You are on page 1of 101

2007:128 CIV

M A S T ERS T H E SI S

Including Solar Load in CFD


Analysis of Temperature Distribution
in a Car Passenger Compartment

JONAS JONSSON

MASTER OF SCIENCE PROGRAMME


Mechanical Engineering
Lule University of Technology
Department of Applied Physics and Mechanical Engineering
Division of Fluid Mechanics

2007:128 CIV ISSN: 1402 - 1617 ISRN: LTU - EX - - 07/128 - - SE

Acknowledgements
This thesis was carried out at the CFD centre at Volvo Cars in Gothenburg, Sweden, with the
supervision of PhD sa Uddheim. I would like to give special thanks to this hard working,
recently mother hooded woman, whom this project would have been nothing without. She has
helped me in any way she could possibly have and guided me to in the end satisfactory
results.
I would also give thanks to all the other highly qualified CFD engineers at this department.
Especially I would like to give thanks PhD Andreas Borg for his theoretical expertise and
PhD Robert Moestam for many long and helpful discussions regarding my work.
I would also show my gratitude towards the very helpful support staff at FLUENT Sweden.
Robert Nyreide has been a key person in helping me understand the many settings and models
used in the software FLUENT and T-grid, but also in analysing the many problems and speed
bumps along the way.
My academic supervisor Staffan Lundstrm has not been a large part in my thesis work but is
a person that would without doubt have helped me if help was needed, and for that I want to
thank him.
Help and support is of course important professionally but even more personally. Without the
people closest to me I dont think this thesis work could have been possible without running
into a hard, bone breaking wall. I would like to bow before all my close friends that have kept
me above water with many long encouraging Thursday dinners and helpful conversations.
Salut!
Last but not least I would like to say: thank you family! Without you nothing would be
possible. Love you!

Abstract
In this thesis a numerical model has been developed that integrates solar load when
computing the temperature distribution in a passenger compartment. The actual compartment
used belongs to a Volvo S80 for which geometry, mesh and results from tunnel tests already
exists. The thesis is divided into two sections, model development and model validation. In
the former study focus is set on means to incorporate the solar load in a proper way and to
understand the software used. In the latter study the numerical procedure derived is applied to
the test case, the Volvo S80, and the results obtained are compared to wind tunnel tests. Three
cases were simulated throughout the thesis: a steady state comfort, a transient soaking, and a
transient cooling.
It is found that the temperature distribution for the steady state case is strongly dependent on
the solar load while realistic results from the soaking case were not obtained due to severe
divergence problems regarding continuity. These problems can be attributed to a very coarse
mesh and mesh discrepancies in fluid zones inside doors, floor and roof. The temperatures at
several positions at the back were in many of the comfort cases much higher than
experimental data being most likely an effect of solid cells in the back seat. For the cooling it
is even more important to use the model developed. The validation of the model yielded that
the steady-state cases capture the trends in temperature within the compartment usually within
a few degrees. The validation of the cooling simulations was very successful except in the
front toe position for the driver. It was also shown that the systems thermal inertia could be
increased by increasing the thickness in the sensitive areas to more realistic total masses. It is
finally found that the thermal inertia greatly affects the cooling of the compartment and is one
of the more important properties to focus on in future work.

Table of Contents
TABLE OF CONTENTS ................................................................................................................................... 1
1 INTRODUCTION ........................................................................................................................................... 3
1.1 PREVIOUS WORK ......................................................................................................................................... 3
1.2 OUTLINE OF THESIS WORK ......................................................................................................................... 4
2 THEORY........................................................................................................................................................... 5
2.1 GOVERNING EQUATIONS ............................................................................................................................. 5
2.2 BUOYANCY USING THE BOUSSINESQ APPROXIMATION .............................................................................. 6
2.3 CONVECTIVE HEAT TRANSFER COEFFICIENT ............................................................................................. 7
2.4 TURBULENCE ............................................................................................................................................... 7
2.4.1 Reynolds Decomposition........................................................................................................................... 7
2.4.3 Boussinesq Hypothesis the Eddy Viscosity Model .................................................................................... 8
2.4.4 K-epsilon Realizable ............................................................................................................................... 8
2.4.5 Wall Functions ....................................................................................................................................... 9
2.5 SOLAR LOAD MODEL ................................................................................................................................. 11
2.5.1 Solar Ray Tracing Algorithm.................................................................................................................. 11
2.5.2 Surface to Surface Radiation Model (S2S)................................................................................................ 12
3 METHOD MODEL DEVELOPMENT................................................................................................... 13
3.1 MODEL GEOMETRY ................................................................................................................................... 13
3.2 NUMERICAL SETUP .................................................................................................................................... 14
3.3 SOLAR AND RADIATION SETTINGS ............................................................................................................ 16
3.4 CASES SIMULATED ...................................................................................................................................... 18
3.4.1 Simulation Case: Steady state Comfort.................................................................................................... 18
3.4.2 Simulation Case: Soaking ...................................................................................................................... 19
3.4.3 Simulation Case: Cooling ...................................................................................................................... 19
4 METHOD MODEL VALIDATION ........................................................................................................ 21
4.1 WIND TUNNEL TEST .................................................................................................................................. 21
4.1.1 Measurement Points............................................................................................................................... 22
4.1.2 Test Cases............................................................................................................................................ 23
4.1.3 Conclusions from Wind Tunnel Test....................................................................................................... 24
4.2 NUMERICAL SIMULATIONS ........................................................................................................................ 25
4.2.1 Simulation Procedure ............................................................................................................................. 25
4.2.2 Simulation Case: Steady State Comfort................................................................................................... 27
4.2.3 Simulation Case: Soaking ...................................................................................................................... 27
4.2.4 Simulation Case: Cooling ...................................................................................................................... 28
4.3 VERIFICATION OF SIMULATIONS ................................................................................................................ 28
5 RESULTS MODEL DEVELOPMENT ................................................................................................... 30
5.1 SIMULATION CASE: STEADY STATE COMFORT .......................................................................................... 30
5.2 SIMULATION CASE: SOAKING ..................................................................................................................... 32
5.3 SIMULATION CASE: COOLING .................................................................................................................... 35
6 RESULTS MODEL VALIDATION......................................................................................................... 39
6.1 SIMULATION CASE: STEADY STATE COMFORT .......................................................................................... 39
6.1.1 Comfort Loop 1 and 3........................................................................................................................... 39
6.1.2 Comfort Loop 2 and 4........................................................................................................................... 42
6.2 SIMULATION CASE: SOAKING ..................................................................................................................... 44
6.3 SIMULATION CASE: COOLING .................................................................................................................... 45
6.3.1 Cooling Loop 1..................................................................................................................................... 45
6.3.2 Cooling Loop 2..................................................................................................................................... 46
6.3.3 Cooling Loop 3..................................................................................................................................... 47
6.3.4 Cooling Loop 4..................................................................................................................................... 49

7 DISCUSSION.................................................................................................................................................. 54
7.1 RECOMMENDATIONS ................................................................................................................................. 55
8 CONCLUSIONS............................................................................................................................................. 56
8.1 FUTURE WORK .......................................................................................................................................... 57
9 REFERENCES ................................................................................................................................................ 58
APPENDIX 1: BOUNDARY CONDITIONS FROM OLD SIMULATIONS ....................................... 59
APPENDIX 2: BOUNDARY CONDITIONS IN VALIDATION SIMULATIONS ............................. 59
APPENDIX 3: FLUENT MODEL SETTINGS............................................................................................ 60
APPENDIX 4: JOURNAL FILES ................................................................................................................... 66

1 Introduction
In the automotive industry it is important to be able to optimize passenger comfort. One
critical situation is when a car has been parked in the sun and has to quickly be cooled down.
The average temperature in the vehicle has during the solar exposure become very high and it
is important to efficiently cool down the cabin hence improving the comfort of the passengers.
One approach to optimise this process is by using numerical methods, as Computational Fluid
Dynamics (CFD), when developing the climate systems. By using CFD improvements can be
initiated at a much earlier stage and climate evaluations could be performed before the car has
been built, resulting in a substantial reduction in both development costs and time. CFD is for
the moment not commonly used when evaluating passenger comfort of this sort but will in the
future be a natural part of the process.
The improvement of climate control systems here at Volvo Cars has so far mostly been
dominated by tests performed in a climate wind tunnel. Thermal couples are positioned inside
the compartment and studied during different cooling test cases. Numerical simulations have
been used as a complement but without including the effects of solar load. The large seize of
the computational mesh and the number of parameters needed to describe the passenger
compartment has made a numerical model describing the flow very demanding. This includes
both the requirement of computational resources and the development time of the numerical
approach. Also, including a heat source as solar load in these models is alone complicated and
has up till now been difficult to manage. This has mostly been because of lack of appropriate
solar and radiation models to use or integrate in the commercial software. However, the
climate in a car is highly dependent on the level of solar exposure and it is therefore important
to include this in a numerical analysis.

1.1 Previous Work


Previous work in this area of computational fluid dynamics (CFD) has mostly been performed
using very simplified geometrical models and/or in-house software. Correlation between test
data and numerical data has however in a few known cases been achieved with an error
margin of only a few degrees [1, 3]. The most difficult part in a solar soaking and cooling
simulation cycle seems to be the hot soaking case where the flow is strictly driven by
buoyancy and convective heat transfer [1]. Another difficulty in the analysis is to correctly
model all the thermal boundary conditions and also determine the most essential materials and
their properties to be used in the model.
Work here at Volvo Cars has until now not been too extensive regarding modelling solar load
and thermal comfort. Studies were performed in early 2000 [4] by Zenitha Chroner to model
the temperature distribution and the cool down process in the compartment, but without
including the effects from solar radiation. Moreover, the results did not fully manage to
describe the cooling process. The cooling simulations had similar behaviour as experimental
data but with a tendency to accelerate the process, i.e. the compartment was cooled faster in
the simulations compared to in the tests. The geometry used in this analysis was modelled
with double surfaces in floors, doors and roof to better capture the insulation of the
compartment. This was done because the temperatures in the foot region were lower than in
the rest of the compartment (the same compartment geometry has been used in the present
thesis work).

A thermal comfort study has previously been performed at Volvo Cars [2]. The study was
carried out as a thesis project to investigating the temperature and velocity distribution in a
passenger compartment and its effects on thermal comfort. This study did not account for the
effects from an external solar heat source but only discussed the effects of draught rates and
other comfort related factors. However, one attempt has been done to include solar load in a
passenger compartment simulation, [9]. The computational domain was extremely simplified
and described mostly the driver section of the compartment. The first attempt in this study
was to use a user defined function (UDF) provided by the FLUENT support. The UDF was to
be used in a beta version, but did not meet the demands or goal for the implementation.
Another approach in this study was to use both commercial software FLUENT and RadTherm
to include the radiation. The combination gave results showing a clear difference for
simulations with and without sun but the process was very demanding and not sufficiently
robust when looping was needed to reach convergence.

1.2 Outline of Thesis Work


This thesis is focused on developing a method of implementing the existing Solar Load Model
in the 6.2.16 version of FLUENT on a Volvo S80 model. The goal is to optimize the use of
the solar load model and to improve the numerical model to the extent of correctly simulating
trends in temperature distribution in the compartment due to alteration in configuration. Three
simulation cases are studied in this thesis: stationary simulations representing a test after a
long time, transient soaking simulations, and transient cooling simulations. The numerical
results are compared to experimental results from old wind tunnel tests and a test performed
during the work. These experiments were used to validate the numerical simulations and
thus the numerical model. When the goal with the thesis is to correctly simulate trends within
5 degrees compared to test the validation works more as a pilot study than a complete
validation of the model.

2 Theory
In this chapter a brief summary of the theory and approximations used in this thesis will be
presented. The governing equations of fluid mechanics are described as well as the realizable
k-epsilon turbulence model. Also, the models used in FLUENTs version 6.2.16 to include
solar radiation are presented and shortly explained.

