooden block with the
plunger extending downward. The block was then placed on top of the concrete test
blocks across the extended joint. The amount of neckdown of the extended sealant was
read directly from the dial gage. The cycling tension and compression test was per-
formed in a machine designed and built specially for the purpose by R. J. Schutz.
Specimens for this test were prepared with concrete test blocks. Lengths of the block
a nd, consequently, of the specimen were 3 in. and the joint depth was variable up to
2 in.
The creep test, which furnishes the constants for the stress equation, was accom-
plished by static loading. Specimens, 6 in. long, were formed between sections of 2
 19

by 2-in. aluminum angles, 7½ in. long·. Creep curves, which are plots of strain vs
time at constant load, were plotted for a range of shapes, loads and hardness values.
The curves are included in Appendix A.
Specimens for the modulus of elasticity test were identical with the creep specimens.
The test was performed in a 30, 000-lb capacity pendulum-type Riehle testing machine.
The modulus curves plotted by the machine in units of load vs deformation and with a
chang·e in scale only are also included in Appendix A. Strain rate for the modulus test
was 0. 1 in. /min.
The load-deformation (or stress--strain) curves for these materials were, of course,
nonlinear functions and the value of modulus used for computation was the value of
stress at 100 percent strain.
The stress relaxation test yielded a plot of stress vs time at constant strain. A con-
stant strain of 50 percent was used throughout the work. The test was performed on
the same pendulum machine as the modulus test.
The photoelastic work consisted of a verification of the stress relaxation time as
determined on the pendulum machine. Specimens were 1 by 1 in. in cross-section and
6 in. long. The material was a translucent poly sulfide epoxy with the manufacturer's
formulation adjusted to vary the Shore hardness. Creep curves were plotted and then
identical specimens were tested on the pendulum machine and in the polariscope (Ap-
pendix A, Figs. 29 and 33).
The peel test used was somewhat similar to the ASTM Peel Test. The sealant was
spread on 1- by 6-in. rigid substrate. A 1-in. wide strip of heavy canvas was im-
pregnated with sealant and then placed on the substrate and rolled until the depth of
sealant and canvas was 1/1 6in. The canvas strip was then folded back through 180" and
the canvas and rigid strip were placed in opposite jaws of the pendulum machine. The
rate of peel used was 3 in. /min and maximum load was read directly from the machine.

DISCUSSION
The sample problems for the elastic sealant (Appendix B) illustrate that shear, when
combined with extension, may be neglected for all practical purposes. However, shear
in combination with compression of the sealant produces a dangerous condition. Mid-
span deflection of a bridge structure is shown to produce practically no shear but does
increase the tensile strains in the sealant by as much as 25 percent.
Leadermann (7) describes time as the main character in his work on creep of poly-
mers. This same description might be paraphrased by stating that in dealing with joint
sealants, time has been the forgotten parameter. The sample problems shown in Ap-
pendix B indicate that when time, or better stated, when stress relaxation enters the
picture, stresses in the sealant drop to almost negligible values. The problem then
becomes one of shape and recovery rather than stress.
In addition to the tests mentioned previously, curves included in Appendix A show
the variation of hardness with temperature and the variation of modulus of elasticity
with hardness. All experimental work in the present investigation was done at room
temperature and these curves mentioned are an attempt to extrapolate these results to
other temperature values. For example, if it is desired to use a sealant which has a
hardness of 5 at room temperature for some application at -30 degrees its behavior
can be predicted. At this temperature, the A-5 material has a hardness of 20. The
creep, modulus of elasticity and stress relaxation values can be found from the re-
spective curves for the A-20 material. This extrapolation method is admittedly a
limitation on the present study but it represents at least a starting point for predicta-
bility.
The creep curves included afford a measure of creep predictability. The higher
modulus materials show a much better compliance with the creep curve shape than the
softer materials. All curves are shown as smooth curves, but the softer materials
are quite erratic in behavior, and the curves represent the average of many retests
and the elimination of some obviously faulty values.
It must be remembered that when a structure moves, the forces exerted on the
sealant are quite high and the sealant must comply with the imposed strain program.
20

