Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Thermo-Mechanical Simulations of Rear Lamps 2011-01-0109

Published
with CFD 04/12/2011

Flavio Cimolin and Michele Rabito

AMET S.R.L.

Andrea Menotti
OLSA S.P.A.

Copyright © 2011 SAE International

doi:10.4271/2011-01-0109

ABSTRACT INTRODUCTION
Designing a rear lamp for the automotive industry is getting
A complete methodology for the thermo-mechanical analysis
more and more challenging: the best compromise between
of optical devices for the automotive industry is presented.
lightness and resistance to heat loads needs to be achieved, in
The objective is to predict the thermal field all over the lamp,
order to let it become an actual style component of the
highlighting the zones with risk of melting, and the
vehicle. Strong heat loads could be responsible of severe
deformations and stresses associated with it. The proposed
plastic deformations of both the body and the external lens,
approach is based on a Computational Fluid-Dynamic (CFD)
resulting in a global damage for the whole optical component.
simulation capable of capturing all the heat transfer
While the use of thinner and lighter materials allows the
phenomena occurring inside and outside the lamp:
designers to reach really innovative configurations in terms of
conduction between different components of the device,
elegance and style, on the other hand the problem of
natural convection associated with density changes in air
highlighting the thermally most critical zones becomes
(buoyancy effects), and radiation heat transfer. The latter
everyday tougher.
requires a fairly complex modeling strategy in order to
provide a satisfactory (and conservative) treatment for the
Prior to accept a newly developed rear lamp, it must come
source of power, i.e. the filament, which can be obtained by
through a series of tests representing its behavior in the worst
means of a proper inclusion of transparency. The radiation
environmental conditions (heat, rain, wind or moisture loads).
model is verified according to a theoretical-numerical
The test could even schedule a sequence of power on and
comparison on a schematic test case; the whole methodology
power off events for different bulbs associated to different
is then validated on a simple prototype of lamp with the aid
services inside the same lamp, leading to the necessity of
of experimental investigations. For a more accurate
following a time-dependent process instead of a steady-state
description of the boundary conditions for the lamp, it is
one. As an example, consider the case of switching on and off
possible to include the external environment, on which
a lightbulb at cycles of 5 minutes, while keeping another one
natural convection arises too. The treatment of unsteady
always on, leaving the lamp inside a ventilated oven set at a
simulations is discussed, with the description of a suitable
working temperature of 55°C for 90 minutes.
adaptive timestep algorithm capable of reducing the
computational costs and thus keeping the simulation feasible.
The whole methodology is finally tested on a complex
industrial lamp.
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Figure 1. Example geometry of a complex rear lamp (foglamp and backup light).

importance of handling specular reflection in order to

accurately track the incident rays on the internal parabolic
surfaces. Moreover, many works can be found in literature
for what concerns the problem of condensation formation on
the lens, due to unsufficient moisture removal by vents
openings (see e.g. [15,16,17,18,19,20]), usually more relevant
for front lights.

Our work, which follows the lines of [21], aims at reviewing

the whole numerical resolution of the problem in order to set
a robust and reliable methodology for the simulation of the
thermo-mechanical behavior of a rear lamp. Focusing on the
modeling of the filament-bulb interaction, that is reproducing
accurately the source of power inside the lightbulb without
Figure 2. Example thermo-fluid-dynamic results any rough simplification, our approach turns out to reproduce
(temperature field on the lamp and velocity field on reality with fairly good agreement as well as perfect energy
vertical sections). balance on the whole system under analysis. This is achieved
by means of a detailed description of the radiation heat
transfer, obtained by employing a surface-to-surface model
Thermal simulations represent an interesting approach to capable of including transparency effects and specular
obtain at an early development stage of the optical device a reflection. Moreover, by means of a CFD simulation it is
prediction of its thermo-mechanical field in working and possible to describe the airflow associated with natural
testing conditions, highlighting the most critical zones on convection, which are responsible for the most critical
which there is concrete risk of melting. Moreover, a clever temperatures directly above the bulbs. For a more accurate
analysis of the results can usually provide suggestions for the description of the boundary conditions for the lamp, it is
resolution of those criticalities, such as for example the possible to include the external environment too, which can
introduction of interior lens or other conductive objects to be still air as well as a ventilated oven, often used in testing
improve the heat transfer, or the cutting of holes on its body conditions.
in order to optimize the recirculation of the air.
A detailed verification and validation of the approach will be
Many studies have been already accomplished in order to presented and discussed, which has been assessed by means
enhance such predictive capabilities, starting from the of both theoretical considerations and experimental
experimental investigations of [1, 2] and going towards the investigations. The numerical-experimental comparison turns
numerical simulations of radiation heat transfer [3, 4] as well out to show a pretty good agreement and perfect energy
as natural convection [5,6,7]. Experimental-numerical balance, assessing the reliability and robustness of the
comparisons indicate that the prediction of peak temperatures methodology. For what concerns the treatment of unsteady
due to buoyancy can be obtained with quite good accuracy simulations, we will provide a suitable algorithm that has
[8,9,10]. Interesting results have been obtained in [11, 12] for been devised in order to keep the computational cost above
“open” lamps too, that is for lamps in which the internal air is feasibility. Indeed, it is necessary to find a compromise
in direct communication with the external one. Dealing with between the planned timestep and the desired accuracy: the
radiation modeling, the analyses of [13, 14] showed the best choice is to adapt the timestep as the solution advances,
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

