Modelling of Heat Transfer across Bolted Joints in

Abaqus/CAE
Volkan Coskun, Emir Ozkokdemir, Bulent Acar
ROKETSAN Missile Industries
Abstract: Bolted joints are frequently used connections in many engineering systems, and crucial
parts of the heat transfer path in structures. During the design phase, heat transfer across the
joints must be modelled properly in order to determine the accurate temperature distribution on
components. Usually in finite element models, perfect thermal contact between mating surfaces is
assumed. However; this approach is inadequate and may bring on misleading results owing to the
equality of temperature assumption between surfaces. Complex configuration of the joints and
non-uniform contact pressure distribution due to bolt preload make the prediction of transient
heat transfer characteristics across the joints arduous. In this study, transient conduction heat
transfer through the bolted joints was modelled in Abaqus/CAE by defining gap conductance
between the abutting surfaces. Two correlations were utilized to estimate the thermal contact
resistance at joints. Impact of bolt preload and resulting pressure distribution on heat transfer and
temperature distribution were investigated. Experiments were performed to verify the analyses
results by conducting various temperature measurements. Finally, coupled temperature
displacement analysis of a missile guidance section under aerodynamic heating was conducted
and results were compared with the results of an uncoupled heat transfer analysis in which perfect
thermal contact was assumed at all interfaces.
Keywords: Bolted Joints, Gap Conductance, Thermal Contact Resistance

1. Introduction
Customer requirements lead to more complex missile systems and using components susceptible
to temperature. One of the most challenging duties to ensure mission success of missiles is to keep
some components such as avionics within their narrow operating temperature range during flight.
To determine whether employing thermal management systems is necessary or not, temperature
distribution on the components shall be accurately estimated. Accurate estimation of the
temperature distribution also enables us to decide which and to where thermal protection systems,
e.g. ablative shields and phase change materials, shall be implemented.
Major thermal load a missile is subjected to during its flight at supersonic and hypersonic regimes
is the aerodynamic heating which can be briefly described as conversion of the kinetic energy of
air into thermal energy due to viscosity (see Figure 1). Aerodynamic heating brings on very high
heat flux rates q”, up to several millions W/m2 which makes thermal contact resistance’s effect at
interfaces significant. The higher the heat flux is, the higher temperature drop across an interface
takes place due to the following relationship,
( )

2015 SIMULIA Community Conference 1

www.3ds.com/simulia
where T1 and T2 denote the temperatures of the mating surfaces and kgap is the thermal contact
conductance which is also called as gap conductance (see Figure 2).

Figure 1. Aerodynamic heating during flight (courtesy of Roketsan).

Figure 2. Temperature drop across a joint, (Gilmore, 2002).

In this study, heat transfer through bolted joints was modelled in Abaqus\CAE by defining gap
conductance between mating surfaces. Two models were utilized to estimate the thermal contact
conductance at joints. Impact of bolt preload and resulting pressure distribution on heat transfer
and temperature distribution were investigated. Experiments were performed to verify the analyses
results by conducting various temperature measurements. Finally, coupled temperature
displacement analysis of a missile guidance section under aerodynamic heating was conducted and
results were compared with the results of uncoupled heat transfer analysis in which perfect thermal
contact was assumed at all interfaces.

2 2015 SIMULIA Community Conference

www.3ds.com/simulia
2. Heat transfer in bolted joints
Bolted joints are crucial parts of the heat transfer path in missile structure. In general, perfect
thermal contact between mating surfaces is assumed. However, this approach may bring on
misleading results especially when heat fluxes are very high. Complex configuration of the joints
and non-uniform pressure distribution due to bolt preload make the prediction of heat transfer
characteristics across the joints arduous.
Figure 3 depicts a typical bolted joint construction (a), pressure distribution at the interface (b),
and heat flow through the joint (c). Pressure is at its peak near the bolt hole and assumed to be
zero outside of the pressure cone. Since gap conductance is a function of pressure, heat flow lines
constrict in the region where the pressure is high.

Figure 3. Heat transfer in a bolted joint ((a) a bolted joint, (b) pressure distribution
at the interface, (c) heat flow), (Madhusudana, 2014).
There exist several correlations for thermal contact conductance in the literature. They can be
broadly classified as ones with elastic deformation assumption and ones with plastic deformation
assumption (Hasselström, 2012). They relate gap conductance to the interface pressure, surface
quality parameters and material properties.