It may seem ridiculous to state that the sealant never moves the structure, but at this
time many spalling failures are blamed on the sealant because it pulled the concrete
apart. The values of creep, modulus and stress relaxation included here should help
to dispel this notion. A spalling failure is simply due to poor concrete, not a strong
sealant.
The creep behavior does vary with shape. Under equal stresses, a shallow seal will
creep more than a deep one. This result is not surprising when compared to the para-
bolic length of the sealant. It is also important that within the 200 percent strain limi-
tation the creep curve will superimpose when subjected to a variation in stress. More
research is needed to determine whether the new polysulfide formulations show any
tendency towards work hardening under repeating cycling.
The modulus curves are purposely left in terms of load vs elongation to make them
easily readable. When plotted as nominal stress-strain curves (stress based on original
cross-sectional dimensions), they are virtually coincident which shows that modulus
values are practically independent of shape. The modulus value by definition is stress
at 100 percent strain. The modulus values are plotted for a strain rate of 0. 1 in. / min.
They will vary with strain rate because the steep gradient of the modulus curve at the
beginning of the test corresponds to the steep downward gradient of the stress relaxa-
tion curve for very short times.
No values of ultimate cohesive strength are reported because of the fear of erroneous
interpretation. Every sealant tested was extended well beyond 200 percent strain, and
some as far as 1, 000 percent, before failure. Above 200 percent strain neckdown is no
longer parabolic and cross-sectional area reduces to almost zero. Consequently, any
stress (which is load divided by area) would show values so high that they appear
ridiculous.
The stress relaxation curves are perhaps the main contribution that this paper has
to offer. The curves follow a generally exponential shape. Tobolsky (5) has shown that
the polysulfides in general follow the relation: -

f = f 0 exp -k 't (7)

1
He states that the curve is, in general, steeper than exponential and relates k to the
rate of bond rupture. His method for fin.ding k' is a curve-fitting technique. The ex-
perimental curves offered here are also steeper than exponential and a curve-fitting
scheme is also used. The equation which fits the curve is really a family of equations
of the form:

f = f 0 exp (t/ r)n (8)

in which n varies with T. For low values of T (about 1 hr), n approaches unity. For
higher values of T, n varies between 0. 5 and 1. 0. The highest value of T obtained in
this work was 6 hr. This was for a high modulus sealant. The middle range (A-12)
sealants showed a relaxation time of 3 hr. The erratic soft-range materials showed
great variation in values and also wide variation for single points on each curve.
Further study of stress relaxation should lead to a definitive relationship by which T
and n could be predicted for different sealants.
The important fact to be recognized with polysulfide sealants is the order of magni-
tude of the stress relaxation time. In just a few hours after an imposed strain, the
stress has relaxed to less than half of its initial value. The fact that the relaxation is
steeper than exponential is also noteworthy. In the event of an imposed strain ( E0 ) of
large value, the sealant can actually relieve the stress by as much as one-third within
the fir st 20 min.
The peel test used was, to some extent, new and needs refinement. The test speci-
mens, 1 in. wide by 6 in. in gage length, were 1/16 in. thick. Eleven of the twelve
specimens tested did not actually fail by peeling away clean from the substrate but by
 21