2.1 Governing equations


The governing equations in fluid dynamics are the Navier-Stokes equations [6]. Continuity
(2.1), momentum (2.2) and energy equations (2.3) are in CFD solved with various
computational approaches.

u i
=0
+
xi
t
u i u j
+

x j xi
T
T Tu j
k
=
+
c p x j x j
x j
t
u i u i u j
P

+
=
+
t
x j
xi x j

(2.1)

(2.2)
(2.3)

where is the density, the dynamic viscosity, u the velocity, P the pressure, T the
temperature, t is time, xi the x, y, z-position tensor, cp the specific heat capacity and k the
thermal conductivity.
For incompressible flow the density does not change with time and can then be written in the
form,

u i
=0
xi
u i u j
+

x j xi
Tu j
T
T
k
=

+
c p x j x j
x j
t

u i u j
u i
P

+
=
+
t
x j
xi x j

(2.4)

(2.5)
(2.6)

2.2 Buoyancy Using the Boussinesq approximation


Buoyancy driven flows are caused by density changes within the fluid domain. Combined
with a dependency of body forces, e.g. gravitation or centrifugal forces, the fluid particles will
gain momentum and thus change position. When the fluid flow is mostly due to temperature
differences in the boundary layer it is called free convection. The warm fluid will have lower
density than the cold and will then rise in a negative gravitational direction. This flow
behaviour can be described with the equation
u

2u
g
u
u
,
+v
= ( ) +

y 2
x
x

(2.7)

where the first term on the right side is the buoyancy force caused by the density variation.
The equation above is derived through assuming a steady, 2-dimensional, incompressible flow
with constant properties except for density change in the buoyancy forces [2]. These
assumptions are the base of what is called the Boussinesq buoyancy approximation. The
equation is for motion in x-direction (negative gravitational direction) and with the pressure
gradient substituted with g .
As mentioned before, the density variation is due to the temperature variation in the fluid and
it is therefore necessary to include the temperature in equation (2.7). This is done by
introducing the thermal expansion coefficient . This is a measure of how much the density
changes with variation in temperature at constant pressure.

1 p
T

1
T T

(T T )

(2.8)

The final momentum equation in gravitational direction then reads:

2u
u
u
u
+v
= (T T ) +
y 2
x
x

(2.9)

For an ideal gas is simply the inverse of the total temperature. As seen in equation (2.9) the
thermal expansion coefficient has large importance on the extent of influence the density
change will have on the fluid flow. For an ideal gas the value of can be calculated as
follows:

p
RT

(2.10)

(2.10) in (2.8) gives

1 p
T

=
p

1
p
=
2
RT
T

(2.11)

2.3 Convective Heat Transfer Coefficient


The convective heat transfer coefficient is a measure of how much the convective heat flux
depends on the temperature difference on the surface and in the fluid. It is used to describe the
convective heat transfer process and the equation formulation is:
q = h(Ts T )

(2.12)

Where q is the convective heat flux, h the heat transfer coefficient, Ts the temperature on the
surface and T the temperature in the fluid (in FLUENT the free stream temperature).
The heat transfer coefficient can be calculated using theory from flow over a 2-dimensional
plate, see equations (2.13, 2.14). The coefficient is described as a function of the Prandtl
number (Pr), the Nusselt number (Nux), the plates conductivity (k), the plate length (x), the
fluid viscosity (v), and the velocity above the plate (U) [5]:
Nu x =

hx
k

(2.13)
1
3

Ux
Nu x = 0.0288 Pr
v

0 .8

(2.14)

By calculating the heat transfer coefficient using this approach the velocity outside of the
compartment can easily be integrated in the present numerical model. The level of heat
transfer is clearly dependent of the velocity and is important to include.

2.4 Turbulence
Fluid flow is divided into two main categories: laminar flow and turbulent flow. Turbulent
flow is difficult to describe using numerical methods because of its time dependence and high
variations in flow due to fluctuations. The flow is usually characterised as turbulent when it
has reached a certain Reynolds number for that specific scenario, e.g. for flow in a pipe
turbulence is said to arise when 2000 < Re < 2500 [6]. To solve the governing equations
without making simplifications and approximations is difficult and very resource demanding.
This leads to different approximate approaches to in a both robust and low error way simplify
the solving process. There are numerical methods solving the 3D Navier Stokes equations
directly, e.g. DNS (Direct Numerical Simulation). However, these numerical approaches are
unprofitably time consuming and hence not used in industrial analysis.

2.4.1 Reynolds Decomposition


In the Reynolds decomposition the instantaneous variables in the governing equations are
decomposed into one fluctuating and one averaged part. The velocities and pressure will then
read
u i = U i + u i
p = P + p ,

(2.15)
(2.16)

where U denotes the mean value and ui the fluctuation for both variables.

The time averaged continuity and momentum equations are obtained by substituting (2.15)
and (2.16) into (2.4) and (2.5) and averaging the final equations. The result will be Reynolds
averaged Navier Stokes (RANS) and are written as

U i
=0
xi

(2.17)

U i
U i

U i
P+
+ U j
=
u iu j

t
x j
xi
x j x j

(2.18)

The term uiu j is called the Reynold stress tensor and describes the correlation between the
fluctuating velocities of the turbulent flow. Without taking in account for the Reynold stress
tensor the momentum equation combined with the continuity equation contains four
unknowns and four equations and can thus be solved without a problem. The stress tensor
contributes to another six unknowns to be solved, making it a total of ten unknowns and only
four equations. Assumptions and approximations must now be made regarding the nature of
the turbulent flow to be able to obtain a solution to this system of equations.

2.4.3 Boussinesq Hypothesis the Eddy Viscosity Model


One common method of relating the Reynold stresses mentioned above to the mean velocity
gradients is through the Boussinesq hypothesis. The stress tensor is said to be equal to the
viscosity times the mean velocity gradients plus a negative term to balance the equation [7]:
U i U j
u iu j = t
+
x
xi
j

2
ij k
3

(2.19)

ij is the Kronecker delta and k the turbulent kinetic energy, which can be written as:
k=

1
u iu i
2

(2.20)

The turbulent or eddy viscosity t can be written as an expression of a constant C, a turbulent


velocity scale q ( q = 2k ) and a mixing length lm depending on the geometry of the flow:

t = C ql m
(2.21)

2.4.4 K-epsilon Realizable


The turbulence model used in this thesis was the k-epsilon realizable. The k-epsilon
turbulence models are robust and fairly precise for many non-boundary governed flows and
therefore preferable. To be able to describe the turbulent eddy viscosity mention in the
preceding chapter one must solve two transport equations: one for the turbulent dissipation
and one for the turbulent kinetic energy. The relation between the three is:

t = C ql m = C

k2

(2.22)

The two transport equations are:

(k ) + (ku j ) = + t
k
t
x j
x j

+ G k + Gb YM + S k
x j

( ) + (u j ) = + t
t
x j
x j

+ C1 S C 2
k + v
x j

+ C1

where

(2.23)

(2.24)

C 3 Gb + S


C1 = max 0.43,
,
+ 5

=S ,

S = 2 S ij S ij

In the equations Gk is a measure of the generation of the turbulence kinetic energy from the
mean velocity gradients, Gb the turbulence kinetic energy produced buoyancy and YM is a
measure of the so called dilation dissipation in compressible turbulence and is important for
high-Mach-number flows. C1 and C1 are constants and k along with are turbulent Prandtl
numbers for k and , respectively [8].
The realizable k-epsilon is proven to be more appropriate for homogeneous shear flows, flows
as jets and mixing layer, and even channel and boundary flows than the standard k-epsilon
model. The two major differences between the models are a new formulation of the viscosity
and the dissipation transport equation. When calculating the turbulent viscosity the value of
C is no longer constant but is a function of the kinetic energy, dissipation and the mean strain
and rate of rotation of the system [8].
The model constants are:

C1 = 1.44

C2 = 1.9

k = 1

= 1.2

2.4.5 Wall Functions


Wall functions are used to approximate variables as turbulence, velocity and pressure in the
region near a wall, called the boundary layer. To be able to resolve the fluid behaviour
without using some kind of assumption or simplification the mesh in this region must be very
fine resulting in some cases unnecessarily long computation times. Therefore it is of great use,
in cases where the boundary layer is of less importance, to use some kind of wall function
approximation. The wall functions rely on the existence of a logarithmic region in the velocity
profile, in the logarithmic layer the profile is

u+ =

vt
1
= ln( y + ) + B
u

(2.23)

where vt is the mean velocity parallel to the wall, u the shear velocity and is calculated by

u = w / ( w is the shear stress at the wall). Moreover, is the Karman constant which
is equal to 0.41, B an empirically derived constant related to the thickness of the viscous sub
layer and y + is the dimensionless distance from the wall [7].
y+ =

u y

(2.24)

The assumption for the turbulent kinetic energy originates from stating that the production of
turbulence and dissipation is nearly equal [7]. This results in:
2

u
k=
C

(2.25)

The expression for dissipation is derived as the value in the centre of the control volume and
is not applied in the control volume next to the wall.
3

u
=
y

(2.26)

10

2.5 Solar Load Model


The solar load model available in FLUENTs version 6.2.16 consists of a combination of a
solar ray tracing algorithm and a radiation model called surface to surface (S2S). The solar
ray tracing algorithm works as the source of the solar heat and the S2S radiation model
accounts for the internally scattered energy. The sources for this energy are the surfaces
exposed to the solar rays and thus soaked with solar heat energy. This method of modelling
solar load of course has its limitations. The S2S radiation model is based on computing a
view-factor file containing the data concerning the included irradiative surfaces relation to
each other. The computation of this file is very resource demanding and in this early stage
very sensitive to parallelization (this is the first version the solar load model is available in).
However, the combination of these models should be able to give a realistic representation of
a real case.

2.5.1 Solar Ray Tracing Algorithm


The Solar Ray Tracing uses both a positioning vector for the solar location and two
illumination parameters to set the solar intensity, to represent solar load. These properties can
be applied to any boundary conditions on the model, and the algorithm is coupled to FLUENT
by including a source term in the energy equations. To specify the magnitude and direction of
the sun the solar calculator can be used. The solar calculator needs specifications of the
coordinates of where the model is located, date and time of day to calculate the sun vector and
the magnitudes of the illumination parameters. The combination of these parameters gives a
solar ray that would appear at this certain location in the world at that given time.
The two illumination parameters used to describe the intensity of the solar load consist of two
irradiation terms, direct solar irradiation and diffuse solar irradiation. The direct solar
irradiation uses a two-band spectral model and accounts for material properties in IR- and
visible bands. The diffuse solar irradiation uses a single-band hemispherical averaged spectral
model only present in actual sun light.
Absorptivity and transmissivity are to be specified for all three bands, direct visible, direct IR
and diffuse hemispherical, when applying the solar load in the boundary condition.
Transmissivity is only apparent on those boundaries that are set as semi-opaque (such as
glass) and not for those set as opaque. These two parameters are mainly set to represent the
correct color of the boundary and to get the correct properties of the semi opaque materials.
Both absorptivity and transmissivity are defined for normal incident angle to the surfaces but
are recomputed/interpolated in FLUENT to represent the correct angle.
A scattering fraction can also be set in FLUENT Text User Interface (TUI) to include the
diffuse irradiation from scattered energy. The reflection from the incident rays will however
not be reflected locally but to all surfaces participating in the Solar Load Model. This
scattering of energy is mostly apparent when the air inside the domain is very humid.
A vector geometry based shading calculation is used in the solar ray tracing algorithm to
include the presence of shading. A method of reducing the ray tracing algorithm is also used
to insure the absence of unnecessarily long computation time. This is applied for mesh with
more than 104 faces and has a default value of 7. For mesh with more than five million faces
an increase of this value will reduce simulation time.