in such a way that at each time advancing iteration the velocity increases. Natural convection occurs because of the
maximum gain is achieved under suitable controlling movement induced by the different densities of the fluid
conditions. associated to different pre-existing temperatures, thus is
ultimately caused by the influence of gravity [23].
The whole methodology is finally tested on a real complex
lamp, namely the backup light and foglamp represented in The adimensional parameter that better characterizes natural
Fig. 1. convection is the Rayleigh number:

PHYSICAL ASPECTS
It is well-known that the three fundamental heat transfer
phenomena are conduction, convection and radiation [22]. (3)
The structure of a lamp is so complex that all of them need to
be taken into account if one aims at fully reproducing g being the gravity acceleration, α the volumetric expansion
numerically the thermal field of all the components. The main coefficient of the fluid medium, µ the dynamic viscosity
heat transfer phenomenon is radiation, since the source of coefficient, ΔT the temperature difference between the hot
heat is the filament of the light bulb, which at first and the cold zone, and L the distance among them. For values
approximation does not touch any other component. of Ra less than 103 no movement occurs, since the buoyancy
Moreover, because of high temperature differences between forces have not sufficient strength. For values of Ra between
the bulb, the body of the lamp and the surrounding air, 103 and 106 a primarily ordered and laminar movement is
natural convection arises both inside and outside the lamp, observed, structured according to the so-called Bénard
which is responsible for the most critical observed convective cells. For higher values of Ra turbulence occurs,
temperatures, usually located just above the bulb. leading to more chaotic and unstructured movements of the
fluid.
CONDUCTION
Conduction is the simplest among the heat transfer Dealing with a rear lamp, assuming the fluid to be the air and
mechanisms: it represents the diffusion of heat inside a body as temperature difference and potential vertical distance
from the region where the temperature is higher towards between hot and cold zones ΔT=200 K and L=10 mm
those where it is lower. The reference mathematical model respectively (with respect to the flow inside the lamp), one
for its description is obtained starting from the balance obtains as indicative value Ra=105, therefore expecting
equation for the energy, to which it is necessary to associate natural convection to arise in a structured way with laminar
the Fourier's constitutive law flow.

The mathematical model capable of describing the evolution

(1) of natural convection consists in the Navier-Stokes equations
for incompressible fluids [24]:
which establishes the direction of the heat flux to be
proportional to the gradient of the temperature T, k being the
thermal conductivity coefficient and q the heat flux vector.
The resulting heat equation is therefore the following one:
(4)

where v and p are the velocity and pressure field respectively,

and f the term associated to the volumetric forces, i.e. gravity
(2) in our application. In order to model natural convection it is
where ρ is the density of the medium, cp the specific heat and necessary to provide a constitutive state equation ρ=ρ(p,T),
Δ stands for the Laplacian operator. The above linear like for example the ideal gas law, and to couple Eq. (4) with
equation is suitable to represent the temperature distribution the following modified version of Eq. (2), which takes into
for solids, liquids and gases under the hypothesis that no account the advection of the temperature:
movements or phase transitions occur inside.