tearing of the sealant. The average value of tearing strength was 42. 5 lb for the 1-in.
wide strip at a testing machine .r ate of 3 in. / min. This value is very close to the values
of peel strength and tear resistance which are available from the various manufacturers.
Further work on the peel test should include an investigation of the effect of specimen
thickness and also a long-time test at low loads, somewhat after the fashion of the
Bike rm an ( 8) test.
The results of the photoelastic work have to be classed as both good and poor. The
original intention was to find a birefringent material with the same properties as the
actual sealant to make a comparative study. Finally, several formulations of epoxy-
polysulfide combinations were developed in the desired hardness ranges, but these
showed a shorter relaxation time, a faster creep rate and more permanent set than the
actual sealants. Differences were expected, of course, but it was hoped that a pro-
portionality constant could be derived such that actual stresses in a sealant could be
checked by photoelasticity. However, the characteristics of the photoelastic material
are somewhat contradictory when compared to sealant behavior. More research is
needed to determine the factor which can be applied to the two materials so that a quan-
titative prediction can be made.
The success achieved with the photoelastic method was the development of several
materials which exhibit stress relaxation according to an exponential law. Identical
specimens of material were tested in the Riehle machine and the polariscope; both of
these tests show very good agreement with the theoretical curve. Quantitative values
for the photoelastic curve were obtained by counting the total number of stress fringes
which passed a given point during the rapid loading process and equating this fringe
order to the maximum load value shown by the testing machine. One end of the speci-
men was blocked at a constant 1 in. width and the other end of the specimen was strained
to 25 percent. This method of obtaining clear fringes assumes a linearity of stress in
the specimen. This linearity should be checked by strain gaging if the method is to be
used further.
Since so much emphasis has been• placed on shape, some new shapes were tested
to find one which would not extrude above the roadway surface. Two such shapes are
shown in Figure 11. The trapezoidal joint would help to eliminate spalling failures,
but it also makes a wider joint and, consequently, would cause a car to make an ob-
jectionable thump when riding over the joint. When formed with a flush top, this joint
extrudes above the roadway surface, but not in a parabolic shape, under compression
loading. When formed with a slight depression in the surface, the sealant no longer
extrudes under compression, but the poor riding quality is still objectionable.
The hollow shape shown in Figure 11 is actually formed by filling the joint to half
depth, inserting a dumbbell-shaped strip of soft urethane foam, and then filling the
joint. For wide joints ( 1 in. and wider) such as might be encountered in bridge work,
this shape has possibilities. However, because it is difficult to form and the foam has
a tendency to float, the few specimens tested showed rather erratic behavior. These
problems might be solved by experiment and experience.
The ordinary rectangular shape formed with a depression in the top also offers
possibilities. This shape does not extrude above the roadway for shallow joints.
Deeper joints do tend to bulge upward under compressive loads.

RECOMMENDATIONS
Although the polysulfide sealants have many properties which make them desirable
as sealants, such as high bond strength, extensibility, and excellent short-term mem-

Figure 11. Experimental joint shapes .

22

ory, the sealant problem is not yet solved. There are many aspects of sealant be-
havior that have not yet been investigated. Among the many questions still to be
answered are the following:
1. What effect does work hardening have on the physical properties of a sealant?
2. Does the aging and weathering of a sealant affect its mechanical behavior?
3. What is the response of a sealant in a skewed expansion joint?
A great deal of additional research is necessary so that these and other vital questions
may be answered, not only in terms of the polysulfides, but also with the other avail-
able sealant materials.

ACKNOWLEDGMENTS
The complete study from which this paper was taken was presented as a Doctoral
Dissertation at Rensselaer Polytechnic Institute. The author wishes to extend his
thanks to Professors Egons Tons, J. F. Throop, J. T. Watkins, J. Hollingsworth,
R. H. Trathen and R. M. Lewis for their encouragement and direction.
The joint sealant industry on the whole has been very helpful. Particular thanks
are extended to Raymond Schutz, Sika Chemical Corp.; C. A. Peters, Allied Ma-
terials Corp.; Norbert Hochreiter, H. B. Fuller Co.; Joseph Amstock, Products
Research Corp.; Aaron Kaplan, Lewis Asphalt Engineering Co.; H. V. Wittenwyler,
Shell Chemical Co.; and A. Shuman, Polarizing Instrument Co.
The author's deepest debt is to Harold B. Britton of the New York State Department
of Public Works, without whose constant faith, encouragement and material assistance
the project never could have been undertaken.