11

Glazing of materials can also be included when using solar ray tracing. For this to work the
transmissivity and reflectivity has to be specified in the wall boundary conditions. Depending
on the incident angle the glazing optical properties varies, significantly for larger angles than
40. Furthermore, it is possible to specify the ground reflectivity which is set to 0.2 as a default
value.

2.5.2 Surface to Surface Radiation Model (S2S)


The surface to surface radiation model in combination with the solar ray tracing algorithm is
used to completely model the solar load. The radiation model is used to include the internally
reflected energy from the sun.
The model accounts for the radiation exchange in an enclosure of grey diffuse surfaces. The
key parameters affecting the exchange are separation distance between the surfaces, seize of
the surfaces, and orientation of the surfaces. They are all accounted for in a so called viewfactor. The view-factor calculation is a very resource demanding process and should be
performed in parallel using more than one computation node. The surfaces included in
radiation should therefore also be kept to a minimum to keep computational time reasonable.
The S2S model uses the assumption that the surfaces are grey and diffuse, i.e. that the
surfaces are independent of the incident wavelength and, by Kirchoffs law, that the
emissivity and absorption are equal [8]. Another assumption in this radiation model is that the
medium separating the affected surfaces have no influence on the irradiative energy (E); the
energy incident on a surface is either reflected ( E ) or absorbed ( E ) and a fraction is
transmitted ( E ). Because most of the surfaces are treated as opaque the transmissivity can be
neglected and conservation of energy gives:

+ =1

(2.27)

since is equal to the emissivity ( ), and = 1 .


The energy equation used in the S2S model is composed of directly emitted and reflected
energy. The energy reflected from a surface is dependent on the energy incident on the
surface and can thus be expressed as a function of the energy flux leaving all radiating
surfaces. The energy flux from surface k is described as

q out ,k = k Tk + k qin ,k
4

(2.28)

where q out ,k represents the energy flux leaving the surface, qin ,k the energy flux incident on
the surface, k the reflected energy, k the emissivity, is the Boltzmanns constant and T
is the temperature.

12

3 Method Model Development


The methodology has in this thesis work been divided into two separate parts: one describing
the model development study and one presenting the model validation. A lot of time was
spent developing the numerical approach (see chapter 3.2.2.2) and understanding the usage of
the FLUENT software. The methodologies will in these two following chapters be presented
in a short but sufficient way.
Three types of cases have been simulated throughout this thesis. The cases included one
stationary simulation representing conditions after a long period of time (not evaluating
human comfort), one transient soaking case resembling a car being heated standing still
without a running engine, and one transient cooling situation with maximum flow from the
AC. Investigation of the boundary conditions was not to be a part of this initial study even
though a few problems made it to some extent impossible to avoid. A short description of the
three cases will later be given in this chapter.

3.1 Model Geometry


The compartment geometry used during the entire thesis work was for a Volvo S80. The
compartment is not a fully detailed model; parts like pedals, nozzles and details under the
seats are excluded from the model. The effects of most of these simplifications are negligible
and essential to reduce model simulation time. Furthermore, the doors, floors and roof consist
of two zones separated by a layer of air to increase insulation. The layers inside the floor and
the roof 10 mm thick where as the layer inside the doors is about 80 mm. The rest of the
geometry is only a shell of the interior of the passenger compartment, see Fig. 1.

Figure 1: The model geometry used in all simulations.

13

3.2 Numerical Setup


The mesh used in the development study consisted of approximately 1.8 million volume cells
and was created in Ansa and T-Grid. The surface mesh had a maximum skewness of 0.84 and
the volume mesh 0.95, and was strictly tetrahedral for both surface and volume mesh. The
mesh is the same as used in earlier studies [5] (except for refinement of the floor) as well as a
parallel heating project and was thus provided with at the start of the thesis.

Figure 2: The mesh in a plane passing through the middle of the driver seat.

The solver used in this project, as mentioned earlier, has been FLUENT version 6.2.16. The
specified solver settings were:

Segregated solver
Steady/non-steady state
K-epsilon turbulence model
o Enhanced wall treatment thermal effects
Solar Ray Tracing
Surface to Surface radiation model
PISO pressure velocity coupling
Boussinesq approximation for including buoyancy

The discretizations used for all simulations were for the solving of momentum a second order
upwind and for energy a first order scheme. For the transient calculations a first order implicit
scheme was used for the time discretization.
The boundary conditions set originated from earlier studies performed here at Volvo, see
Appendix 1. Three different thermal boundary conditions were used consistently throughout
this thesis:
- Convective
- Heat Flux
- Coupled
14

The convective boundary condition allows you to include the effects of the convective flow
outside the vehicle. This is done by calculating the convective heat transfer coefficient ()
dependent on the outside velocity. The theory used to do this is from flow over a plate and
details can be found in chapter 2.3. For the simulations in the development study different
values of was set to represent the different outside velocities. This was done for both the
steady state case and the transient cooling. For the steady state simulations was set to 100
representing an outside velocity of 110 km/h and for the cooling simulations 30 representing a
velocity of 40 km/h. These values are computed for a two dimensional plate with the length of
0.5 m (see chapter 2.3). The convective boundary conditions were also used when including
regional heat sources as from the engine compartment and exhaust system. The free stream
temperature, which is the temperature on the outside, was then set to 60 C and 70 C
respectively.
The heat-flux boundary condition allows you to set surfaces as constant energy sources by
specifying a heat flux in W/m2. This was tried in earlier cooling projects but showed to give
unreasonably high temperatures and has therefore not been used in this thesis [5]. This
boundary setting was only used when setting unimportant surfaces as adiabatic and thus
excluding them from the heat exchange calculations.
The coupled boundary condition is only used, or even available, when there is a fluid-facefluid relation. This happens, in the present geometry, in the seats, doors, floor, and the roof.
This boundary condition couples the two fluid regions and gives the opportunity to set the
separation surface with a thickness and material.
Shell conduction is an option that allows heat conduction in the plane of the boundary and is
in this early stage only used in the most sensitive areas. To include this physical property
FLUENT creates in the solver a prism layer which acts as if the boundary would have been a
solid material. This prism layer needs a thickness specified to be able to estimate its
conductive properties and this should therefore always be set in the boundary condition
settings.

15

Free strm temp


Wall thickness
BC
Type
Material
Shell cond

[C]
[m]
Windows
convective
glass
yes
30
45
0.005
Floor inside
coupled foamrubber
no
0
17
0.001*2
Floor outside convective
steel
yes
30
45
0.0015
Roof inside
coupled
canvas
no
0
17
0.0007*2
Seats
coupled
aluminum
no
0
17
0
Doors inside
coupled
canvas
no
0
17
0.0007*2
Doors outside convective
steel
yes
30
45
0.0007
Body
convective
steel
yes
30
45
0.0007
Instrmt panel convective foamrubber
yes
50
45
0.005
Parcel shelf
convective
steel
yes
50
45
0.0015
Wall engine
convective
steel
no
50
70
0.0015
Roof outside convective
steel
yes
30
45
0.0007
Tunnel consol convective
steel
no
50
60
0.0015
A-B-C pillars convective
steel
no
30
45
0.0007
Steering wheel heat-flux
aluminum
no
0
17
0
Other BC:s
heat-flux
aluminum
no
0
17
0
Table 1: Boundary condition of the most important surfaces set in the initial study for a cooling case with outside
air velocity 40 km/h and an outside temperature of 45 C.

The inside of the seats are set to solid materials to better represent the material properties of a
real seat. Previously performed studies have shown that the material inside the seat had large
influence on especially the cooling of the compartment [5]. The material set in all simulations
in this thesis was called seatfoam, see Table 2.
Material
Density [kg/m3]
Cp [J/kg K]
aluminium
2719
871
steel
8030
502.48
canvas
157
1873
glass
2700
840
foamrubber
80
1800
carpet
1600
1465
seatfoam
500
1075
Table 2: Material data used throughout the thesis.

Thermal Conductivity [W/m K]


202.4
16.27
0.4
0.78
0.3
0.3
50

The coupled boundary condition for the seat was set with a thickness of 0 to allow the
information from the inside of the seat to be directly coupled with the surrounding air. This
was done both because specifying the outer seat material is difficult and due to the fact that
the solid temperature inside the seat is a large source of heat and should not be enclosed by a
material that is hard to correctly specify.

3.3 Solar and Radiation Settings


The solar settings used in this thesis have had the intention of resembling the solar irradiation
conditions from wind tunnel tests. The solar position has therefore not been set using the solar
calculator function available in FLUENT but instead by using a constant direction vector. It
was initially set perpendicular to the plane parallel to the road, e.g. the sun is located straight
above the vehicle with a direction vector [0,0,1]. However, an angle of 15 was later set to
more precisely imitate the setting used during the wind tunnel test (for better understanding of
this angle see chapter 4.1 and Appendix 3).
16

The two key parameters controlling the solar representation are the solar illumination vector
and the magnitude of solar irradiation. It is available to set two different solar illumination
parameters in FLUENT: direct solar irradiation, which accounts for both IR- and visible-band
radiation, and diffuse solar irradiation, that accounts for a hemispherical averaged spectrum.
The diffuse irradiation is not present in the irradiation caused by the solar cells in the wind
tunnel and is therefore not used in these simulations. The value of direct solar irradiation for
all simulations is as in tunnel tests set to 1000 W/m2. The fraction of IR vs. visible radiation
was set to a default value of 0.5. Other parameters like ground reflectivity and solar scattering
were set to the default values of 0.2 and 1 respectively.
The surfaces included in the solar ray tracing were:
Outside roof
Outside doors
Inside doors
Body
Instrument panel
Tunnel consol
Seats
Parcel shelf
Windows
Steering wheel (only in validation simulations)
List 1: Surfaces included in the Solar Ray Tracing.

To minimize the extremely resource demanding view-factor calculation only the most
contributing surfaces was included in the Surface to Surface (S2S) radiation computation. The
S2S radiation model accounts for the internally distributed energy due to the solar exposure
(see theory chapter 2.6). The surfaces set as radiating were:
Instrument panel
Parcel shelf
Seats (only one of the coupled boundaries were included)
Inside roof (only one of the coupled boundaries were included)
Inside doors (only one of the coupled boundaries were included)
Windows
Steering wheel (only in validation simulations)
List 2: Surfaces included in the internally distributed energy using the S2S-radiation model

Due to the view-factor calculation being very time consuming a parameter study regarding the
clustering of the included surface elements was carried out. When calculating the surfaces
relation to each other clustering of the surface elements is essential. The parameter studied
was the number of faces per surface cluster and its effects on the result.

17

Three simulations were carried out clustering 100, 1000 and 10000 faces per surface cluster
using the same steady state case as described earlier. The results in all air measurement points
(breath, belt and foot) did not differ more than tenths of degrees for all three cases. Also, the
calculation time for all the cases lasted around 7.5 hours on 16 computational nodes.

Temperature [C]

45
40
100 faces

35

1000 faces
30

10000 faces

25
20
belt rear

belt front

floor rear

floor front

brth rear

brth front

Figure 4: Temperature plot of the air points for three different numbers of faces per surface cluster, 100, 1000,
and 10000.

The conclusion drawn from is that in this case the clustering has little importance to the
outcome. This can be a result of the very coarse surface mesh, making clustering not as
essential as for case with many more cells to cluster. It can also be due to the relatively small
irradiative energy source and few surfaces hit by the sun. Therefore, 1000 faces per surfaces
cluster was used in all simulations. This was partially due to its small effect on the view factor
computation time and also because of this is the same number of surfaces already being used.

