A Proposed Model for Creep

Relaxation of Soft Gaskets in

Bolted Joints at Room
Ali A. Alkelani Temperature
Basil A. Housari
A mathematical model is proposed for predicting the residual clamp load during creep
Sayed A. Nassar and/or relaxation in gasketed joints. An experimental procedure is developed to verify the
proposed model for predicting the gasket relaxation under a constant compression, gas-
Fastening and Joining Research Institute, ket creep under a constant stress, and gasket creep relaxation. To study gasket creep
Department of Mechanical Engineering, relaxation, a single-bolt joint is used. The bolt is tightened to a target preload and the
Oakland University, clamp load decay due to gasket creep relaxation is observed over time under various
Rochester, MI 48309 preload levels. Experimental and analytical results are presented and discussed. The
proposed model provides an accurate prediction of the residual clamp load as a function
of time, gasket material, and geometric properties of the gasket. A closed form solution is
formulated to determine the initial clamp load level necessary to provide the desired level
of a steady state residual clamp load in the joint, by taking the gasket creep relaxation
into account. 关DOI: 10.1115/1.2826430兴

Keywords: gasket creep relaxation, clamp load loss, gasketed bolted joints

Introduction more accurate modeling of a gasketed bolted joint, this paper

provides a closed form equation for evaluating the gasket relax-
Gaskets are used to create and maintain a seal between two
ation, creep, and creep relaxation in single-bolt joint. This pro-
separable flanges. Theoretically speaking, flanges should be per-
posed model provides an accurate prediction of the residual clamp
fectly flat, smooth, and rigid, so they could be bolted together and
load as a function of time, gasket material, and geometric proper-
seal without the need for a gasket. However, this is rarely the case
ties of the gasket.
in most applications; hence, a gasket is needed to fill the cavities
between the two flanges and prevent joint leakage. A host of de-
sign and analysis challenges are associated with gasketed joints.
Among these challenges is the reaction of the gasket material to
Mathematical Modeling of the Gasket
the sealed medium, the pressure that the gasket can withstand in
the radial direction, the environmental effect on the gasket mate- In this work, a mathematical model is proposed to investigate
rial such as temperature and humidity, and more importantly the the gasket material behavior when subjected to an external pres-
change in the gasket thickness and stresses, which is usually re- sure. The gasket material selected for this study is styrene butadi-
ferred to as gasket creep relaxation 关1兴. ene rubber. It is a low cost gasket material that offers moderate to
Although the effect of gasket creep relaxation on the clamp good performance for low pressure applications. It conforms eas-
load loss in bolted joints has been always recognized, the creep ily to uneven flange faces. Some of the mechanical properties of
relaxation of gasketed joints remains a vital research subject. In this material, according to the ASTM-D-2000, are shown in Table
the literature, very few papers address analytically the effect of 1. The gasket material is considered viscoelastic. Figure 1 shows
gasket creep relaxation. Creep analysis of bolted connections is the presented proposed model of the gasket; it consists of three
presented in Refs. 关2–4兴, where steady creep was assumed for the mechanical elements in series 关1,11兴. The first element is an elas-
joint but the gasket creep was ignored. More recent studies 关5,6兴 tic spring with stiffness K1. The second element is pure viscous
show that the gasket creep relaxation is more important than the dashpot with damping rate C2. The third element is a viscoelastic
bolt/joint material relaxation and cannot be ignored. Kraus 关7,8兴 element, which consists of an elastic spring K2 and a dashpot C2
proposed a model to predict the time required for the bolts to relax connected in parallel. The gasket model in Fig. 1 can be used to
from an initial stress level to a steady state level. However, the explain its load-deflection characteristics. When a gasket is sub-
effect of the gasket and other joint structures was not considered. jected to a compressive force, the various model elements will
Most of the previously mentioned work focused on the bolt/joint behave differently, as shown in Fig. 2. First, the elastic member
material relaxation and ignored the effect of gasket creep relax- represented by the spring K1 will compress by ⌬1 and it will fully
ation. Bazergui 关9兴 and Bouzid 关10兴 considered the effect of gas- decompress once the compressive force is removed. Second, the
ket creep relaxation. Bazergui 关9兴 showed based on experimental viscoelastic element represented by the parallel combination of K2
data that, for most types of gaskets, a linear relationship between and C2 will compress by ⌬3 and it will recover after a period of
displacement, due to creep, and the time could be constructed on time if the load is removed; therefore, this element is described as
a semilogarithmic plot. Bouzid 关10兴 also adopted the same proce- time-dependent and fully recoverable element. Third, the viscous
dure to evaluate the relaxation in bolted flanged connections. For element represented by C1 will also compress by ⌬2 but when the
load is removed, it will not recover; hence, this element is de-
scribed as time-dependent and unrecoverable element. This ele-
Contributed by the Pressure Vessel and Piping Division of ASME for publication ment accounts for the long term gasket creep relaxation.
in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received July 20, 2006;
final manuscript received March 1, 2007; published online January 30, 2008. Review When a gasket is loaded, its equivalent stiffness Kg changes
conducted by William Koves. with time causing the gasket compression ␦g to increase and the