3. Test sample and FE model

Test sample used in experiments is similar to a pipe flange construction (see Figure 4). While one
end of the sample is closed, the other end is open. The test sample was heated at the closed side for
a certain time with a hot air gun whose heating power can be adjusted up to several kW, while all
other surfaces were insulated. Temperature distribution on the test sample was measured with K
type thermocouples. A series of thermocouples placed on the sample were aligned with the bolt
centerline, and another series of thermocouples were aligned with an artificial line lying along the
mid of the two neighboring bolts.

2015 SIMULIA Community Conference 3

www.3ds.com/simulia
 Figure 4. Test sample.
Before tests were conducted, the arithmetic average roughness and flatness of the mating surfaces
of the flange were measured. Flatness of the surfaces were 0.013 and 0.005 mm. The RMS
roughness of the surfaces were calculated as 0.65 and 0.78 μm by assuming a Gaussian
distribution of asperities.
Tests were carried out at three different torque levels, i.e. Γ, 2Γ, and 4Γ Nm. After each test, the
sample was let cool down for a time to reach ambient temperature and have uniform temperature
distribution. Afterwards, torque level was adjusted with a torque meter and a new test was started.
After tests were completed, FE model of the test sample was created in Abaqus/CAE. Since
pressure is zero at interfaces in uncoupled heat transfer analysis, coupled temperature-
displacement analysis must be conducted to model the pressure dependent thermal contact
conductance. Therefore, linear solid coupled temperature-displacement elements (C3D8T) were
used to build the model. Since the model is symmetric, only 1/12 of the test sample was built (see
Figure 5). Surface-to-surface contacts were defined for all mating surfaces except tie contact was
used for the interface between the nut and bolt.

4 2015 SIMULIA Community Conference

www.3ds.com/simulia
 heat flux
hot side of
flange
bolt force

symmetry
restraint

cold side
of flange

Figure 5. FE model of the test sample.

Two correlations taken from the literature were utilized to define the thermal contact conductance
between the mating surfaces. The first correlation, i.e. Mikic’s correlation (Gilmore, 2002), is
based on the elastic deformation assumption, and the second one, i.e. Tien’s correlation
(Hasselström, 2012), is based on the plastic deformation assumption. GAPCON subroutine was
used to define the gap conductance in Abaqus/Standard, because GAPCON offers flexibility in
specifying dependencies of gap conductance on the interface pressure, surface quality parameters
and temperature dependent material properties.

4. Results
Three different torque levels and two different thermal contact conductance correlations were used
in the simulations. In total, six simulations were conducted and results were compared with the
test data. Temperature drop across the two different sections of the flange is chosen for
comparison since temperature drop is the major variable related to the gap conductance and
interface pressure (see Figure 6). Figure 7, 8 and 9 show the temperature drop across the flange for
different torque levels of Γ, 2Γ and 4Γ Nm, respectively.

2015 SIMULIA Community Conference 5

www.3ds.com/simulia
 point 1 point 3

point 4

point 2

Figure 6. Locations where temperature drops were compared.

Figure 7. Temperature drop across the flange (Γ Nm) ((a) between point 1 and point
2, (b) between point 3 and point 4).

Figure 8. Temperature drop across the flange (2Γ Nm) ((a) between point 1 and
point 2, (b) between point 3 and point 4).

6 2015 SIMULIA Community Conference

www.3ds.com/simulia
 Figure 9. Temperature drop across the flange (4Γ Nm) ((a) between point 1 and
point 2, (b) between point 3 and point 4).
As seen from the figures, temperature drop between the points 1 and 2 is lower than the
temperature drop between the points 3 and 4. This is due to the higher interface pressure and gap
conductance in the proximity of the bolt hole. Figure 10 shows the temperature distribution and
contact pressure at the flange interface of the cold side when the bolt torque is 4Γ Nm and time is
Θ. Higher temperature and contact pressure values are observed near the bolt hole.

o
Figure 10. Temperature (in C) and pressure distribution (in MPa) at the interface.
Fully coupled heat transfer analysis with Abaqus/Standard results in acceptable results compared
to the test data especially when time is less than 0.8Θ. The Mikic’s correlation yields more
satisfactory results than the Tien’s correlation when torque level is lower than 4Γ Nm. Difference
between the two correlations becomes negligible as torque level is increased.
Figure 11 demonstrates the effect of bolt torque on temperature drop across the flange. As torque
level is increased, temperature drop across the flange reduces. It is an expected result since the gap
conductance is almost proportional to the interface pressure which is determined by the bolt
torque.