CONVECTION
Convection is a heat transfer mechanism characterized by a (5)
mass flow: it is the fluid itself that with its movement carries
around the heat from different places of the medium. It can be The simplest way to include buoyancy effects into the
extremely efficient with respect to conduction as the fluid Navier-Stokes equations consists in following the so-called
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Boussinesq approach: the density is considered variable only

for what concerns its action as forcing term in the momentum
equation (4), where even small differences can be responsible (9)
for the triggering of the convective motion.
A final remark needs to be discussed for what concerns
RADIATION reflection, which can be specular or diffuse (or both, in
different percentages). Depending on the type of surface on
Radiation is a heat transfer phenomenon completely different
which an incident ray collides, it could be reflected into a
from the previous two, since it occurs between two surfaces
single outgoing mirror direction (specular reflection) or
at different temperature (even located distant from one
equally distributed in all the possible directions (diffuse
another) almost independently with respect to the interposed
reflection). Since we are dealing with lamps with really
medium, which could eventually be vacuum. The classical
marked optical properties, expecially for what concerns the
theory of radiation [22, 25] rests upon the balance between
parabolic internal surface of the body, specular reflection
the fraction a of total incident energy absorbed by a body, the
turns out to be really important to adequately describe the
fraction r of energy reflected and the fraction t of energy
reflection of the rays going from the bulb towards the lens.
transmitted (that is which passes through), which simply
writes (as functions of the different wavelengths λ):
STRESSES INDUCED BY HEAT LOADS
The influence of the temperature field, which can be
(6) computed according to the previously described heat transfer
mechanisms, on the body of the lamp can be discussed on
Moreover, for simple equilibrium considerations, it turns out two different levels. First of all, according to the type of
that the total amount of absorbed energy must be equal to the material of which the lamp is made of, there will be a
total amount of emitted energy ε, which often goes under the maximum threshold temperature Tmax above which melting
name of Kirchhoff's law of thermal radiation. This can be will occur. This can be assessed by analyzing the thermal
read in Eq. (6) as a balance between the emitted, reflected field alone, looking for the maximum temperature in the
and transmitted energy translating in the common fact that critical zones, which should not exceed Tmax.
poor reflectors are good emitters and vice versa. Under the
“black body hypothesis” α = ε = 1, the distribution function On the other hand, by the structural viewpoint, the thermal
e(λ,T) of radiant energy emitted at thermodynamic field can be seen as a heat load which gets into the whole set
equilibrium at the temperature T can be computed according of constraints that keep tied the component and connect it to
to quantum mechanics by means of Planck's law: the rest of the vehicle, generating deformations that could
potentially lead to damages of the lamp. In a linear isotropic
elastic material the constitutive relation between the tensorial
stress field σ and the tensorial strain field ε is given by [26]:

(7)
(10)
where h is Planck's constant, c0 is the speed of light and kb is
Boltzmann's constant. Integrating over all the wavelengths
the radiation spectrum one obtains that the total power E and v being the Young modulus and Poisson coefficient
emitted by a black body of area A at temperature T is respectively, α the coefficient of thermal expansion, Tref the
expressed by Stefan-Boltzmann's law: expansion reference temperature (namely the temperature at
which the thermal strain vanishes), tr(·) the trace operator and
δ the identity matrix. The first term of the right hand side of
Eq. (10) represents the elastic stress (due to mechanical
(8) loads), while the second one represents the thermal stress
(due to heat loads). Recalling that by definition
where σb=5.67·10−8 W/(m2K4) is the Stefan-Boltzmann
constant. However, this relation, which is true only for a
theoretical black body, needs to be corrected by weighting the
integral with the emissivity coefficients ε(λ)≠1. The “gray
body” hypothesis simplifies the computation assuming no (11)
dependence of ε on λ, which leads to our final formula for the
u being the displacement vector, the constitutive equation
radiant power output emitted by a surface of
(10) can then be employed in order to evaluate u from the
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Figure 3. Polyhedral mesh of the lamp and of the surrounding air.