3.4 Cases Simulated


Three types of simulations were performed during the model development study: a steady
state comfort, a transient soaking, and a transient cooling.
In this model developing study all the flow in the inlets are set with a direction normal to the
surface of the inlet. But, it was later known that the air flow should be directed towards the air
measurement points in the front and the back of the compartment. This was done in the
validation simulations. Furthermore, all simulations were set in recirculation (REC) mode.
This means only using the air inside the vehicle in the process of cooling the compartment.
The difference from a situation where out side air (OSA) is used is that all air exits in the front
at the passenger side instead of in the rear of the compartment (see Fig. 1).

3.4.1 Simulation Case: Steady state Comfort


The steady state comfort case was the starting point in this model study. To be able to fully
understand the works of this Solar Load Model one must start with the simplest of cases and
then work from there. The goal of this initial study was to make sure that the solar load was
included in the calculations in a correct way and to evaluate the model and the usage of it.
The same numerical setup, except for solar load, was used as for the previously performed
climate simulations here at Volvo, see chapter (3.1.2). The values used for the inlet velocities
and temperatures were extracted from a wind tunnel test performed on a Volvo S80 in the
18

year 2000. There were uncertainties with the coordinates of the measurement points which
may have had impact on the comparison with the test.

3.4.2 Simulation Case: Soaking


The soaking case represents a car parked in the sun without any air flow from the ventilation
nozzles. The flow is now strictly induced by free convection and buoyancy forces and is hard
to capture without sufficiently resolving the boundary layer. The largest differences from the
steady state case is modelling the inlets as walls instead of velocity-inlets and simulating
using a transient approach. Except for those changes the cases coincide.
Initially the soaking was, as mentioned above, simulated using the same numerical setup as
the preceding steady state comfort case. Due to problems with divergence some of the key
parameters controlling the buoyancy forces were altered and a few different approaches were
investigated. They consisted of:
a) Changing the thermal expansion factor controlling how much the changed density
affects the flow to a factor of 0.001 instead of 0.003 (0 = no buoyancy included)
b) Starting with gravitational force 0.0981 m/s and gradually increasing the value to 9.81
m/s to initially minimize the effects of gravitational force
c) Simulate without buoyancy
d) Simulating steady state but including buoyancy
The time step used in the transient simulations was a constant value of 1 s. To minimize
computational time it is in this case preferable to use an adaptive time step. Unfortunately, a
converging transient solution including buoyancy was not achieved and the use of an adaptive
time step not implemented.

3.4.3 Simulation Case: Cooling


The cooling simulation also originates from the same numerical settings as for the initial
steady state comfort simulation. The difference from the previous comfort simulations is the
use of transient inlet boundary conditions included using a User Defined Function (UDF). The
total cooling time simulated in this developing part of the thesis was set to 600 s because most
of the cooling occurs during the first few minutes; the smallest time scales and largest velocity
gradients are also apparent during the first couple of minutes and are therefore crucial to
resolve for the general prediction. A time step of 0.1 s was used for the first three seconds of
the simulation but was increased to 1 s for the remaining time.
The transient inlet boundary condition uses the registered temperatures from a cooling test
performed in a wind tunnel. It is then included in the calculations by using the UDF
mentioned above in specifying the temperature variation with time, see Appendix 4.
Used as initial condition for the cooling simulations were initially the results from one of the
soaking simulations. Due to the incapability to produce a correctly modelled soaking case,
including buoyancy, the final simulations instead used a patched temperature. The
temperature varied linearly from 55 C at the lowest point to 75 C at the highest point in the
compartment. The equation for this was created using temperature data from tunnel tests and
resulted in good representation of the real distribution; the average temperature was in the
linear distribution 59.8 C and in test 60 C. The downfall with using this method is the lack
of solar soaking in the instrument panel and parcel shelf. However, test results from the wind
19

tunnel test showed that excluding the radiation through the front wind screen did not affect the
cooling of the compartment much, hence making this approximation more than realistic.

Figure 5: The temperature distribution used as an initial temperature guess in the cooling simulations. The plane
is located in the middle of the compartment.

20

4 Method Model Validation


The model validation study consisted of two parts: a wind tunnel test performed on a Volvo
S80 and simulations compared with the results from this tunnel test. This chapter will first
describe the tunnel test and the results from it, followed by the method used for the
simulations.

4.1 Wind Tunnel Test


The car used in the present tunnel test was a dark blue Volvo S80 diesel. The interior of the
car was light in colouring; the compartment included almost white (creamy colour) leather
seats and a greyish colour on the instrument panel and parcel shelf.
When performing a wind tunnel test to compare with simulated data it is important to make a
few measurements before the actual test in the tunnel. This is done to both insure that the
vehicle is in a good condition, engine wise, and to get values of air leakage, total mass flow
into the compartment, and the flow distribution in the separate air vents. These data are
essential to obtain correct inlet data for the simulations.
The leakage measurement is done by sucking air from the compartment to obtain a desired
zero pressure in the compartment. This is done for different modes and fan voltages in a
stationary vehicle. In this test the mode was set to OSA with Vent settings (the inlets used
are the same as listed in Table 6). But, this should not give different results than for REC,
which is used in the simulations, and the measured mass flows were noted as:
Knob setting
Mass Flow [l/s]
1
28.4
2
44.4
3
58.9
4
71.5
5
86.2
6
101.1
7
112.7
Table 3: Total air flow for different knob settings.

A measurement of the mass flow in the individual ventilation nozzles was also performed
when the AC-system was set with Vent settings. It is not always the same distribution in the
ventilation channels and it must therefore be measured to be able to set the correct air flow in
the simulations. The measurement resulted in the following air distribution:
Inlet
Distribution [%]
Front Left
21.2
Front Right
19.6
Front Middle Left
21.7
Front Middle Right
18.7
B-pillar Left
9.6
B-pillar Right
9.2
Table 4: Air flow distribution in all inlets used in simulations.

21

There is a difference in distribution when comparing the left and the right side of the
compartment. This is consistent for the inlets in the vehicle and is very important to take into
account when setting up the simulations. To get a better feel of where the inlets are located
see Fig 1.
When excluding the flow in the b-pillars, which is done in one of the test cases (see chapter
4.1.2), the air will be redistributed to the front inlets and the new distribution will then
become:
Inlet
Distribution [%]
Front Left
26.0
Front Right
24.0
Front Middle Left
26.7
Front Middle Right
23.3
Table 5: Air flow distribution without flow in b-pillars.

4.1.1 Measurement Points


About 200 measurement points were placed in the S80 to measure air and surface
temperatures using air and surface thermal couples. The air point locations were set by Ford
standards in breath, floor, and belt level both in the front and in the back, see Fig. 6. Some of
the extra air points included were placed in:
Toe position on both driver and passenger side
Inside the trunk
Inside the driver seat
List 4: Air points included apart from standard air points.

Most of the thermal couples were set on surfaces throughout the passenger compartment. The
most interesting points regarding the thesis were placed on:
Outside roof
Instrument panel
Parcel shelf
Door panels (where the sun hits)
On seat cushion
On engine wall by feet (on the metal not on the carpet)
On the tunnel consol
On the canvas on the roofs inside
List 5: Position of surface thermal couples used in the tunnel test.

To be able to accurately compare the test data with simulation values the air temperature
points from the tunnel test, including breath, floor, toe and belt, were carefully measured in
the real compartment and implemented in the CFD model.

22

Figure 6: The air measurement points used to compare tests and simulated values. The points coordinates were
measured in the S80 vehicle used in the test and then placed in the Ansa model accordingly.

4.1.2 Test Cases


Three different cases were investigated during this tunnel test but only two were chosen to be
simulated in the validation study. The three tests will in the report be referred to as:
-

Test case 1
Test case 2
Test case 3

All the cooling tests had a constant outside temperature of 43C and an outside velocity of
100 km/h and the solar cells were set at an angle of 15 in the direction where the driver side
of the compartment gets more exposed. Also, the solar irradiation was set to a constant value
of 1000 W/m2. The two cases used in the simulations were:
I. Cooling with standard settings (test case 1)
Before the cooling test took place a soaking of the car was performed. When the mean
value of the air temperature inside the compartment had reached a value of 60C the
soaking stopped and the cooling test started. The mean value was calculated by adding all
the values of the air points in breath, belt, and foot level and dividing that by the number
of points. The cooling was performed with knob setting 7 (see Table 5) with a resulting
total mass flow of 112.7 l/s and the total time span of the cooling was approximately 30
min (referred to as cooling case 1). The cooling test was followed by constant air flow
settings for another 30 minutes resulting in comfort test case 1. After this phase, the
outside velocity was lowered to an idle value of 4-5 m/s but without turning the car off.
Instead the knob setting was lower to 4 and a total mass flow of 71.5 m/s and continued
like this for another 30 min. This last case is in the validating simulations referred to as
comfort case 2.
II. Cooling with out flow in b-pillar (test case 2)
This second cooling case was performed in the same manner as the preceding test with a
few exceptions. The largest difference was the shutting off the flow in the b-pillars
(referred to as cooling case 2). Another difference was, as a result from turning of the bpillar flow, a bit larger mass flow in the air inlets in the front. Also, an additional small

23

difference was a continued outside velocity of 100 km/h in the last comfort test. The two
comfort cases are later referred to as comfort case 3 and comfort case 4.

4.1.3 Conclusions from Wind Tunnel Test


Some important conclusions could be drawn from the results of the different test cases
investigated. The most significant observation was the very high temperatures on the outside
of the roof, on the instrument panel and on the parcel shelf. The temperature reached about
100 C after the soaking and stayed very warm during the entire cooling procedure. Some of
the remarkable observations from the tests, which were also implemented in the simulations,
were:
i.

The unimportance of the instrument panel temperature during the cooling. In test case
3 the windshield was covered with cardboard to exclude the instrument panel from the
solar load. The cooling was surprisingly consistent for the two cases with and without
covered windscreen despite a surface temperature difference of almost 30 C. One
conclusion to be drawn from this is that the surface temperature has less importance
when cooling the compartment. The thermal inertia as well as the starting temperature
of the system did not change for the two cases and this seems to be the governing
factors in the process of cooling the compartment. One factor that could have affected
the cooling in the case with covered windshield, could have been the fact that in the
first cooling test the instrument panel was soaked to a very high temperature and some
of that heat was still present. However, the surface temperature had dropped to about
60 C before the next soaking had begun making at least the surface temperature
correct.

ii.

The slow cooling of the temperature inside the front driver seat. There have been
discussions regarding using solid cells inside the seats when simulating cooling
because of the cooling of the seats supposedly is slower than the surrounding air. The
results from the tunnel tests showed that the temperature inside the seat did differ from
surrounding air temperature and solid cells should be used in a numerical simulation
(see Fig. 7).

70

Inside seat

Temperature [C]

65

Above seat

60
55
50
45
40
35
0

Time [min]

Figure 7: A graph showing the difference in cooling inside and just above the seat from Test Case 1.

24

10

4.2 Numerical Simulations


The surface mesh was for the validations simulations refined with purpose to reduce problems
with limited temperatures and to better resolve the solar sensitive areas. The new surface
mesh had a maximum skewness of 0.75 and the volume mesh had a total of 2.6 million cells
with a skewness of 0.95. An attempt to use a T-Grid generated Hex-core mesh was also done
but without success. The Hex-core mesh had to be run in a newer version of FLUENT that
only existed in a beta-version. This resulted in many problems related to bugs not yet taken
care of in the FLUENT code. Other than the mesh the same numerical settings were used
when performing these final computations. But, some of the boundary conditions were
slightly changed for these simulations, see Appendix 2.