Journal of Pressure Vessel Technology Copyright © 2008 by ASME FEBRUARY 2008, Vol. 130 / 011211-1

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28489/ on 07/14/2017 Terms of Use: http://www.asme.org/a

 Table 1 Mechanical properties of styrene butadiene rubber 1

冋 册
Kg共t兲 = 共4兲
Temp. Tensile Elongation K1 + K2 t 1
Polymer range 共°F兲 Hardness 共psi兲 共%兲
+ − e−共K2/C2兲t
K 1K 2 C1 K2
Red rubber −20 to + 170 75⫾ 5 400 250 where the four constants K1, K2, C1, and C2 are determined ex-
perimentally; a creep test is conducted in order to obtain these
constants. The gasket is rapidly compressed to a predetermined
load value; then the change in deflection with time is recorded.
The stiffness of the elastic element K1 is obtained by substituting
stress to decrease. The equivalent gasket thickness at time t can be the force and compression values at t = 0 into Eq. 共3兲, as shown by
derived using the model shown in Fig. 1. For a constant compres- the following equation:
sion, the force F may be expressed as follows: F
K1 = Kg共0兲 = 共5兲
F = K1⌬1 = C1⌬˙ 2 = K2⌬3 + C2⌬˙ 3 共1兲 ␦g共0兲
As the time t becomes very large, the exponential term in Eq. 共3兲
where ⌬˙ 2 and ⌬˙ are the rates of change of ⌬ and ⌬ with respect
3 2 3 approaches zero and the equation reduces to

冋 册
to time.
For constant spring rates and damping coefficients, Eq. 共1兲 K1 + K2 t
yields the solution for ⌬1, ⌬2, and ⌬3 subject to zero initial con- ␦g共t兲 = + F tⰇ0 共6兲
K 1K 2 C1
ditions, as follows:
Measuring the total gasket deflection ␦g共t兲 at two sufficiently large
F times t1 and t2, the two constants K2 and C1 are determined by
⌬1 = 共2a兲
K1 using Eq. 共6兲 as

⌬2 =
t
C1
F 共2b兲
␦g共t1兲 = 冋 K1 + K2 t1
K 1K 2
+
C1
册F t1 Ⰷ 0 共7兲

⌬3 =
F
K2
− e−共K2/C2兲t
F
K2
共2c兲
␦g共t2兲 = 冋 K1 + K2 t2
K 1K 2
+
C1
册F t2 Ⰷ 0 共8兲

Finally, after the constants K1, K2, and C1 have been experimen-
Equations 共1兲 and 共2a兲–共2c兲 yield the total deflection of the gasket
tally determined from Eqs. 共5兲, 共7兲, and 共8兲, the fourth constant C2
共␦g = ⌬1 + ⌬2 + ⌬3兲 as follows:

冋 册
may also be determined experimentally. After a short time t3 from
K1 + K2 t 1 applying a known constant compression to the gasket, the gasket
␦g共t兲 = + − e−K2/C2t F共t兲 共3兲 deflection ␦g共t3兲 is measured; substituting the measured values
K 1K 2 C1 K2
into Eq. 共3兲 would yield the value of C2 by using

冋 册
An equivalent gasket stiffness Kg = F / ␦g is obtained from Eq. 共3兲
as follows: K1 + K2 t3 1
␦g共t3兲 = + − e−共K2/C2兲t3 F 共9兲
K 1K 2 C1 K2
The values obtained for the four constants 共K1, K2, C1, and C2兲 are
used in the mathematical model to evaluate the behavior of the
gasket material under different loading conditions.

Experimental Verification
In order to use the proposed mathematical model to predict the
gasket behavior, the effect of gasket stress and gasket thickness on
the gasket constants K1, K2, C1, and C2 needs to be investigated.
The gasket is loaded up to three different stress levels; the con-
stants were determined from one experiment 共at a given gasket
stress level兲 and used to predict the gasket behavior at the other
stress levels. To investigate the effect of gasket thickness on the
calculated constants, three different thicknesses of the same gasket
Fig. 1 Mechanical model of a typical gasket †1,11‡ material 共red rubber兲 are investigated, namely, 1 / 16 in., 1 / 8 in.,
and 3 / 16 in. For the experimental verification, creep tests are used
to evaluate the proposed gasket model.

Creep Test
In the creep test, the gasket load remains constant throughout
the test while the gasket compression 共thickness reduction兲 is
monitored, as illustrated in Figs. 3 and 4. An MTS fatigue testing
machine is used to conduct the creep test. The gasket material is
cut into 3 ⫻ 3 in.2 specimens that are placed between two steel
plates, as shown in Fig. 3. The two steel plates are centered be-
tween the two jaws of the MTS machine, and then the upper jaw
is brought down slowly until it touches the specimen. Then the
desired force is quickly applied and maintained for 40 min. Real
time data on the gasket load and displacement are collected using
a data acquisition system. A curve like the one shown in Fig. 4 can
Fig. 2 Gasket force compression †1‡ be obtained and used to calculate the gasket constants K1, K2, C1,

011211-2 / Vol. 130, FEBRUARY 2008 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28489/ on 07/14/2017 Terms of Use: http://www.asme.org/a

 Fig. 3 Gasket creep and relaxation test setup

and C2. These constants can be used to construct the gasket creep
versus time plot using the derived equation 共Eq. 共3兲兲 and compare
the results to the ones obtained experimentally.
Figures 5–7 show the analytical gasket creep predicted by Eq.
共3兲 versus the experimental data for various gasket thicknesses. Fig. 6 Experimental and analytical creep results for 1 / 8 in.
For each thickness, three levels of initial gasket loads are used; thick gasket
experimental data from one arbitrarily level is used for determin-
ing the four gasket constants and the other two levels are used to
verify the general validity of the model represented by Eq. 共3兲; used for determining the gasket constants for various thicknesses.
also the same constants values are used later to predict the gasket However, it was found that the gasket constants C1 and C2 are
relaxation and gasket creep relaxation behavior. The analytical significantly affected by the gasket thickness, but there was no
and experimental results are in good agreement; they both show clear correlation between those constants and the thickness.
that the stress level does not affect the gasket constants. Hence, an
arbitrary stress level that varies between 450 psi and 1000 psi is Relaxation Test
In the relaxation test, the gasket is compressed and maintained
at a fixed compression throughout the test, while the decreasing
gasket load is monitored in real time for 40 min. The test setup is
similar to that used in the creep test 共Fig. 3兲. However, in the
relaxation test the MTS machine is set up to apply the constant
deflection. Figures 8–10 show the analytical and the experimental
results for the relaxation test for gasket thicknesses of 1 / 16 in.,
1 / 8 in., and 3 / 16 in., respectively. Each figure shows the gasket
relaxation behavior for three levels of initial stress. The gasket
constants obtained from the creep test for each gasket thickness
are used to predict the relaxation behavior of the gasket 共Eq. 共3兲兲.
The analytical and experimental gasket creep results are in good
agreement. The test shows that the mathematical model presented
can be used successfully to predict the gasket relaxation behavior
as well as the creep behavior of the studied red rubber gasket.
Fig. 4 Typical creep curve