REFERENCES
1. Tons, E. A Theoretical Approach to Design of a Road Joint Seal. Highway
Research Board Bull. 229, pp. 20-53, 1959.
2. Treloar, L. R. G. The Physics of Rubber Elasticity. Oxford Univ. Press, 1958.
3. Billmeyer, F. W. Textbook of Polymer Chemistry. New York, Inter science
Publishers, 1957.
4. Alfrey, T. Mechanical Behavior of High Polymers. New York, Interscience
... .
~ ... ,.,, ,.._~

.t'UUU::mer::;, .Ll:1':tO.
5. Stern, M. D., and Tobolsky, A. V. Stress Relaxation in Polysulfide Rubbers.
J. Chem. Phys., Vol. 14, No. 93, 1946.
6. Cook, John P. A Study oI the Adi on uI Ela::;Lic J uinl Seal Male rial::; i11 lhe
Expansion Joints of Bridge Decks. Doctoral Diss. Rensselaer Polytechnic
Inst., Dept. of Civil Eng., 1963.
7. Leadermann, H. Elastic and Creep Properties of Filamentous Materials and
Other High Polymers. Washington, D. C., Textile Foundation, 1943.
8. Bikerman, J. J., and Yap, W. Rheology of Peeling in a Newtonian Liquid.
Trans. Soc. of Rheology, Vol. 2, pp. 9-21, 1958.
 23

Appendix A
GRAPHICAL RESULTS

soo
500

400

400

300
.:JOO
~

c::,
JOINT DEPTH 2111.
JOINT DEfJTH 2 in.
"'
c::, eOO JOINT DEPTH 1t ,n
-J
JOINT 0EfJTH IYz in.
/00 ){)/NT DEPTH I /n .
JOINT DEPTH I in
/00 JOINT DEPT/-/ /2 in.
JOINT DEPTH Yz in.
() OS /.0 1.5 2.0 ,?S ~o
0 ,c___.,_ __.___ _._____,.__~ _ _,
ELON<jATION (in.)
lO 2.0 30
ELONGATION (in J Fibure 13. Load elongation curves; Shore
A-5, 1-in. wide joint.

Figure 12. Load elongation curves; Shore

A-5, ½-in. wide joint.

500

500
JOINT DEPTH 2 in.
400
400
JOINT DEPTH 11/z in .
.:JOO
JOINT DEPT/I Z ,n.
300
~ JOINT DEPTH I in , JOINT DEPTH I~ in.
~
~
'--
c::,
200 C)
'< 200
c::,
-J
JOINT DEPTH Yi in. "
C)
')
JOI/VT DEPTH /1~.

/00 /00 JOINT DEPTH Ye in.

0
/0 z.o 3.0 0 0.5 /,0 1.5 20 Z.5 J.0

ELONGATION fin .) ELON<;ATION On)

Figure 14. Load e longation curves; Shore Figure 15. Load elongation curves; Shore
A-12, ½-in. wide joint. A-12, 1-in. wide joint.
24

JOINT DEPTH 2 in . 500

500
JOINT DEPTH 2 in
400
400 JOINT DEPTH 1¥2 in
JOINT DEPTH I Y2 in.
300
~
600 JOINT DEPTH I in
:"! c:,

c:, JOINT DEPTH I in . "'

C)
......
'<
C)
200
JOINT DEPTH Vi in .
JOINT DEPTH 112 in

/.0 2.0 3.0

30 ELONGATION (tn)

ELONGATION (1n)
Figure 17. Load elongation curves; Shore
Figure 16 . Load elongation curves; Shore A-20, 1-in. wide joint.
A-20, ½-in. wide joint.

5QO
I JOINT DEPTH Zin
/ J O I N T DEPTH /~i/7.
500[
400 / /
JOINT DEPTH 2 in

JO'Nl o,p,a I Yz in.

JOINT DEPTH /Ir,,.

~ 300 f2 .J0//1/T OE.fin-! J i'n
,:::: '-

Cl C,
'S;
Q zro
JOINT DEPTH ::f In "'C)
__,
'-I
JOINT DEPTH 1/2 in.