4.2.1 Simulation Procedure


During the model developing study a numerical approach/process was slowly taking form.
When simulating solar load there are a few additional steps compared to simulations without
and the finally used approach was:
I. General settings, materials, solar load and boundary conditions
First of all, the general settings, material properties, solar load configuration and all
individually optimized boundary conditions must be set according to the case to be
simulated. A good way to minimizing errors and time is to use journal-files to set these
configuration based settings, see Appendix 4. In the journal, commands used in the
FLUENT Text User Interface (TUI) are typed in a text file and then read in the FLUENT
Graphic User Interface (GUI).
When all the settings are correctly specified a final case file will be saved to be used when
performing the next step. When performing the case-setup it is also important to specify
any monitoring of points or extraction of i.e. temperatures at specific intervals during the
simulation. Furthermore, it is important when using journal files to read the settings and
boundary condition settings to set all the models and specified settings first and last of all
read the boundary condition file.
II. Creating a View Factor file
The case created in the first step is here directly used and merely a command saying create
view factor file is the difference. The view-factor file is a pre calculation of how the
Surface to Surface (S2S) radiation model will treat the surfaces included in the radiation
calculation inside the vehicle. The view factor calculation is a very resource demanding
process and the number of surfaces included should be limited in any way possible; only
the most important boundaries should be included in the S2S model. The boundaries
included in the radiation calculation are specified when setting the boundary conditions in
the first step.
III. Creating solar data
After the view-factor file is created the solar data must be computed and saved. This is
done by running your simulation but without solving any of the governing equations. This
must in the present version of FLUENT be done serially hence the auto-save solar data
function is not available in parallel. If the solar load is constant in direction it is only
necessary to iterate a few iterations or time steps, which is the case in the present model,
but if it is required to capture the suns repositioning during a specific time period it
essential to iterate the same time span as in the simulation where the solar data will be
used.
25

IV. Initializing, Reading Solar Data, and Simulating


The last step before starting these simulations is to correctly initialize the solution, in this
case in terms of velocities and temperatures, or read data used as initial guess. The solar
data created in step 3 must also be read during the simulation. If the repositioning of the
sun with time is wanted the solar data must be read with the same time interval as when
creating the solar data. But, if the solar load is modeled as constant it is only necessary to
read the solar data at the start of the simulation. In some transient simulations it may be
necessary to start with a few time steps with the minimum temperature limit set to the
initialized temperature. This is needed due to that discrepancies in the volume mesh
results in limited temperatures in some zones and initial temperatures of 1 Kelvin is not to
prefer.

26

4.2.2 Simulation Case: Steady State Comfort


Four different steady state configurations were simulated to represent four of the comfort
tests performed in the wind tunnel (results in chapter 6.1). They consisted of two different
mass flows and the two configurations with and without flow from the b-pillars.

Comfort Loop 1:

Representing the conditions after 30 min of cooling and 30 min of


constant inlet temperature with 112 l/s of total mass flow (comfort
case 1)

Comfort Loop 2:

Representing the conditions in (1) but after additionally 30 min of


constant air flow but this time with 72.5 l/s of total mass flow
(comfort case 2)

Comfort Loop3:

Representing the conditions without flow in b-pillar nozzles after


30 min of cooling and 30 min of constant inlet temperature with
112 l/s of total mass flow (comfort case 3)

Comfort Loop 4:

Representing the conditions in (3) but after additionally 30 min of


constant air flow but this time with 72.5 l/s of total mass flow
(comfort case 4)

When simulating the steady state configurations they were performed in two steps. Firstly
8000 steady state iterations were done to develop a satisfactory velocity profile. This was
followed by simulating using unsteady state settings to achieve converged air temperatures
(for more details of the unsteady behaviour see chapter 3.3). The time step settings used for
the unsteady simulations were:
Time step
30 sec
Total time
100 time steps
Iterations/time step 30
Table 6: Time step settings for the transient part of the comfort simulations.

4.2.3 Simulation Case: Soaking


An attempt to simulate the soaking was also done in the validation part of the thesis. The
refined mesh used in all validation simulations was specifically refined in the region were
problems appeared during the previous soaking simulations, see chapter 3.5. However, due to
the same problems of divergence but in a new location the following cooling simulations used
the same linear temperature distribution as before as the initial guess.

27

4.2.4 Simulation Case: Cooling


Two cooling simulations were performed resembling the two cooling tests presented in the
preceding chapter 4.1.2 (results in chapter 6.3). The time step used is the same as for the
simulations done in the developing study: 0.1 s the first 3 seconds and 1 s the remaining 600
s. Also, to increase insulation in the model the layer of air inside the doors, floor and the roof
were set as the same solid material as in the seats (seatfoam). The two cooling simulations
will be referred to as:

Cooling Loop 1:

Representing the conditions of 10 min of cooling with 112 l/s of


total mass flow (cooling case 1)

Cooling Loop 2:

Representing the conditions without flow in b-pillar nozzles for 10


min of cooling with 112 l/s of total mass flow (cooling case 2)

To further investigate the importance of thermal inertia when simulating cooling, one more
alteration was tested for both cooling configurations. This configuration included changing
the thermal mass of the model. The density of foamrubber was changed from 80 kg/m3 to
1000 kg/m3, increasing the weight of the doors and floor. Also, the thickness of the instrument
panel was set to 0.1 m representing a total weight of 145 kg and the body was set to a
thickness of 0.01 m making its total weight 236 kg. The total weight of the passenger
compartment will with theses settings have a total mass of approximately 700 kg which is
reasonable. These two boundary settings had the intention of affecting the cooling mostly in
the area around the breath and belt air points in the front and all air points in floor level,
which are the most sensitive areas as observed earlier. These two configurations will be
referred to as:

Cooling Loop 3:

Representing the conditions of 10 min of cooling with 112 l/s of


total mass flow with extra thermal mass (cooling case 1)

Cooling Loop 4:

Representing the conditions without flow in b-pillar nozzles for 10


min of cooling with 112 l/s of total mass flow with extra thermal
mass (cooling case 2)

4.3 Verification of Simulations


To be able to rely on the numerical simulations a verification study should be performed. A
verification study intends to investigate the iterative convergence in a solution as well as the
mesh independence of the solution. In this thesis a mesh independent solution was not
achieved nor aimed to be. When implementing a new model, as the solar load model, it is
important to be able to simulate correct trends even if the deviation is large. A mesh
independent solution would have taken much too long and would have made the
implementation of the solar load harder.
A study of the iterative convergence in the steady-state comfort case was however
performed. To do this an air point in breath level in the front and in the back of the
compartment was monitored in temperature with iterations. The temperatures did not
converge to a specific value with iterations but instead had an oscillating behaviour with
temperature differences up to 10 C. This was consistent behaviour even after 15000
iterations. To investigate if a steady state solution is non-existing a non-steady state

28

simulation was performed with the result after 8000 iterations from the steady state case as
initial condition.

Figure 8: Plots of temperature with number of iterations. On the left hand side the simulation is steady state up to
8000 iteration and on the right hand side the simulation is a continued non-steady state simulation of the left
result.

The tendency of converging once using non-steady state settings is in Fig 8 very clear and
leads to the conclusion that one must simulate this steady state comfort case transient. One
approach could be to first achieve an almost fully developed velocity profile using steady
state setting but then switching to non-steady state to completely converge the solution. Or,
the simulation could be performed entirely with non-steady state settings as a continuing of a
cooling simulation.
The verification of the transient cooling simulations were mostly insuring that every time step
had converged to a criterion of 0.001 in all residuals except for energy where the criterion was
1E-6, see Fig 9.

Figure 9: The convergence of the transient cooling simulations with iterations.

29

5 Results Model Development


The results will be presented in the same fashion and order as the methodology in chapter 3;
first the results from the model development study will be displayed followed by the results
from the validation simulations. The results will be illustrated using both contour plots of the
temperature in the compartment as well as graphs of the temperature in the air measurement
points from the wind tunnel tests.
The results presented in the model development section are from the final configurations
simulated in this study. The settings and numerical approach are stated in the methodology
chapter.

5.1 Simulation Case: Steady state Comfort


The results from the steady-state comfort simulations showed clear differences in
temperature distribution for simulations with and without modelling the sun (Fig. 10). The
largest difference when comparing air point values was observed at breath level in the rear of
the compartment and is clearly an effect of generally lower velocities in that region as well as
the seat in the back acting as a very large energy source. The parcel shelf also generates of
heat in the rear due it being exceedingly exposed to the sun.
30

Temperature [C]

27
including solar load

24

not including solar


load

21
18
15
Torso
rear R

Torso
front R

Floor
rear R

Floor
front RL

Floor
front RR

Breath
rear R

Breath
front R

Figure 10: A plot of the temperatures in the air points for simulation with and without solar load.

The effect of solar radiation is very obvious when comparing the temperature distribution in a
plane in the middle of the compartment, see Fig 10. It is also confirmed when comparing
surfaces exceedingly exposed to the solar rays. The solar energy flux at the solar surfaces had
reasonable values with a maximum of about 800 W/m2 and the highest values are found on
the instrument panel and the parcel shelf, see Fig 11.

30

Figure 11: Temperature plots of a plane in the middle of the car (top left and right pictures) and of the most solar
affected surfaces in the compartment (bottom left and right pictures).

When changing the angle of the solar direction vector by 15 degrees there is an obvious
difference in solar exposure. The results of this can be seen mostly on the seats and a little on
the parcel shelf, see Fig 12.

Figure 12: Pictures of the solar heat flux of the most exposed parts of the compartment for both a vector straight
above the vehicle (left) and a vector with a 15 angle in positive y direction.

The surface temperatures were also studied and showed a general over prediction on the
surfaces where extra heat sources were added and a under prediction on surfaces like
windows and instrument panel (see Fig 13).

31

Temperature [C]

55
50
45
40
35
30
25
20

from test
from simulation

Cabin Ave

Foot rear R

Torso rear R

Brth rear R

Foot front RL

Foot front RR

Torso front R

Brth front R

Window rear

Window front

Windows sides

IP

Carpet roof

Carpet back

Carpet front

Wall warm

Body

Figure 13: Temperature distribution in air points and surface points after 8000 iterations from both test and
simulation.

5.2 Simulation Case: Soaking


To correctly simulate a soaking situation buoyancy and free convection must be correctly
modelled. As mentioned in the model development methodology the soaking case suffered
from steep divergence in continuity when including buoyancy.

Figure 14: The residuals from the soaking simulation including buoyancy. The value diverging is the continuity.

The continuity divergence was in this case due to very high values of the turbulent kinetic
energy and turbulent dissipation causing a mass imbalance in the left front door. The problem
originated from a poorly optimised mesh in the affected area.

32

Figure 15: Picture of where high values of turbulent kinetic energy causing the divergence problem in the
soaking case.

Results were obtained for the soaking case but only when excluding the buoyancy term from
the equations. The temperature distribution did in this case not in any way resemble the
temperature from test. The temperatures were very high in the area surrounding the seats
(modelled as solid cells) and of course by the instrument panel and the parcel shelf. But, the
tendency of an increasing temperature in roof level was not seen. The results from tests had a
difference of 10-15 C between foot level and head level where the temperature increased
gradually from foot to head.

Figure 16: Temperature distribution in a plane in the middle of the compartment for a soaking simulation without
including buoyancy.

Results were also acquired for the soaking case when running a steady state simulation with
included buoyancy. The solution did not as in the transient case diverge in continuity.
However, the results did not entirely show correct tendencies in temperature distribution, see
Fig 17.

33

Figure 17: Temperature distribution in a plane in the middle of the compartment for a soaking simulation using a
steady state approach and including buoyancy.

34

5.3 Simulation Case: Cooling

Temperature [C]

The results from the developing study had a fairly good correlation to results from an old test
performed on a S80 in 2000. The cooling is initially a bit too fast for all air points. The
temperatures are even a bit too high in both breath and belt level after a few minutes, see Fig
18. However, this is most likely due to a miss-direction of the airflow from the nozzles. In this
first developing study all nozzles are set with a directional vector normal to its surface and in
tests the air flow is to be directed toward the air measurement points.
60
55
50
45
40
35
30
25
20

Breath front ave Test


Breath front ave Sim
Breath rear ave Test
Breath rear ave Sim

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 18: Plot of temperature with time for breath level in the front and in the back with both simulated results
and values from tests.