solid momentum equation of the classical linear elasticity solution, it is recommended to leave at least 3-4 cells across
theory the thickness of the thin solid components of which the lamp
is made of, which can be achieved by defining there a so-
called “thin mesh” made of a predominantly structured
prismatic mesh together with a circumscribed unstructured
(12) one that has the scope of linking the former in the most tough
areas. Following such a strategy, it is possible to bound the
where ρ is the solid density and F is the body force vector. total number of cells even keeping a size of 2-4 mm on the
“flat” direction of components with a thickness of less then 1
METHODOLOGY OF ANALYSIS mm (i.e. the bulb). Finally, for what concerns the mesh of the
air, we define a suitable prism layer in proximity of all the
STEADY-STATE THERMO- no-slip walls (see Fig. 3).
MECHANICAL SIMULATION SETUP
The Navier-Stokes equations are solved with the SIMPLE
The numerical solution of the equations detailed in the above algorithm following a segregated approach [28], as well as
section has been carried out with a finite volume approach by for the thermal energy one. Natural convection is
means of the commercial CFD package Star-CCM+ [27], contemplated by activating the ideal gas model (or the
capable of handling conjugate heat transfer phenomena Boussinesq one), which links the fluid-dynamic solution to
between different bodies as well as radiation and solid stress. the thermal one. If the lamp is closed and only inner air needs
For the sake of simplicity, we assume the mechanical analysis to be modeled, turbulence effects are quite irrelevant and
to be decoupled from the thermo-fluid dynamic one: the laminar flow is sufficient. On the other hand, if external air is
thermal field obtained as output of the latter is given as input included, the flow can be suitably described in a RANS
to the former. fashion by means of the Realizable k-ε turbulence model with
“all y+” wall treatment.
The finite volume discretization takes advantage of the
possibility to employ generic polyhedral cells, which For what concerns radiation heat transfer, it is fundamental to
represent the most flexible and cost-effective type of cells. identify the boundaries and interfaces involved in the process.
Moreover, in order to guarantee a proper accuracy to the
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Their faces are grouped into “patches”, thus setting up the

discretization of radiation. Following the “gray body
assumption”, on each boundary it is required to define the
proper coefficients of emissivity, reflectivity and
transmissivity. Moreover, for this kind of application it is
fairly important to take advantage of the specular reflection
model, according to which the view factors are calculated by
ray-tracing.

The boundary conditions of the problem are quite

straightforward to define, at least for what concerns the
outside of the lamp: if the external air is modeled, we simply Figure 4. Scheme of the filament-bulb radiation
impose the temperature on its boundary, while in absence of interactions.
external air it is possible to set a Robin convection condition
with a fixed heat transfer coefficient. It is also possible to
extend it to the case of more complex situations, like for The above scheme can be easily solved by taking into
example putting the lamp inside a ventilated oven. The account the definition of emissivity and transmissivity of the
modeling of the source of power on the filament(s) and inner surface of the bulb
bulb(s) is less trivial and will be explained in the next section.
A steady-state simulation for a model of about 1-2 millions
cells can be carried out in a few hours taking advantage of (13)
parallel computation.
the balance of fluxes on the bulb at steady state
FILAMENT-BULB RADIATION
Let us now deal with the heat transfer phenomena occurring
inside a light bulb, in order to provide a methodology to (14)
accurately model the source of power. The goal is to follow a
the balanced subdivision of the emitted radiation flux
conservative approach imposing on the pair “filament/bulb”
between the external and the internal side of the bulb (that
exactly the total power P desired, without setting any
should be true for a thin glass)
Dirichlet boundary condition on the temperature. In the
following we present a theoretical-numerical comparison that
stands for both a validation of the method and a verification
(15)
of the software.
and the definition of emissivity and transmissivity of the bulb
Referring to Fig. 4, we consider a filament (F), supposed to inner surface applied to the radiation emitted by the bulb
be a point source of power, a thin partially transparent bulb towards itself
(B) and an external ambient environment (A) that absorbs all
the radiation that arrives. Note that between the filament and
the bulb there is nothing (vacuum). Let us neglect for the
moment the reflectivity of the bulb, and consider only its (16)
emissivity ε and its transmissivity t, such that ε + t = 1.
Eq. (13), (14), (15), (16) are 6 equations in 5 unknowns,
The radiation emitted by the filament, corresponding to the which show to be consistent with respect to the connection
whole power dissipated on it, gets partly absorbed by the bulb between ε and t given by ε + t=1. The physically most
(Pfb), while the remaining part pass through it towards the relevant quantity associated with this scheme is the net
ambient (Pfa). The bulb emits both in direction of the ambient radiation heat flux Qrad absorbed by the bulb, which turns out
(Pba) and in the opposite one (inside): this latter portion of to be (as a function of the transparency t):
radiation is again split into two parts, one being re-absorbed
by the bulb (Pbb) and the other one going across it (Pbba).

(17)

The schematic configuration described above has been

implemented on a numerical model (of course without
satisfying exactly the assumptions of pointwise filament and
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

infinitely thin bulb) and solved for the steady-state • Variable timestep: a full implicit unsteady simulation is
configuration in Star-CCM+. The results are displayed in Fig. performed, in which the timestep is not kept fixed but instead
5, which shows really good agreement between the adapted in order to have a fast advance when allowed by the
theoretical values and the numerically-obtained ones. temperature variations.