/00

o as 10 1., z.o zs 3o
3.0
EL ON'7AT/ON {in.)
ELONGATION ( in.)

Figure 18. Load elongation curves; Shore Figure 19. Load elongat ion curves; Shore
A-30, ½-in. wide joint . A-30, 1-in. wide joint.
 25

5 00 ?
(::_
lo-
~ ii
"100 ~
~ /3

300
"t
~ 1.2
'<:)

'
'-- ~
~ 20/J ~ /./

"
~
'J
c:i
~ /0
/00 .3 5 T .9 II lam. 3 s 7 9 II IA"'- 3
CLOCK TIME

Figure 21. Rate of bridge movement; normal

joint width lin. at 76 F, 100-ft bridge
0 050 JOO /.5'0 z.oo span, 22 to 66 F temperature variation.
DE.FOA'MAT/ON (117.)

Figure 20 . Load-deformation c urve-com-

p ress i on; Shore A-12, 3½-in. diamet er, 4-
in. high specimen.

JOINT DEPTH
Ve in. °JOINT DEPTH
I in.

Q6

Q4
.!:,

-~'
'-'
0.2

~ 0
~
>-- /,0 -
0)
JOINT DEPTH JOINT DEPTH
1/lz in. 2 in.
0.8

Q2
a~_.__ _.__ _,__.___.__ _,
2 ,3 4 5 6 2 3 4 5 G
TIME (HOURS)

Figure 22. Creep curves; Shore A-5, ½-in. wide joint, 8-psi constant load .
26

JOINT DEPTH

'l
JOINT DEPTH
1/2 in. I in.
/,0

0.8
06
,..._ 04
~
.s 02

::' 0
«
c,:
>--- I.O JOINT DEPTH
'1) JOINT DEPTH
I /le in. 2 in
0.8
0.6

0.4
0.2

0
2 3 4 5 6 0 2 3 4- 5 G
TIME CHOU!/3)

Figure 23, Creep curves; Shore A-5, 1-in. wide joint, 4-psi constant load.

1.2 JOI/VT DEPTH JOINT OE/ITH

0? in I /n.
/.0

Oil .......-1
[
~
-~
'--
-~
0.6

Od -

azl
0 - ~
L - L __l__J
l
<
'CC JOINT DEPTH JOINT DEPTH
'>:
1---_ !!12 In 2 If/.
V")

=[
0.4
f
02 -

0 2 5 4 5 6 0 2 J 4 5 G
TIME <HOURS)
Figure 24. Creep curves; Shore A-12, ½-in. wide joint, 12-psi constant load .
 27

/0 JO/NT OE/JTH JOINT OE/JTH

1/2 in. I in.
0.8

O.G
0.4 ,
<:;
;:: 0.2
-~
0
<
q J.O JOINT DEPTH JOINT OEPTH
""
...__ I.IP in. 2 in.
0-, 0,8

0.6
0.4

0.2
_ _ __ __ _ ------L..-.J
0
l2i345G 012345G
TIME ( HOURS)

Figure 25 . Creep curves; Shore A-12, 1-in. wide joint, 6-psi constant load.

/.0 - JOINT OE/JTH JOINT DEPTH

lh in. I in
08
QG

04
.~ 0.2 •
......_
-~
0
:a::
;:
<:,:
...__ /.0 JOINT 0£/JTH JOINT OE/JTH
(/)
I½ in. 2 in.
0.8

0,6

0.4

O.Z -

0
I 2 3 4 5 G 0 2 3 4 5 6
TIME ( HOURS J

Figure 26. Creep curves; Shore A-20, ½-in, wide joint, 16-psi constant load.
28

lO JOINT DEPTH JOINT DEPTH

/IE! in. I in.
0.8
06

0.4
:~- o.z
'-
.!:;
0
~ 1.0
<( JOINT DEPTH JOINT DEPT/I
,_
c,:
112 in. c in.
er, 0.8

O.G

0.4 -
0,2

0
2 3 4 5 0 O I 2 3 4 5 6
TIME {HOURS!