The worst correlation is seen in the floor region and is most likely due to lack of thermal mass
in that area. The deviation is in the beginning of the cooling up to 10 C (see Fig. 19). Also,
there seems to be better correlation in the back of the passenger compartment consistently for
all results.

Temperature [C]

55

Foot front ave Test

50

Foot front ave Sim

45

Foot rear ave Test


Foot rear ave Sim

40
35
30
25

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 19: Plot of temperature with time for foot level in the front and in the back with both simulated results
and values from tests.

35

The solutions had good convergence except for the first few time steps where the velocity
gradients are large; the solution did in that early stage converge but needed more iterating per
time step. The velocity field had a very reasonable development with time even if it is
probable it would have needed a higher convergence criterion than 0.001 to have fully
converged each time step. The temperature development in the compartment also had an
acceptable behaviour, see Fig 20.

Velocity with time

Figure 20: Pictures of the velocity, from 0 to 1 m/s, in a plane in the middle of the compartment at 3 , 60, 300
and 600 seconds.

36

Temperature with time


b)

Figure 21: Pictures of the temperature from 15 to 30C in a plane in the middle of the compartment at 3, 60, 300
and 600 seconds.

The results from the simulations done with a solid material set in the fluid zones in doors,
floor panels and the roof showed that these areas in some way affect the cooling of the
compartment. The temperatures in general increased for the whole domain and the cooling
was not as aggressive as before. However, the temperatures were in this case in general too
high especially for areas in torso and breath level, see Fig 22.
60

Breath front ave Test

Temperature [C]

55

Breath front ave Sim

50

Breath rear ave Test

45

Breath rear ave Sim

40
35
30
25
20

630

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 22: Plot of temperature with time for breath level in the front and back with both simulated results and
values from tests.

37

The layer of air set as solid inside the door is quite large in volume and could be the reason
for the large change in cooling between configurations with and without solids. Due to this
the configuration used for the validation simulations did not include solid material inside the
door, only inside the floor and roof.

38

6 Results Model Validation


The results from the model validation will be presented beginning with all the steady state
comfort simulations followed by a short soaking summary and last the results from the four
cooling simulations. Model validation means in this case to compare the simulated results
with tunnel test than rather to completely validate the numerical model.

6.1 Simulation Case: Steady state Comfort


First the results from Comfort Loop 1 and 3 (with and without flow in b-pillars with
maximum flow) will be displayed and compared, followed by Comfort Loop 2 and 4.

6.1.1 Comfort Loop 1 and 3


In the first section the results from Comfort Loop1 (comfort case 1 with all vents and
maximum flow) will be presented on the left hand side and the results from Comfort Loop3
(comfort case 3 without flow in the b-pillars and maximum flow) on the right hand side.

Figure 23: Pictures comparing the two simulated steady state configurations loop1 (left side) and loop3 right
side)

When comparing the two configurations for a flow rate of 112 l/s and with and without bpillar flow, the result is not as expected. It seems like the configuration without b-pillars is
cooler than the configuration with. This could be an effect of the redistributed air to the front
inlets and not accounting for leakage and losses when doing this. Also, this is not a mesh
independent solution and the temperature is this case depends on the velocity profile. Notice,

39

also, that the results differ locally up to 10 C and might not be ready to indicate these kinds
of variations in configuration.

Temperature [C]

40
35
30

Test comfort case 1

25

Sim comfort loop 1

20
15
belt b

floor pas b

floor dr b

brth pas b

brth dr b

toe pas

toe dr

belt f

floor pas f

floor dr f

brth pas f

brth dr f

Figure 24: A plot of the temperatures at the air measurement points from both comfort test case 1 and the
Comfort Loop1-configuration. The points are located at breath, floor, and belt level in both the front and the back
as well as in toe level in the front.

The goal with this thesis was to be able to simulate the correct trends in temperature
distribution inside a passenger compartment. The results from Comfort Loop1approximately
follow the same trend as the test results, see Fig 24. The temperature is in the front of the
compartment slightly over predicted, except for at belt level, and even more in the back. The
reason for the high temperature in the back is most probable to be the choice of solid material
in the seats. The back seat is a very large energy source and should thus have large effects on
the results. The solid material should however have larger effects in a cooling situation but
might not have reached a steady state therefore having a bit too high temperature.
The much larger temperature at toe level on the driver side is most likely the result of the
added heat source in that area. The source is supposed to represent the heat from the engine
on to the front body incident to the foot area, on both passenger and driver side. This implies
that the extra heat source added in that same location might need some reconsideration.

40

Temperature [C]

35
30
Test comfort case 3

25

Sim comfort loop 3

20
15
belt b

floor pas b

floor dr b

brth pas b

brth dr b

toe pas

toe dr

belt f

floor pas f

floor dr f

brth pas f

brth dr f

Figure 25: A plot of the temperatures at the air measurement points from both comfort test case 3 and the
simulated Comfort Loop3 configuration. The points are located at breath, floor, and belt level in both the front
and the back as well as in toe level in the front.

The results from the simulated comfort loop 3 seem to have almost the same temperature
distribution as in the comfort loop 1. The reason for this is unclear. The air points in the front
are in both configurations directly hit by the cooling air and due to a coarse mesh might not be
able to show on the difference in inlet mass flow. The air point in the back might be too
influenced by the heat from the seats and roof top to be able to predict the change in with and
without flow in b-pillars. Moreover, the increase in air flow in the front inlets results in more
air flowing over the seats instead of staying in the front which could result in the cooler
temperatures in the rear.

41

6.1.2 Comfort Loop 2 and 4


The simulated results from Comfort Loop2 (comfort case 2 with all vents and medium flow)
and Comfort Loop4 (comfort case 4 without flow in b-pillar and medium flow) also had fairly
good resemblance with the tested values. The results from Comfort Loop2 and 4 are presented
in the same fashion as for 1 and 3, respectively.

Figure 26: Pictures comparing the two simulated steady state configurations comfort case 2 with flow in the bpillars(left side) and 4 without flow in b-pillars (right side)

Temperature [C]

40
35
Test comfort case 2

30

Sim comfort loop 2

25
20
belt b

floor pas b

floor dr b

brth pas b

brth dr b

toe pas

toe dr

belt f

floor pas f

floor dr f

brth pas f

brth dr f

42

Figure 27: A plot of the temperatures at the air measurement points from both comfort test case 2 and the
Comfort Loop2-configuration. The points are located at breath, floor, and belt level in both the front and the back
as well as in toe level in the front.

For the simulated Comfort Loop2 the temperatures still show the same trends as for the test
results, see Fig 27. The temperatures are now slightly under predicted, instead of over
predicted as in the previous configuration, especially in breath-level in the back of the
compartment. The temperature has however locally increased in temperatures when
comparing the two simulated configurations high and low mass flow i.e. Comfort Loop1 and
2.

Temperatures [C]

40
35
Test comfort case 4

30

Sim comfort loop 4

25
20
belt b

floor pas b

floor dr b

brth pas b

brth dr b

toe pas

toe dr

belt f

floor pas f

floor dr f

brth pas f

brth dr f

Figure 28: A plot of the temperatures at the air measurement points from both comfort test case 4 and the
Comfort Loop 4-configuration. The points are located at breath, floor, and belt level in both the front and the
back as well as in toe level in the front.

The same conclusion is drawn regarding the results from comfort loop 2 and 4 as for the
comparison of 1 and 3. The results are predicting the same temperature trends regarding the
temperature distribution but are unable to show the difference when altering configurations in
ventilation settings (with and without b-pillar flow). One difference is however observed
when comparing the latest two configurations. The temperature in the breath level in the back
has for comfort loop 4 been increased compared to comfort loop 2. This is a wanted behaviour
when there is no flow in the b-pillars for comfort loop 4.
The total computational time was for the steady state comfort simulations approximately 55
hours. The time was for the three simulation steps as follows:
1) View factor calculation = 6.5 hours using 8 computational nodes
2) Steady stat simulation for 8000 iterations = 36 hours using 16 computational nodes
3) Transient simulations for 100 time steps and 3000 iterations = 12.5 hours using 16
computational nodes

43

6.2 Simulation Case: Soaking


The soaking simulation did, even when using the refined mesh, still diverge in continuity.
During the model development study the mass imbalance causing the divergence was located
inside the left front door and has with this mesh been eradicated. However, the mass
imbalance was not fully eliminated and has now surfaced in a small region in the roof
fluid/solid zone. This indicates that the coupled areas (fluid-wall-fluid) are very sensitive to
mesh quality and should be refined with utter most care and precision to eliminate these
problems. Preferable is to mesh this separate zones with some kind of structured prism
approach to avoid this problem in the future.

Figure 29: Picture of where high values of turbulent kinetic energy causing the divergence problem in the
soaking case.

Due to lack of time the mesh was not once more refined to dissolve the present problem.
Instead a patched temperature consisting of a linear distribution calibrated using the
temperatures in foot and head level from the wind tunnel validation tests, was used. Having
the results form the tunnel test in mind, the heating of the major solar surfaces should not
have significant influence on the cooling process. Therefore this approach would give a more
than appropriate initial approximation, see Fig. 5.

44

6.3 Simulation Case: Cooling


The results from the cooling simulation representing cooling test case 1 and 2 will be
presented for two different configurations. The first two configurations representing cooling
test case 1 and cooling test case 2 with the same thermal inertia as in the previous steady state
simulations. The two other configurations has increased thermal mass but the same as the first
two in other settings, see chapter 4.2.4 for further details.

6.3.1 Cooling Loop 1


The results from Cooling Loop 1 had locally good correlation with test data, see Fig 29-31.
Breath and belt level in the rear of the compartment only showed on deviation of a maximum
of 5-8 C the first two minutes but revealed about the same behaviour as the test data the
following 7-8 minutes. The points in the floor regions all cooled too quickly compared to the
test data and especially in the front of the compartment (see Fig 31).

Temperature [C]

75

Breath front dr ave Test

65

Breath front dr ave Sim

55

Breath rear dr ave Test


Breath rear dr ave Sim

45
35
25
15
600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 29: Plot of temperature with time for breath level in the front and rear with simulated results, Cooling
Loop1, and values from tests case 1.

Temperature [C]

75

Belt front ave Test

65

Belt front ave Sim

55

Belt rear ave Test


Belt rear ave Sim

45
35
25
15

Figure 30: Plot of temperature with time for belt level in the front and rear with simulated results, Cooling
Loop1, and values from tests case 1.

45

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Temperature [C]

65
60
55
50
45
40
35
30
25

Foot pas front ave Test


Foot pas front ave Sim
Foot pas rear ave Test
Foot pas rear ave Sim

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 31: Plot of temperature with time for foot level in the front and rear with simulated results, Cooling
Loop1, and values from tests case 1. The results are for the passenger side only.

6.3.2 Cooling Loop 2

Temperature [C]

In general the same results were obtained for the Cooling Loop 2 configuration (without flow
in the b-pillar). The biggest different from the simulations with b-pillar flow is the increase of
temperature in the rear of the compartment. This of course is a wanted behaviour because of
the lack of air flow in the rear and shows that the difference in ventilation settings can be
shown in simulations.
85

Breath front dr ave Test

75

Breath front dr ave Sim

65

Breath rear dr ave Test


Breath rear dr ave Sim

55
45
35
25
15

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Temperature [C]

Figure 32: Plot of temperature with time for breath level in the front and rear with simulated results, Cooling
Loop2, and values from tests case 2. The results are for the driver side only.