2015 SIMULIA Community Conference 7

www.3ds.com/simulia
 Figure 11. Effect of the bolt torque on temperature drop ((a) between point 1 and
point 2, (b) between point 3 and point 4).

5. Aeroheating analysis of a missile guidance section

As stated above, temperature of the avionics shall not exceed the maximum operating temperature
during the flight. Therefore, in order to decide the necessity of implementing thermal protection,
accurate thermal analysis shall be made. Figure 12 shows the quarter model of a missile guidance
section built in Abaqus/CAE. The model consists of two casings fastened with bolts, plate and
avionics. Bolted joint connections are also used to fasten the plate to the casing2 and the avionics
to the plate (which is not shown in the figure).

bolts casing2

casing1

bolt

avionics
plate

Figure 12. Quarter model of the guidance section.

Aerodynamic heating environment the missile is subjected to was determined via CFD analysis in
the form of recovery temperature and heat transfer coefficient versus time. This thermal load was

8 2015 SIMULIA Community Conference

www.3ds.com/simulia
defined in Abaqus/Standard by introducing surface film condition on the outer surfaces of the
casings. All other surfaces were assumed to be adiabatic.
Linear solid coupled temperature-displacement elements (C3D8T) were used to build the model.
Surface to surface contacts were defined between all mating surfaces except tie contact was used
between the threaded regions of the bolts and holes. GAPCON subroutine was utilized to describe
the thermal contact conductance according to Mikic’s correlation, because Mikic’s correlation
yields more satisfactory results at lower torque levels and the differences between the correlations
becomes negligible at higher torque levels. So it covers broader range of torque levels.
Figure 13 shows the temperature distribution on the guidance section at the end of the flight. Heat
flows to the avionics through the bolted joint connection. Therefore, heat transfer in the bolted
joints and the thermal contact conductance have a direct impact on the temperature of the avionics.

o
Figure 13. Temperature distribution (in C) on the guidance section.
The same aeroheating analysis was also conducted with heat transfer elements (DC3D8) after
removing bolt loads and converting all surface to surface contacts to the tie contacts. This is the
case in which perfect thermal contact is assumed between all mating surfaces. Figure 14 and
Figure 15 compare the results obtained from the fully coupled temperature-displacement analysis
and uncoupled heat transfer analysis.

o
Figure 14. Temperature distribution (in C) on the plate and avionics ((a) coupled
temperature-displacement analysis, (b) uncoupled heat transfer analysis).

2015 SIMULIA Community Conference 9

www.3ds.com/simulia
 error ~8 oC

error ~5 oC

Figure 15. Volume average temperature variation of the plate and avionics.
Figure 14 and Figure 15 show the temperature distributions on the plate and avionics, and volume
average temperature variation of the plate and avionics, respectively. As seen clearly, there exists
a significant difference between the results of the analyses when taking into account the narrow
temperature margin of the avionics. Perfect thermal contact assumption results in higher
temperature estimation on the avionics which may lead to the unnecessary use of thermal
protection.

6. Conclusion
This paper presents the modelling of heat transfer in bolted joints in Abaqus/CAE by defining
pressure dependent thermal contact conductance between mating surfaces in a fully coupled heat
transfer analysis. Several tests were conducted by taking numerous temperature measurements on
the test sample and measurements were compared with the results of the fully coupled heat
transfer analysis. Fully coupled temperature displacement analysis in Abaqus/Standard gave
satisfactory results compared to the test data which emphasizes the flexibility and trustworthiness
of Abaqus. Furthermore, fully coupled and uncoupled heat transfer analyses of a missile guidance
section were conducted. Results indicate that there is a significant difference between the
estimated results, as perfect thermal contact assumption overestimates the temperature on the
components.

7. References
1. Gilmore, D.G., “Spacecraft Thermal Control Handbook Volume I: Fundamental
Technologies,” The Aerospace Press, El Segundo, California, 2002.
2. Hasselström, A. J., and Nilsson, U. E., “Thermal Contact Conductance in Bolted Joints,”
Diploma Work, Chalmers University of Technology, Sweden, 2012.

10 2015 SIMULIA Community Conference

www.3ds.com/simulia
3. Madhusudana, C. V., “Thermal Contact Conductance,” Springer International Publishing,
Switzerland, 2014.
4. Abaqus Users Manuel, Version 6.13-1, Dassault Systemes Simulia Corp., Providence, RI.

2015 SIMULIA Community Conference 11

www.3ds.com/simulia