Let us give a brief description of how it is possible to set up a

variable timestep unsteady simulation by suitably scripting
the time advancing algorithm. For an example of application
of this technique on the industrial rear lamp described in this
paper, we demand to the Results section later on. Let us call
t0,t1,t2,…,tn,… the discretized time instants, at which the
solution in terms of temperature and velocity is (T0, v0), (T1,
v1), (T2, v2),…, (Tn, vn),…, the timestep at time tn being Δtn
= tn − tn−1.

The key idea of the advancing scheme is to use the computed

Figure 5. Behavior of the radiation heat flux on the bulb solution up to the time tn in order to update the timestep Δtn
Qrad as a function of the transparency t, for the +1, the objective being to perform the “longest” possible step

theoretically-computed values (solid line) and the values which guarantees a certain fixed accuracy. To this aim we
obtained by numerical simulation (dots). define the two monitors

TRANSIENT ANALYSIS TECHNIQUES

As already discussed in the introduction, transient analyses (18)
are really useful when trying to reproduce complex testing
conditions with different lights turning on and off at
scheduled times (eventually of the order of dozens of
minutes). Most commercial CFD packages allow to
(19)
numerically reproduce a transient by means of an implicit
unsteady simulation in which at each timestep the new that can be interpreted as the “average temperature variation”
solution is computed with several “inner iterations” at the and the “maximum velocity variation” occurred during the
same manner as for a steady state computation. In order to previous timestep n. Then it is possible to set reasonable
accurately solve the whole transient, the timestep must be values for the desired average temperature variation δT* as
taken small enough to be able to capture the rapid variations well as for the desired maximum velocity variation δv*, thus
of the temperature field on the bulb in the power on and using the functions
power off instants (eventually of the order of fractions of
seconds). Unfortunately, for grids of several millions cells
this results in a fairly huge computational cost, often leading
to unfeasible simulations.
(20)
To overcome the problem it is possible to employ one of the
following strategies, ordered by increasing complexity (and to update the timestep, according to the simple rule that if ηn
accuracy): > 1 and ζn > 1 than the timestep can be increased (since the
variation of the solution is less than the maximum one
• Scaled power: the unsteady computation is replaced by a accepted), otherwise it must be reduced. If the above
standard steady-state one in which the power provided on the condition is satisfied, then dealing with the thermal field the
filament(s) is replaced by the weighted averaged according to
the lighting and standby times. ideal timestep would be , while dealing with
• Freeze velocity: the transient simulation is performed the velocity field it would be Hence we set
solving only the thermal equation and keeping fixed (i.e. n n n
ζ = min{η , ζ } and update the timestep with the formula
“freezing”) the velocity field previously computed for an
equivalent scaled power steady-state solution. This allows to
take into account buoyancy-related effects in an
approximated way, but without fully computing the whole (21)
fluid-dynamic solution.
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Figure 6. Experimental apparatus and its numerical reproduction. The distance between the bulb and the body of the box has
been deliberately taken high, in order to highlight buoyancy effects.

β being a suitable under-relaxation factor. If instead of measured ambient temperature (that in the numerical
increasing the timestep, one establishes it to be reduced, we simulation has been imposed on the boundaries of the
set µn = max{ηn,ζn} and replace (21) with external air) was Tenv = 298 K.

The quantitative comparison of the maximum temperatures

on the four significant faces of the cube is detailed in Tab. 1,
(22) showing good agreement except that on the lateral face
(although not exceeding 5 K difference). The reason could be
As a final remark in view of a cheap simulation, we observe associated to a non correct modeling of the filament
that wherever possible the external air should be removed geometry: the toroidal surface employed in this case could be
from the simulation, leaving as heat transfer coefficient responsible of an uncorrect directioning of the radiation
exactly the one computed by an equivalent steady-state towards the lateral faces.
solution.
By means of the numerical simulation it is also possible to
VALIDATION OF THE carry out a detailed heat flux analysis, showing perfect
conservation of energy:
METHODOLOGY
In view of a numerical/experimental comparison for the • Power dissipated on the filament: 16.00 W
validation of the complex thermo-fluid-dynamic • Heat flux absorbed by the internal surface of the bulb: 3.08
methodology described above, we have considered a really W
simple prototype of lamp, consisting of a cubic box of 86 mm
with thickness 3 mm made of Lexan, and of a W16 lightbulb • Heat flux on the external surface of the bulb towards the air:
sealed up inside the box (see Fig. 6). The same geometry has 2.83 W
been reproduced for the numerical simulation, leading to a
model of about 300 000 polyhedral cells (including internal • …of which by radiation: 0.68 W
and external air). • Heat flux exiting from the box towards the environment:
15.97 W
A Flir ThermaCAM T360 thermo camera has been employed
to investigate the temperature contours at steady state • …of which by radiation: 8.48 W
(namely after 30′) when the bulb is powered with 16W. The
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Figure 7. Numerical/experimental comparison of the steady state temperature field on the top face.