Fig. 5 Experimental and analytical creep results for 1 / 16 in. Fig. 7 Experimental and analytical creep results for 3 / 16 in.
thick gasket thick gasket

Journal of Pressure Vessel Technology FEBRUARY 2008, Vol. 130 / 011211-3

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28489/ on 07/14/2017 Terms of Use: http://www.asme.org/a

 Fig. 8 Experimental and analytical relaxation results for Fig. 10 Experimental and analytical relaxation results for
1 / 16 in. thick gasket 3 / 16 in. thick gasket

For the various levels of applied preload, the steady state load
relaxation is increased with the increase in gasket thickness. The deflection 共pure relaxation兲. For accurate prediction of the clamp
steady state clamp load losses as percentages of the applied pre- load loss in gasketed joints, however, the uncoupled relaxation
load during the relaxation test were 4%, 14%, and 21% for gasket and creep tests do not yield a reliable prediction 关14兴. Hence, a
thicknesses of 1 / 16 in., 1 / 8 in., and 3 / 16 in., respectively. The new gasket model is developed to predict the clamp load loss due
increased gasket relaxation 共due to thickness increase兲 would to gasket creep relaxation in a single-bolt joint. The model is
cause a similar trend in the clamp load loss and is consistent with shown in Fig. 11. The gasketed joint system is modeled as a set of
reported findings in Refs. 关1,12,13兴. springs and dashpots. The bolt is presented by a spring in parallel
with the gasket and the flanges. Figure 11共a兲 shows the free 共pre-
Mathematical Modeling of Gasket Creep Relaxation tightening兲 position for the different joint members. When the
The clamp load loss in a gasketed bolted joint is caused by the joint is tightened, Fig. 11共b兲 shows the gasket compression and
creep and relaxation of the gasket material. In most cases, the bolt elongation. As time elapses, gasket creep relaxation takes
threaded fasteners are used to create the required pressure on the place causing further gasket thickness reduction as illustrated by
gasket. Usually, after tightening the bolts, the clamp load starts to Fig. 11共c兲. This additional reduction in gasket thickness from Figs.
decrease. In gasketed joints, creep relaxation is the combined ef- 11共b兲 and Fig. 11共c兲 is not recoverable even after the load is com-
fect of the relaxation of the gasket material, where the load on the pletely removed; permanent gasket deformation is shown in Fig.
gasket decreases with time, and the gasket creep that causes bolt 11共d兲.
relaxation. Gasket creep causes the gasket to become thinner; Once the bolt is tightened, gasket creep relaxation starts. The
therefore, the bolt loses some of its elongation, which translates change in the force is the same in the gasket 共⌬Fg兲, the joint
into a loss in the joint clamping force. Most of the standard gasket members 共⌬Fc兲, and the bolt 共⌬Fb兲. This yields
tests are conducted under constant force 共pure creep兲 or constant
⌬Fg = ⌬Fc = ⌬Fb 共10兲
Using the model shown in Fig. 11, Eq. 共10兲 can be written as
follows:

Kg⌬␦g共t兲 = Kb⌬␦b 共11兲

Figure 11 shows that the change in the bolt elongation is equal to
the summation of the changes in the joint and gasket thickness;
this yields

⌬␦c + ⌬␦g = ⌬␦b 共12兲

Assuming that, for a soft gasket material, the change in the joint
thickness is too small compared to the change in the gasket com-
pression 共␦c Ⰶ ␦g兲, Eq. 共12兲 reduces to