Figure 27. Creep curves; Shore A-20, 1-in. wide joint, 8-psi constant load.

1B
~
~ I Bps, I.? ;<>3I
[ 4.p:z/

0.8 -

0.6

06 ""'
'-: 0.5
0.
~ ~
01
'
'-- ~
<: 0.4 h: 03
~ ~
h: 02
i ~
~
0.2
~ 0./
~
0 10 ,".0 30 4t> St? 60 70 80 90 /(JO
0 I Z. 3 4 s T/1WE (M//1/UTES)
TIME ( HOURS)
Figure 29.Creep curve-photoelastic spec-
Figure 28. Boltzmann superposition curve; imen of clear polysulfide epoxy; Shore A-
Shore A-12, 1-in. wide joint. 20, 1- by 1- by 6-in. specimen.
 29

.50

--- Test values

--- Equation values

---._ 30
'<:)
\:::_

/0

0
?0 40 t,O 80 /(}tJ /2() /60
T/A✓E ( MI/VUTE.5)

0 06
Figure 30. Stress relaxation curves; Shore A-5, f =f 0 exp (t/T) where T = 180 min .

50

Test values
---- Equa1ion values

0 20 -10 60 80 /00 /20 /40 /60

TIME (M/NUTE5)

Figure 31, Stress relaxation curves; Shore A-12, f = f exp (t/T) 0 '
6
where T 18o min.
0
30

TO
Test J!.Q]yn_ __
---- Equation values

/0 -

ON~WM~~m~~-~m~~~~~

T//WE ( M//llu'TE5)

Figure 32. Stress relaxation curves; Shore A-20, f =f0 exp -t/r) 0 '
5
where T 360 min .

/00

90

.!30

70

t[,O
Testing machine values
---- so
~
Photoelastic values
Equation values

~ 40

~ JO

" 20

/0

0 20 40 60 80 /00 /20 /60

T/MF ( MI/VUTE5)
Figure 33. Stress relaxation curves-photoelastic specimen of clear polysulfide epoxy;
Shore A-20, 1- by 1- by 6-in. specimen, 25 percent constant strain.
 31

,ftJ

36
High range
V) 32 Middle range
V)
l<J Low range
28
~
~ 24
~
~
{_ 20
/6
~\J 12
:J::
If) 8

-4'0 -2() 0 20 40 60 60 /{)() /,?{)

TEA.f.PEJ?ATURE (LJR;RE[.5 r.J

Figure 34. Shore hardness vs t emperature .

40
35
~

---
t--
V 30
t --
Cf)
"<:(
--.J
25
l..u
l..l.... 20
C)

V)
:::::,
15
-...J
::::, (0
c:::::i
CJ
~ 5

o s m 0 ro M ~ ~ ~
SHORE HARDNE55
Figure 35. Tensile modulus vs Shore hardness.
32

Appendix B
C'1AT TT,,.,Tl'"'\.'11.T Ar.i C, A 1\/fTIT r,, n n A n T Dl\/fC"
PV.1..JU .1..1.Vl'II V .J.' t.J.n..J.V.L.C-.1..J.L:J r.L\,VLJ.L.J.L:J.J.V.lU