Table 1. Numerical/experimental comparison.

Figure 8. Exploded view of the analyzed industrial lamp.

RESULTS FOR AN INDUSTRIAL The original CAD geometry has been suitably simplified and
TEST CASE arranged in order to correctly define all the interfaces
Let us now consider as industrial test case for the application between the different components (see Fig. 1). The final 3D
of the complete methodology the rear lamp of Fig. 8. It mesh generated by Star-CCM+ (see Fig. 3) consists of about
consists in a multiple service rear lamp with a foglamp 700,000 polyhedral cells, which has been employed to carry
service in the center position and two lateral bulbs employed out the simulation on a 4-cores 64 bit Windows workstation.
as backup lights. The housing is made up of polycarbonate One of the most severe tests through which the lamp comes
while the lens is in polymethacrylate. The circuit holder in consists in providing full power on the foglamp (16 W) for an
polypropylene supports the three lamp sockers in poliammide hour, while turning on the backup lights for 5 minutes and
PA66, each of them holding a W16 lightbulb. Moreover, a then off with period 20 minutes, at ambient conditions.
grid in PC+ABS has been set just behind the lens for
aesthetic purposes (see Fig. 8 for an exploded view of the real Let us first of all consider a “scaled power” steady-state
lamp). The geometry analyzed in this study, which consists in simulation, in which the foglamp is powered with 16 W and
an early step of the design of the lamp, suffers of several the two backup lights each with 4 W. The temperature field
criticalities on the internal part of the body near the central on the lamp is shown in Fig. 9, highlighting the peaks on the
bulb and on the portion of the lens just in front of the backup top of the body associated with buoyancy effects. The motion
lights, which melts if all the lights are set full power for more of the air (both internal and external) can be visualized by
than 15′. means of vectors or streamlines, as illustrated in Fig. 10,
which allows to better understand the importance of natural
convection on the overall heat transfer. Fig. 11 provides an
interesting comparison between the temperatures on the lens
measured with the aid of the thermo camera and the ones
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Figure 9. Contour plot of the temperature of the lamp.

Figure 10. Visualization of the internal and external air motion associated with natural convection.

coming from the simulation. The agreement if fairly good

both qualitatively and quantitatively, with just a slight
difference in the form of the peak associated with the
foglamp, which actually depends on a little misalignment of
the parabola carried out on the real geometries for optical
reasons that causes a duplication of its geometrical focus and
that for the sake of simplicity has not been reproduced in the
numerical model.
Figure 11. Comparison of the measured and simulated
temperatures of the lens.

As an example of result for the stress analysis, consider the

displacement magnitude contour plot of Fig. 12, obtained by
constraining the back portion of the circuit holder and
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

Figure 12. Displacement magnitude contour plot.

applying the thermal load computed by the steady-state off event raises, till another of such events occurs. With the
analysis. adaptive timestep algorithm the total number of advancing
time iterations carried out is about 700, which is less than
Without any doubt the steady-state analysis provides really 10% of the number of the same that would have been
useful results, already showing the local criticalities to which required by keeping it fixed to its minimum value.
the lamp is subject. However, for a most accurate analysis it
is necessary to follow the whole transient of the power on and
power off events, which has been carried out by means of the
adaptive timestep algorithm described previously. The plot of
Fig. 13 gives the evolution of the temperatures on three
points of the lens, showing that the most sensible zone turns
out to be that located just above the backup lights (blue curve
in the plot), which actually reaches the melting temperature
of the material after about 15′ (note that this test is rather
severe and that the lamp we are dealing with is not the final
design).

Figure 14. Evolution of the timestep in the adaptive

algorithm for the unsteady simulation.