⌬␦g ⬵ ⌬␦b 共13兲

The gasket compression ␦共t兲 at time t is still given by ␦共t兲
= F共t兲 / Kg共t兲.
The gasket model, which has been presented earlier, can be
used to estimate the change in gasket compression and the clamp
Fig. 9 Experimental and analytical relaxation results for load at any time t. The additional decrease in gasket thickness is
1 / 8 in. thick gasket given by

011211-4 / Vol. 130, FEBRUARY 2008 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28489/ on 07/14/2017 Terms of Use: http://www.asme.org/a

 Fig. 11 Mechanical model of single-bolt gasketed joint

冉
⌬␦g共t兲 = ␦g共0兲 −
F共t兲
Kg共t兲
冊 共14兲
gasketed joint. Results show good correlation between the experi-
mental data and the analytical model. The slight difference of
⫾5% between the model results and the experimental data could
where ␦g共0兲 is the gasket compression at initial tightening, Kg共t兲 is be due to normal variation in the gasket mechanical properties.
the gasket stiffness, and F共t兲 is the clamping force at time t.
Using Eq. 共13兲, the clamping force F共t兲 at time t is given by
F共t兲 = Kb␦b共t兲 = Kb关␦b共0兲 − ⌬␦g共t兲兴 共15兲
where ␦b共0兲 is the initial bolt elongation right after the tightening,
which is obtained from the bolt preload F0 and the bolt stiffness
Kb. Substituting Eq. 共14兲 into Eq. 共15兲, and by manipulation the
clamping force at any time t is given as follows:

F共t兲 = 关␦b共0兲 − ␦g共0兲兴 冉 KbKg共t兲

Kb − Kg共t兲
冊 共16兲

Equations 共4兲 and 共16兲 provide a closed form solution for the
residual clamping force F共t兲 in a single-bolt gasketed joint at any-
time t. These equations are used to construct the creep relaxation
plots for the studied red rubber gasketed joint.
The value of the initial clamp load F0 required to maintain a Fig. 12 A single-bolt gasketed joint test setup
desired level of the residual clamp force Fr, after the gasket creep
relaxation has taken place is determined from Eq. 共16兲 as follows:

F0 = Fr 冉 Kg共t兲 + Kb Kg共0兲
·
Kg共0兲 + Kb Kg共t兲
冊 共17兲

where Kg共0兲 and Kg共t兲 are obtained using Eq. 共4兲.

Experimental Verification of the Creep Relaxation

Model
Figure 12 shows the test setup used for the experimental veri-
fication of the proposed gasket creep relaxation model in a single-
bolt gasketed joint. The bolt is tightened using an electric tool at
127 rpm to minimize the amount of gasket creep relaxation that
takes place during the tightening process itself. A load cell is
inserted between the bolt and the upper plate to monitor the
clamping force during the test. After tightening, the joint is left for
40 min, during which the clamp load is recorded using a data
acquisition system.
The same gasket constants obtained from the earlier creep test
are used in the creep relaxation model. Figures 13–15 show the
analytical results from the proposed creep relaxation model 共Eq. Fig. 13 Experimental and analytical creep relaxation results
共16兲兲 in comparison with the experimental data, for a single-bolt for 1 / 16 in. thick gasket

Journal of Pressure Vessel Technology FEBRUARY 2008, Vol. 130 / 011211-5

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28489/ on 07/14/2017 Terms of Use: http://www.asme.org/a

 needed to produce a desired level of a steady state residual clamp
load Fr, by taking the gasket creep relaxation into consideration
共Eq. 共17兲兲.