85

Belt front ave Test

75

Belt front ave Sim

65

Belt rear ave Test

55

Belt rear ave Sim

45
35
25
15

Figure 33: Plot of temperature with time for belt level in the front and rear with simulated results, Cooling
Loop2, and values from tests case 2. The results are for the passenger side only.

46

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

The temperatures in belt level in the rear are during the first two minutes very accurately
following the test results. However, when the velocity field has been developed into the rear
of the compartment, the coarseness of the model probably causes a local increase in cooling.
This could be a reason for the temperature drop around the 3 minute mark. The tendency of
rapid cooling compared to test is even more evident in this configuration, especially in the
front torso/breath level and could be coupled to the higher velocity gradients in this
configuration (see both Fig 32 and Fig 33).

Temperature [C]

65

Foot front pas ave Test


Foot front pas ave Sim

60

Foot rear pas ave Test


Foot rear pas ave Sim

55
50
45
40
35
30

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 34: Plot of temperature with time for foot level in the front and rear with simulated results, Cooling
Loop2, and values from tests case 2. The results are for the passenger side only.

If comparing this cooling configuration with the previous, the cooling tendencies in the front
are faster for the configuration without flow in the b-pillar. This is probably caused by
increasing the air flow in the front ventilation nozzles when redistributing the air flow that
should have come in the b-pillars. The coarse mesh could also have influenced the outcome in
this later configuration, and probably in the others as well.

6.3.3 Cooling Loop 3


The results from the Cooling Loop 3 using more realistic thermal mass (see chapter 4.2.8),
displayed improvements compared to the previous loop (see Fig 35). The cooling in the front
has now better correlation to the speed of cooling for the empirical values. The area most
affected by this change in thermal inertial was the frontal area and the floor region just as
expected.

Temperature [C]

75

Breath front dr ave Test

65

Breath front dr ave Sim

55

Breath rear dr ave Test

45

Breath rear dr ave Sim

35
25
15

Figure 35: Plot of temperature with time for breath level in the front and rear with both simulated results,
Cooling Loop 3, and values from tests case 1.

47

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Temperature [C]

75

Belt front ave Test


Belt front ave Sim

65

Belt rear ave Test

55

Belt rear ave Sim

45
35
25
600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 36: Plot of temperature with time for belt level in the front and rear with both simulated results, Cooling
Loop 3, and values from tests case 1.

When comparing the simulated values in belt level with the test results it shows that the
simulated values have a much slower cooling behaviour. Changing the thickness of the
instrument panel and the car body obviously had impact in this simulated configuration.
65

Foot pas front ave Test


Foot pas front ave Sim

Temperature [C]

60

Foot pas rear ave Test


Foot pas rear ave Sim

55
50
45
40
35
30

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 37: Plot of temperature with time for foot level in the front and rear with both simulated results, Cooling
Loop 3, and values from tests case 1.

Even the temperatures in foot level seem to be closing the empirically obtained values. By
changing the thermal inertia of the model it should be possible to get even better correlation
with the test data. When comparing the foot values in the rear the increasing trend in the
beginning of the cooling is almost the same except for a global difference in temperature.

48

6.3.4 Cooling Loop 4


The results for Cooling Loop 4 did not show the same clear difference with thermal mass
settings. The rear breath level increased marginally but the rest of the breath levels kept the
same behaviour as before.

Temperature [C]

80

Breath front dr ave Test

70

Breath front dr ave Sim

60

Breath rear dr ave Test


Breath rear dr ave Sim

50
40

30
20
600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 38: Plot of temperature with time for breath level in the front and rear with both simulated results,
Cooling Loop 4, and values from tests case 2.

The belt level point in the front had small increase compared to the other configuration but
did not show as extreme differences as for the first cooling configuration. The rear also
showed a slight increase of temperature but as with the same significance as in the front, see
Fig 38-39.

Temperature [C]

70

Belt front ave Test

65
60

Belt front ave Sim


Belt rear ave Test

55
50

Belt rear ave Sim

45
40
35
30
25
600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 39: Plot of temperature with time for belt level in the front and rear with both simulated results, Cooling
Loop 4, and values from tests case 1.

The behaviour is consistent for all measurement points for Cooling Loop 4. This could have
the same origin as stated for the results in the steady state simulations: the redistribution of the
b-pillar air. Using a better numerical mesh would probably help to capture the higher velocity
gradients present in this configuration.

49

65

Foot pas front ave Test

Temperature [C]

60

Foot pas front ave Sim

55

Foot pas rear ave Test

50

Foot pas rear ave Sim

45
40
35
30
600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 40: Plot of temperature with time for foot level in the front and rear with both simulated results, Cooling
Loop 4, and values from tests case 1.

The total computational time was for the cooling simulations approximately 12 hours when
total simulated time is 10 min and approximately 32 hours when total time is 30 min. The
time was for the two simulation steps as follows:
1) View factor calculation = 6.5 hours using 8 computational nodes
2) Transient simulation with total time 10 min = 12 hours using 8 computational nodes
3) Transient simulation with total time 30 min = 32 hours using 8 computational nodes

50

6.4 Summary of the Results


The results from the development study showed that simulations with and without solar load
differed substantially for both the steady state 'Comfort' and the cooling case. The results from
the study showed for the cooling simulations good correlation with the test data when solar
load was included, see Fig. 41. The tendencies that did not correlate with the tests could be
attributed to the miss-direction of the inlet velocities, which was corrected for the 'validation'
simulations.
60

Breath rear ave Test

Temperature [C]

55

Breath rear ave Sim

50

Breath rear ave Sim no sun

45
40
35
30
25
20

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 41: Results from the model development study showing cooling temperatures for a wind tunnel test and
simulations with and without solar load.

The results from the steady state 'Comfort' simulations in the validation study displayed
correct temperature trends in the compartment but could not simulate the alteration in
configuration. When the configuration was changed to have no mass flow in the b-pillar inlets
the temperatures even decreased in the rear of the compartment and this is not a proper
behaviour. Moreover, the simulated results over predicted the temperatures in the rear of the
compartment and also in toe-level in the front. The high 'toe' temperatures could be a result of
the extra heat source added in that area originating from the engine compartment.
For the cooling simulations performed in the validation study the results showed good
correlation with test data in the rear of the compartment but not so good in the front. The
deviation in the front could be an attribute of both coarse mesh and insufficient thermal mass
in that area. The largest deviation was observed in torso level in the front, for Cooling Loop 2
and was about 15 C during the first minutes of cooling but not as much after 10 min. This
was probably due to the coarse mesh and the high velocities in the front. Other than this very
high temperature deviation the larges difference was observed in foot level for Cooling Loop
1, where it was about 10 C. The deviation was other than not larger than 5-6 C. The results
from the cooling simulations with higher thermal mass indicated that this problem could be to
some extent corrected with changing the thermal mass, see Fig. 4; the fast cooling in the front
was slowed down for the simulations with increased mass. The maximum deviation had now
decreased to a value of 7-8 C in foot level and was other than that within the set goal of 5 C.
These results were observed for Cooling Loop 3. The Cooling Loop 4 still showed extreme
deviation in front torso level but still with the same reason why.

51

Temperature [C]

70

Breath front dr ave Test

60

Breath front dr ave Loop 1

50

Breath front dr ave Loop 3 (extra therma mass)

40
30
20
600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 42: Results in breath level in the front of the compartment on driver side from tunnel test, Cooling Loop
1, and Cooling Loop 3 (with extra thermal mass)

65

Belt front dr ave Test

Temperature [C]

60

Belt front dr ave Loop 1

55

Belt front dr ave Loop 3 (extra therma mass)

50
45
40
35
30
25

600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 43: Results in belt level in the front of the compartment on driver side from tunnel test, Cooling Loop 1,
and Cooling Loop 3 (with extra thermal mass)

Temperature [C]

60

Foot pas front ave Test

55

Foot pas front ave Loop 1

50

Foot pas front ave Loop 3 (extra thermal mass)

45
40
35
30
600

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

Figure 44: Results in foot level in the front of the compartment on driver side from tunnel test, Cooling Loop 1,
and Cooling Loop 3 (with extra thermal mass)

52

Temperature [C]

The steady state 'Comfort' simulations were, as mentioned above, not able to show the same
tendencies as tests when altering ventilation configuration (with and without b-pillar flow).
However, the transient Cooling simulations did show a difference when excluding air flow in
the b-pillar inlets, see Fig 45. The cooling was slowed down in the rear of the compartment
which correlated with the tunnel tests.
75
70
65
60
55
50
45
40
35
30

Breath rear dr ave Loop 1


Breath rear dr ave Loop 2

53

600

Figure 45: Simulated results from Cooling Loop 1 and Cooling Loop 2 (without flow in b-pillars).

570

540

510

480

450

420

390

360

330

300

270

240

210

180

150

120

90

60

30

Time [s]

7 Discussion
There are many simplifications in this numerical model that can be discussed. For instance,
the many existing boundary conditions are very hard to in a correct way reconstruct to
coincide with a real S80 vehicle. The most problematic areas are the doors, the instrument
panel, and the floor. It is not easy to assign a material to a boundary that in real life contains
many different materials and parts with all unique heat transfer properties.
The materials and thicknesses chosen in the validation study were mostly selected for their
resemblance of accurate thermal mass, especially in the Cooling Loop 3 and 4. The material
on the inside of the roof was chosen as steel instead of canvas (which is the last layer in a real
car) just to increase the mass. Also the material on inside doors and floor was changed from
canvas and carpet, respectively, to the material foamrubber. This was also to better insulate
the compartment and increase the total mass of the individual parts to a more realistic value.
Finally, the layer of air in the floor and roof was changed to the solid material seatfoam, also
used inside the seat, to increase the insulation in these areas. It can be discussed if this is
really needed. But because the cooling in all simulations was too fast in the beginning of the
process, the conclusion that the mass is at least not too large at this point not unrealistic.
The choice of solid material in the seats (seatfoam) can also be discussed. The material
properties were originally taken from a former study performed here at Volvo for cooling
without included solar load. It was the material that, in that case, gave best correlation to test
data. The material was later changed in density to in the simulations better represent the total
weight of a real seat. A driver seat weighs approximately 30 kg and with the present material
properties so does the simulated seat. The value of Cp could be the parameter to tune when
studying this further.
The choice of (the convective heat transfer coefficient) and its effects can also be discussed.
Using flow over a plate theory is a very rough estimation and should be investigated further
especially when solar load is included in the simulations.
The solar load model also has its limitations. The high temperature on the roof is for instance
not possible to capture when using this model. The boundary was included in the solar ray
tracing but no solar heat flux was noted for any case. The roof reached a temperature of up to
100 C during the tunnel test and this should be included in future simulations.

54

7.1 Recommendations
The solar load model used throughout this thesis should be used when including solar load in
the future as well. However, it is in its early stages of development and many problems
involving parallelisation makes it sometimes difficult to use. Also, when setting up a case it I
recommend to set it up from the beginning; start with reading the mesh and then set all the
models and boundary conditions for present configuration, from the start.
Regarding the many materials and boundary conditions a closer study is needed to really
determine what is suitable for this S80 model. However, I do not recommend the usage of the
low density materials like canvas and carpet for floor and inside doors and roof. It is better to
set the metal or other material that is set underneath the carpet/canvas, the thermal properties
of the material feels more important to capture. Also, the material set in the seats should be
investigated but solid cells should be used and the material set in the validation simulations is
a good starting point.
One of the questions that arose during the analysis was: Is it necessary to include the heat
sources from the engine and the exhaust system when the toe points on the driver side shows
much too high temperatures compared to test data? This is something that should be
investigated more in future work.
The boundary conditions where most of the heat transfer occurs should if possible be set using
shell conduction. This enables the calculation of heat transfer in the plane, better capturing the
heat transfer phenomena. Furthermore, as few surfaces as possible should be included in the
Surface to Surface radiation model. The view factor calculation is, as mentioned throughout
this report, very resource demanding and it highly dependent on the amount of surface
elements included in the radiation calculations.
Finally, it is recommended to, as in this thesis, use journal files to set up the case. This
shortens the time to set up a case from the beginning if or when it is necessary due to the
parallelisation problem mentioned above. But, this can invite new problems when changing
versions of FLUENT and it is therefore important to be thorough when then changing the
journal files.