SUMMARY
A comprehensive CFD-based methodology for the numerical
simulation of the thermo-mechanical behaviour of rear lamps
has been developed and discussed. It embraces in a fairly
accurate way all the three fundamental heat transfer
mechanisms: conduction between solids, natural convection
associated with buoyancy effects, and radiation. The latter
Figure 13. Evolution of the temperature on three sample
includes a suitable treatment of specular reflection and
points on the lens for the unsteady simulation (red line:
transparency effects, which are fundamental issues that need
in front of the foglamp, blue line: in front of one of the
to be taken into account for an accurate reproduction of the
backup lights, green line: halfway between the previous
behavior of lamps.
two).
A particular attention has been directed to the description of
Focusing on the numerical point of view, the evolution of the the source of power, that is to the filament-bulb interaction. A
timestep is reported in Fig. 14, clearly showing how it simple schematic configuration allowed us to verify the code
increases as the temporal distance between the last switch on/ by comparing the output with analytical results. A major
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

validation has been then carried out on a simple prototype of The PIV Measurement and the CFD Simulation By Using a
a lamp, showing good agreement between simulation and Skeleton Model,” SAE Technical Paper 2000-01-0802, 2000,
reality. doi:10.4271/2000-01-0802.
12. Shiozawa, T., Yoneyama, M., Sakakibara, K., Goto, S.,
Dealing with unsteady simulation, a peculiar algorithm has
Tsuda, N., Saga, T., Kobayashi, T., “Thermal air flow
been devised, capable of adapting the timestep in order to set
analysis of an automotive headlamp: the PIV measurement
it to the highest possible gain at each instant, actually
and the CFD calculation for a mass production model”, JSAE
optimizing the computational costs. Both the steady-state and
Review 22, pp. 245-252, 2001.
the unsteady simulation has been finally tested on a real
complex industrial lamp. 13. Moore, W.I., Donovan, E.S., Powers, C.R., “Thermal
analysis of automotive lamps using the ADINA-F coupled
REFERENCES specular radiation and natural convection model”, Computers
and Structures, Vol. 72 (1), July 1999, pp. 17-30(14), 1999.
1. Langebach, J., Senin, S., Karcher, C., “Experimental study
of convection and radiation interaction in a headlight model 14. Moore, W.I. and Powers, C.R., “Temperature Predictions
using pressure variation”, Experimental Thermal and Fluid for Automotive Headlamps Using a Coupled Specular
Science 32, pp. 521-528, 2007. Radiation and Natural Convection Model,” SAE Technical
Paper 1999-01-0698, 1999, doi:10.4271/1999-01-0698.
2. Sousa, J.M.M., Vogado, J., Costa, M., Bensler, H., Freek,
C., Heath, D., “An experimental investigation of fluid flow 15. Moore, W.I. and Hilburger, F.K., “Development of a
and wall temperature distributions in an automotive CFD Model for Simulating Headlamp Humidity Clearing,”
headlight”, International Journal of Heat and Fluid Flow 26, SAE Technical Paper 2001-01-0861, 2001, doi:
pp. 709-721, 2005. 10.4271/2001-01-0861.