Conclusions
Gasket creep, relaxation, and creep relaxation behaviors have a
significant effect on the residual clamping force in gasketed joints.
A mathematical model is presented to predict the gasket behavior
at room temperature, and an experimental procedure is established
to determine the necessary gasket constants for the model. The
effect of gasket stress level and the gasket thickness on the gasket
constants used in the model is investigated. The gasket constants
are found to be independent of the gasket stress level but are
affected by the gasket thickness for the gasket material considered
in this study 共red rubber兲.
A closed form solution is formulated for the clamp load as a
function of the time elapsed after the initial tightening of the joint.
The clamp load formulation has been successfully used to provide
a closed form solution that determines the initial clamp load level
that is necessary to provide the desired level of a steady state
Fig. 14 Experimental and analytical creep relaxation results residual clamp load in the joint, by taking the gasket creep relax-
for 1 / 8 in. thick gasket ation into account. The good agreement between the mathematical
model results and the experimental data suggests that the pro-
posed model can be used to accurately describe the gasket behav-
Hence, the proposed mathematical model may be used to predict ior and the clamp load loss due to gasket creep relaxation.
the gasket creep relaxation behavior in a single-bolt gasketed
joint, provided that the gasket constants have been experimentally References
determined and used in the model. The new creep relaxation 关1兴 Czernik, D. E., 1996, Gaskets Design, Selection, and Testing, McGraw-Hill,
model may also be used to estimate the initial clamp load F0 New York, pp. 51–54, 89–92.
关2兴 Bailey, R. W., 1937, “Flanged Pipe Joints for High Pressure and Tempera-
tures,” Engineering 共London兲, 144, pp. 364–365.
关3兴 Martin, J., 1938, “Stresses and Deformations in Pipe Flanges Subjected to
Creep at High Temperatures,” J. Franklin Inst., 226, pp. 645–657.
关4兴 Waters, E. O., 1938, “Analysis of Bolted Joints at High Temperatures,” Trans.
ASME, 60, pp. 83–86.
关5兴 Nassar, S. A., and Alkelani, A. A., 2006, “Clamp Load Loss Due to Elastic
Interaction and Gasket Creep Relaxation in Bolted Joints,” ASME J. Pressure
Vessel Technol., 128, pp. 1–8.
关6兴 Nassar, S. A., and Alkelani, A. A., 2006, “Effect of Tightening Speed on
Clamp Load Distribution in Gasketed Joints,” SAE Special Publication SP-
2033, pp. 173–187.
关7兴 Kraus, H., 1980, Creep Analysis, Wiley, New York.
关8兴 Kraus, H., and Rosenkrans, W., 1984, “Creep of Bolted Flanged Connections,”
Welding Research Bulletin No 294.
关9兴 Bazergui, A., 1984, “Short Term Creep Relaxation Behavior of Gaskets,”
Welding Research Council Bulletin No. 294, pp. 9–22.
关10兴 Bouzid, A., and Chaaban, A., 1996, “A Proposed Method for Evaluating Re-
laxation in Bolted Flanged Connections,” International Conference on Pres-
sure Vessel Technology, Vol. 2, pp. 123–132.
关11兴 Ehrenstein, G. W., 2001, Polymeric Materials: Structure-Properties-
Applications, Hanser, Munich.
关12兴 Bouzid, A., Chaaban, A., and Bazergui, A., 1995, “The Effect of Gasket
Creep-Relaxation on the Leakage Tightness of Bolted Flanged Joints,” ASME
J. Pressure Vessel Technol., 117, pp. 71–78.
关13兴 Nassar, S. A., and Alkelani, A. A., 2006, “Clamp Load Loss Due to Elastic
Interaction and Gasket Creep Relaxation in Bolted Joints,” ASME J. Pressure
Vessel Technol., 128, pp. 395–401.
Fig. 15 Experimental and analytical creep relaxation results 关14兴 Bickford, J. H., 1995, An Introduction to the Design and Behavior of Bolted
for 3 / 16 in. thick gasket Joints, 3rd ed., Dekker, New York.

011211-6 / Vol. 130, FEBRUARY 2008 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28489/ on 07/14/2017 Terms of Use: http://www.asme.org/a