Sealant as an Elastic Material

To cover the range of field conditions, three cases of stress should be considered:
(a) tension (or compression) alone, (b) tension plus shear, and (c) compression plus
shear. For purposes of illustration, dimensions of a typical highway bridge are used.
The structure is a 60-ft span rolled-beam bridge with a concrete deck. The beams
are 33WF130 spaced at 6 ft cc. The joint sealant is a square cross-section, 1 in . wide
by 1 in. deep.
Tension Alone. -Assume the sealant material is placed at 75 F. Temperature range
is -40 to +120 F, maximum differential is 75 - (-40) = 115, and coefficient of ex-
pansion of bridge is 0. 0000065 (steel). Anticipated movement = 0. 0000065 x 60 x 115 x
12 = 0. 54 in. due to temperature.
Tension Plus Shear. -The case of tension plus shear could be caused in two ways,
first by the impact of a truck wheel bearing on the extreme end of the bridge, or ro-
tation at the support as caused by midspan deflection. A truck wheel bearing on the
extreme end of the bridge tends to deflect the short cantilevered portion of the bridge
which extends beyond the centerline of bearing (assumed to be 1 ft). For a 20-ton
truck, one rear wheel = W(0. 4) = 0 . 4 x 20 x 2 = 16 kips, distribution factor (AASHO)
S/ 5 . 5 = 6 . 0/ 5 . 5 = 1. 09 . Impact at 30 percent is given by P = 16 x 1. 09 x 1. 3 = 22. 7
kips . Assume this load is concentrated at the end of the short cantilever; therefore,

PL
3
22 . 7 X (1) 3 X 1728
Deflection = 3EI 3 X 29 X IQ 3 X 6700 = 0' 000067 in.

This deflection represents a shear-type movement in the sealant but its magnitude is
only one-ten thousandth of the movement in the tensile direction; therefore, shear can
be neglected. To consider the end rotation of the structure, consider the 20-ton truck
at midspan, using the same distribution factor and impact. Therefore, P = 20 x 1. 09 x
! _1 = 2B 4 !_,-.if'"', <>_nrl Pnn rnbtinn Ai<: !:ivPn hy:

0 PL2 28.4 X 60 2 X 144

= 16EI = 16 X 29 X 10
3
X 6700
0.0047 rad

Movement at top of slab = 45 x 0. 0047 = 0. 213 in.; top Wx = length at top of extended
sealant = 1. 54 + 0. 213 = 1. 753 in. For the amount of rotation shown (0 . 0047 rad), the
upwa rd movement at the end of the slab will be 0. 060 in. This is a s hear-type move-
m ent but it i s onl y 8/ioo of the movement in the tensile direction and makes virtually no
change in the stress magnitude.
The implication of these computations is that shear may, in general, be neglected
in combination with tension, but that the case of end rotation is important. However,
this importance comes, not from shear, but from the fact that it causes a further ex-
tension of the sealer in the tensile direction, which in this case is almost one-half as
much (0 . 213 vs 0. 54) as the extension caused by temperature change. This fact should
be given careful consideration in the future design of the service life of any joint sealer.
Using the statistical approach, the stresses in the sealant are determined a s follows:
Total extension = 0 . 54 + 0. 213 = 0 . 753 in.
Joint width = 1. 0 + 0. 753 = 1. 753
A.1 = >.. = 1. 753/1. 0 = 1. 753
A.2 = 1/ >.. = 1/1. 753 = 0. 5 70
A. 3 = 1
W = ½ G [ >.. = 1/ >.. + 1 - 3]
2 2
 33

The only force acting is in the tensile direction; therefore,

f = dw
ct>.. 1]
= G [ >.. - XS"

The value of G (34 psi) is in the value for a high modulus tread stock rubber which is as
hard a material as could be practicably used as a sealant:

f = 34[1.753 - (l.~53 ) 3 ] = [1.753 - -(}.~ 85] 34 = 53.3psi

This force f is the magnitude of the force acting on a unit cross-sectional area in the
unstrained state. For this specific problem it is acting adjacent to the bonding surface
where no neckdown has taken place. This value at the interface must be related to the
peel strength of the material in question. The polysulfides show a peel strength which
averages only 42 psi by test, so that a peel failure seems imminent. The stress "t"
on the necked-down cross-section of the sealant is given by:

Comparison of this value with the true stress-true strain curve indicates that a poly-
sulfide will stand stresses of three times this magnitude at-this strain before failure.
In brief, cohesion looks safe but a peel failure seems imminent.
Compression Plus Shear. -Separate from, but related to, these examples is the case
at the other end of the temperature scale. When the bridge is fully expanded and the
sealer is compressed, the latter extrudes above the roadway surface. At this time,
temperature is high and the sealant has softened somewhat. Experience shows that
these conditions normally result in failure by folding and flattening of the sealant under
the action of traffic. However, since we are working under the assumption that the
sealant is fully elastic, the net effect of traffic will be the depression of the sealant
back into the joint. The stress in the sealant at this time will be the resultant of a
compressive stress caused by temperature and a punching type of shear as caused by
the wheel load. Movement due to compression = (1. 20 - 75) x O. 0000065 x 60 x 1. 2 =
0. 21 in. due to temperature;

1 - 0. 21
0.79
1.0

dw =G>----xs
f = crx [ 1] 42. 5 psi

The amount of shear is determined by writing and differentiating the parabolic equa-
tion of the extruded sealant. For this case shear is the downward movement back into
the joint of a differential element of sealant adjacent to the interface. It is expressed
as the tangent of the angle ¢ which the sealant makes with the pavement: a = tan ¢ =
0. 47 and shearing stress y = G a = 34 (0. 47) = 15. 9 psi. Combining the direct com-
pression stress with the shearing stress gives a principal stress, S max = 47. 7 psi.
This stress, which because of the nature of the loading is related to peel strength, also
exceeds the peel strength of the polysulfides.
34

Sealant as a Flowing Solid

The illustrative problems use the following equivalent model equations (model shown
;n ...li1i,
...... ...
r
.A.t:,•
1fl) •
.... .., , •

(9)

2
G1B G;f3 7J1B + 1)2B ( G18 + G}3 ) (10)
2
( 111B G2B - 7J2B G1B

(11)

(12)

The data for these problems were taken from the creep and stress relaxation curves
included here for the medium range material (Shore A = 12). The strain rate is taken
from a curve of joint movement with temperature change. Hourly fluctuations of tem-
perature were obtained from the U. S. Weather Bureau. The temperature curves are
plotted for the one day in March which showed the greatest temperature variation.
According to this curve, the rate of temperature change and, consequently, the strain
rate remain relatively constant for a period at 0.0014 in ./in./min.
The following data were taken from the creep curve for Model A:

G1A + -1 E1A = 10 psi

3

1 EA
azA 3 2 = 7 psi

rizA 960 lb-min/ sq in.

r,;iA 9,600 lb-min/ sq in.

From the equivalency relations, the constants for Model B are

G1B 7.0 psi

GaB 3. 0 psi

r,1 B 9, 300 lb-min/ sq in.

r, 2B 300 lb-min/ sq in.

T1 r/1 1,330 min

G1

T2 = r/2 100 min

Ga
 35

Since a bridge will normally build up stress until it lurches in a sudden movement,
this amount of movement will have to be assumed. This amount of lurch ( £ 0 ) assumed
is O. 25 in. , slightly more than half the anticipated movement in a day's time.
Case I. -Where £ = £ 0 + ,St, determine the stress at t = 60 min after an initial lurch
ofO.25in.

= [0.0014 (9300) + e- 0· 066 (7.O x 0.25 - 0.0014 x 9300~ + @.0014 (300) +

e-0• 50 (3.0 X 0.25 - 0,0014 X 300~

= 13.02 + -11.6 + 0.42 + 0.18 = 1.92psi

Case II. -In the event that the bridge does move ideally, £0 0. The stress at
t = 60 min reduces to

a = [.ari1 (1 - e-t/r~ 2
+ ~r, (1 - e-t/T~

= 13.02 (1 - e- 0· 066 ) + 0.42 (1 - e- 0· 00) ==O

Case III. --In the event that the bridge moves through an initial lurch and then re-
mains at a constant elongation, i9 = 0 £ 0 = 0. 25 and t = 60 min:

= 1.80 + 0.41 2.2lpsi.

It becomes immediately apparent that when stress relaxation enters the picture, the
entire outlook changes. The stresses shown here are well below the demonstrated peel
strength of the polysulfides; therefore, under normal conditions, the sealant should
not fail.