3. Moore, W.I., Donovan, E.S., Powers, C.R., “Recent 16. Moore, W.I. and Powers, C.R., “Using CFD for
Advances in MSC/PATRAN Pre-Processing Software Humidity Clearing Simulation of a Composite Headlamp,”
Allows Modeling of Complex Automotive Lamp Designs”, SAE Technical Paper 2000-01-1598, 2000, doi:
MSC Americas Users' Conference, Universal City, 1998. 10.4271/2000-01-1598.
4. Nolte, S., Eine Methode zur Simulation der Temperatur- 17. Poppner, M., Maschkio, T., “Development of a Software
und Stromungsverteilung in lichttechnischen Geraten, Dr-Ing Tool for the Simulation of Formation and Clearance of
Thesis, Universitat Paderborn, 2005. Condensation in Vehicle Headlamps”, PAL 2003 - Progress
in Automobile Lighting, Darmstadt, 2003.
5. Fischer, P., “CFD Analysis and Experimental Verification
of an Automotive Fog Lamp,” SAE Technical Paper 18. Preihs, E., “Automotive Headlamp - Analytic Solution
2005-01-1921, 2005, doi:10.4271/2005-01-1921. and Measurements of Condensation inside a Headlamp”,
Nordic COMSOL Conference, Copenaghen, 2006.
6. Moore, W.I. and Powers, C.R., “Utilizing CFD for
Thermal Analysis of a Prototype HID Fog Lamp,” SAE 19. Shiozawa, T., Yoneyama, M., Sakakibara, K., Goto, S. et
Technical Paper 2000-01-0801, 2000, doi: al., “Analysis of Natural Convection Inside an Automotive
10.4271/2000-01-0801. Headlamp by Using CFD Temperature Prediction of the Parts
7. Okada, Y., Nouzawa, T., Nakamura, T., “CFD analysis of by SHT Method,” SAE Technical Paper 2001-01-0860, 2001,
the flow in an automotive headlamp”, JSAE Review 23, pp. doi:10.4271/2001-01-0860.
95-100, 2002. 20. Shiozawa, T., Ohishi, M., Yoneyama, M., Sakakibara, K.
8. El-Khatib, F., Bielecki, J., and Poorman, T., “Design of et al., “Analysis of Moisture and Natural Convection Inside
Experiment Analysis of Thermal Variables that Affect An Automotive Headlamp By Using CFD,” SAE Technical
Automotive Lighting CFD Temperatures,” SAE Technical Paper 2005-01-1449, 2005, doi:10.4271/2005-01-1449.
Paper 2006-01-0490, 2006, doi:10.4271/2006-01-0490. 21. Cimolin, F., Menotti, A., Rabito, M., Vadori, R., “Analisi
9. Poorman, T., Bielecki, J., Chang, M., Belsare, S. et al., termo-meccanica contecniche combinate strutturali e CFD del
“Automotive Lighting Thermal Performance Prediction comportamento di un fanale autoveicoli stico in condizioni di
Methods,” SAE Technical Paper 2001-01-0858, 2001, doi: prova e di utilizzo”, Proceedings of the XXXVII National
10.4271/2001-01-0858. Congress of AIAS, Rome, 2008.

10. Poorman, T., Bielecki, J., and Chang, M., “Methodology 22. Lienhard, J.H.IV, Lienhard, J.H.V, “A heat transfer
to Evaluate Automotive Lamp Venting,” SAE Technical textbook”, Third Edition, Phlogiston Press, 2006.
Paper 2003-01-0648, 2003, doi:10.4271/2003-01-0648. 23. Tritton, D.J., “Physical Fluid Dynamics”, Second
11. Shiozawa, T., Nakanishi, A., Ozawa, T., Oki, T. et al., Edition, Oxford Science Publications, 1988.
“Thermal Air Flow Analysis of an Automotive Headlamp -
 Downloaded from SAE International by Univ of California Berkeley, Sunday, August 05, 2018

24. Batchelor, G.K., “An Introduction to Fluid Dynamics”,

Cambridge University Press, 2000.
25. Siegel, R., Howell, J.R., “Thermal Radiation Heat
Transfer”, Third Edition, Hemisphere Publishing Co, 1992.
26. Sadd, M.H., “Elasticity: theory, applications, and
numerics”, Second edition, Academic Press, 2009.
27. Star-CCM+ User Guide, CD-Adapco, 2007.
28. Ferziger, J.H., Perić, M., “Computational methods for
fluid dynamics”, Third edition, Springer, 2002.

CONTACT INFORMATION
Corresponding author:

Flavio Cimolin
AMET S.r.l.
Environment Park, Via Livorno 60, 10144 Torino, Italy.
[email protected]

ACKNOWLEDGMENTS
The authors wish to thank the support of CD-adapco in the
names of Anthony Massobrio and Lucia Sclafani, for their
fundamental aid in understanding even the most subtle
peculiarities of the software STAR-CCM+ in order to
conveniently reproduce the complex heat transfer phenomena
occurring inside a lamp.

The Engineering Meetings Board has approved this paper for publication. It has Positions and opinions advanced in this paper are those of the author(s) and not
successfully completed SAE's peer review process under the supervision of the session necessarily those of SAE. The author is solely responsible for the content of the paper.
organizer. This process requires a minimum of three (3) reviews by industry experts. SAE Customer Service:
Tel: 877-606-7323 (inside USA and Canada)
All rights reserved. No part of this publication may be reproduced, stored in a
Tel: 724-776-4970 (outside USA)
retrieval system, or transmitted, in any form or by any means, electronic, mechanical, Fax: 724-776-0790
photocopying, recording, or otherwise, without the prior written permission of SAE. Email: [email protected]
ISSN 0148-7191 SAE Web Address: http://www.sae.org
Printed in USA