55

8 Conclusions
The conclusions that can be drawn from this thesis study are:
1) When simulating the steady state comfort case it is not sufficient to simulate
using steady state approach. To obtain a converged result a transient
calculation must be done.
2) The soaking simulation is very hard to obtain good results from and is much
more sensitive in terms of convergence than the other cases. It is also noticed
that the soaking part of the cooling simulation is not as important as initially
thought. By setting a linearly varying temperature distribution in the
compartment a good initial guess for the cooling simulation can be achieved.
3) The solid cells inside the seats have influence on the cooling process and
should be included in the simulations. However, it might be good to perform a
study of choice of material for theses solid cells.
4) It was possible to simulate cooling within a general deviation of 5 C, except
for in foot level.
5) One conclusion is that it was not possible to include external boundaries in the
solar load. The reason for this is at this point unclear.
6) The alteration in configuration was detectable in the cooling simulations but
not in the steady state comfort.
7) It is possible to simulate realistic temperature trends in the passenger
compartment in both steady state conditions as well as in a cooling case.
8) The number of faces per surface cluster was not a sensitive parameter in the
surface to surface model.
9) A final conclusion is that the cooling of the compartment highly depends on
the systems thermal inertia. It is possible to adjust this by setting more material
on the boundaries or increasing Cp or the densities of the already set materials.

56

8.1 Future Work


There is a lot of future work needed to make this numerical model sufficiently good to use in
real Volvo projects. There are areas that have been left out or at least not studied enough in
this work and should be done when working with it in the future. The method to include the
solar load in simulating compartment temperatures is developed and can with ease be used in
these future investigations.
Firstly, it is essential that an excessive boundary study is performed to determine the
importance of the different settings used in these simulations; there are parameters as
absorption and transmission when including the sun that would be very interesting to perform
a study of. It would also be preferable to investigate the different materials used in the model.
Is a correct material used for the solid cells inside the seat or should it be changed?
The solar load model should also be investigated further. There was not sufficient time to
perform an extensive study in changing the settings of the model to see its effects and learn
better its limitations and parameter sensitivity.
When a good model has been achieved in terms of boundary condition and realistic thermal
properties have been set, a better mesh should be generated. When simulating heat transfer it
is important to sufficiently resolve the areas where the heat transfer is large, e.g. at the
windows, instrument panel and roof top. These areas could be resolve using prism layers. It
would also be interesting to refine the area in front of the inlets to better resolve the velocity
field. This could be a reason for the locally bad results in the front, where most of the air is
entering the compartment.

57

9 References
[1]

Lin C-H., Lelli M., Niemiec R., Han T., Hammond D., An Experimental and
Computational Study of Cooling in a Simplified GM-10 Passenger
Compartment, SAE Technical Paper No. 910216, The Engineering Society for
Advanced Mobility Land Sea Air and Space, Warrendale, Pa., USA

[2]

Svensson C., Modelling the Thermal Comfort in a Car Compartment by Using


CFD, Diploma Thesis, Department of Thermo and Fluid Dynamics, Chalmers
University of Technology, Sweden, 2002

[3]

Huang L., Han T., A Case Study of Occupant Thermal Comfort in a Cabin
Using Virtual Thermal Comfort Engineering, EACC 2nd European Automotive
CFD Conference, Frankfurt, Germany, June 2005

[4]

Mezrhab A., Bouzidi M., Computation of Thermal Comfort Inside a Passenger


Car Compartment, Science Direct, Applied Thermal Engineering 26, 16971704, 2006

[5]

Chroner Z., Computation of Air Temperatures in a Passenger Compartment


Using CFD Steady-state and Transient, Volvo Technical Paper No. 700187

[6]

Kundu P.K., Fluid Mechanics, 2004

[7]

Ferziger J. H., Pric M., Computational Methods for Fluid Dynamics: 3rd
Edition, 2002

[8]

FLUENT, User Guide version 6.2.16

[9]

Svensson C., Including Solar Radiation When Modelling the Thermal Comfort
of a Car Compartment, Volvo Technical Paper 700377

58

Appendix 1: Boundary Conditions from Old Simulations


The boundary conditions used initially in the thesis were used in old simulations regarding
heating/cooling of a passenger compartment. The most essential boundaries were set as:
Free strm temp
Wall thickness
BC
Type
Material
Shell cond

[K]
[m]
Windows
convective
glass
yes
100
318
0,005
Floor
convective
steel
yes
100
318
0,0015
Roof inside
coupled
canvas
yes
0
300
0,0007*2
Roof outside
convective
steel
yes
100
318
0,0007
Seats
coupled
aluminum
no
0
300
0
Doors inside
coupled
canvas
yes
0
300
0,0007*2
Doors outside
convective
steel
yes
100
318
0,0007
Body
convective
steel
no
100
318
0,0007
Instrument panel convective foamrubber
yes
50
318
0,005
Rooftop
heat-flux
aluminum
no
0
300
0
Warm engine
convective
steel
yes
50
343
0,0015
Tunnel consol
convective
steel
yes
50
333
0,0015
A-B-C pillars
convective
steel
no
100
318
0,0007
Steering wheel
heat-flux
aluminum
no
0
300
0
Other
heat-flux
aluminum
no
0
300
0
Table 7: The boundary condition settings used in old simulations and also used as first settings in the thesis.

Appendix 2: Boundary Conditions in Validation Simulations


The boundary conditions used in the validation simulations differed from those initially set,
see Table 8.
Free strm temp
Wall thickness
BC
Type
Material
Shell cond

[K]
[m]
Windows
convective
glass
yes
95
316
0,005
Floor inside
coupled foamrubber
yes
0
300
0,01*2
Floor outside
convective
steel
yes
95
316
0,0015
Roof inside
coupled
steel
yes
0
300
0,0007*2
Roof outside
convective
steel
yes
95
316
0,0007
Seats
coupled
aluminum
no
0
300
0
Doors inside
coupled foamrubber
yes
0
300
0.001*1
Doors outside
convective
steel
yes
95
316
0,0007
Body
convective
steel
yes
95
316
0.001
Instrument panel convective foamrubber
yes
50
318
0,005
Parcel shelf
convective
steel
yes
50
323
0,0015
Warm engine
convective
steel
no
50
343
0,0015
Tunnel consol
convective
steel
no
50
333
0,0015
Pillars
convective
steel
no
95
316
0,0007
Steering wheel
heat-flux
aluminum
no
0
300
0
Other
heat-flux
aluminum
no
0
300
0
Table 7: The boundary condition settings used in old simulations and also used as first settings in the thesis.

59

Appendix 3: FLUENT Model Settings


The settings used in FLUENT for the steady state comfort simulations are summarized in the
table below:

Solver Settings
Solver
Formulation
Time
Velocity Formulation
Gradient Option
Porous Formulation

Segregated
Implicit
Steady
Absolute
Cell-based
Superficial Velocity

Operating Conditions
Operating Pressure [kPa]
Reference Pressure Location
Gravity
Gravitational Acceleration [m/s2]

Boussinesq Parameter (Operating Temp) [K]


Variable-Density Parameters
Energy Model
Energy Equation
Viscous Model
Model
K-epsilon Model
Near-Wall Treatment
Enhanced Wall Treatment Options
Options
Model Constants
User-Defined Functions

60

101.324
(0,0,0)
Enabled
x=0
y=0
z = -9.81
288.16
Enabled
Enabled

K-epsilon (2 eqn)
Realizable
Enhanced Wall Treatment
Thermal Effects
Full Buoyancy Effects
Default
Default

Radiation Model
Model
Surface to Surface (S2S)
Partial Enclosure Temperature [K]
318
Flow Iter per Radiation Iter
10
Number of S2S Sweeps
1
Tolerance
0.001
View Factor Parameters
Surfaces
Blocking
Method
Hemi cube
Smoothing
None
Hemi cube Parameters
Resolution
10
Subdivisions
10
Normalized Separation Distance
5
Cluster Parameters
Faces per Surface Cluster (BC:s included)
1000
Faces per Surface Cluster (BC:s not included)
0
Solar Load Model
Model
Solar Ray Tracing
Sun Direction Vector
(0 , 0.25882 , 0.96593)
Illumination Parameters
Direct Solar Irradiation [W/m2]
1000 (Constant)
Diffuse Solar Irradiation [W/m2]
0 (Constant)
Spectral Fraction [V/V+IR]
0.5
Solution Controls
Equations
Flow
Turbulence
Energy
Under-Relaxation Factors
Default
Pressure-Velocity Coupling
Model
PISO
Skewness Correction
1
Neighbor Corrections
1
Skewness-Neighbor Coupling
Enabled
Descretization
Pressure
Standard
Momentum
Second Order Upwind
Turbulence
First Order Upwind
Energy
First Order Upwind
Initialization
Temperature in all Zones [K]
300

61

Boundary Conditions
Inlet Boundary Condition Front Left
Type
Velocity Inlet
Velocity Specification Method
Magnitude and Direction
Reference Frame
Absolute
Velocity Magnitude Maximum Flow [m/s]
2.59
Velocity Magnitude Medium Flow [m/s]
1.64
Coordinate System
Cartesian (X, Y, Z)
X-Component of Flow Direction
0.8644 (Constant)
Y-Component of Flow Direction
0.2994 (Constant)
Z-Component of Flow Direction
0.4039 (Constant)
Temperature Maximum Flow [K]
279.9
Temperature Medium Flow [K]
281.43
Turbulence Specification Method
K and Epsilon
Turbulent Kinetic Energy [m2/s2]
1
Turbulent Dissipation Rate [m2/s3]
1
Internal Emissivity
1
Participates in Solar Tracing
Disabled
Inlet Boundary Condition Front Right
Type
Velocity Inlet
Velocity Specification Method
Magnitude and Direction
Reference Frame
Absolute
Velocity Magnitude Maximum Flow [m/s]
2.38
Velocity Magnitude Medium Flow [m/s]
1.51
Coordinate System
Cartesian (X, Y, Z)
X-Component of Flow Direction
0.8667 (Constant)
Y-Component of Flow Direction
-0.2973 (Constant)
Z-Component of Flow Direction
0.4005 (Constant)
Temperature Maximum Flow [K]
280.45
Temperature Medium Flow [K]
281.8
Turbulence Specification Method
K and Epsilon
Turbulent Kinetic Energy [m2/s2]
1
Turbulent Dissipation Rate [m2/s3]
1
Internal Emissivity
1
Participates in Solar Tracing
Disabled

62

Inlet Boundary Condition Front Middle Left


Type
Velocity Inlet
Velocity Specification Method
Magnitude and Direction
Reference Frame
Absolute
Velocity Magnitude Maximum Flow [m/s]
3.7
Velocity Magnitude Medium Flow [m/s]
2.34
Coordinate System
Cartesian (X, Y, Z)
X-Component of Flow Direction
0.847 (Constant)
Y-Component of Flow Direction
-0.352 (Constant)
Z-Component of Flow Direction
0.3979 (Constant)
Temperature Maximum Flow [K]
278.05
Temperature Medium Flow [K]
279.16
Turbulence Specification Method
K and Epsilon
Turbulent Kinetic Energy [m2/s2]
1
Turbulent Dissipation Rate [m2/s3]
1
Internal Emissivity
1
Participates in Solar Tracing
Disabled
Inlet Boundary Condition Middle Right
Type
Velocity Inlet
Velocity Specification Method
Magnitude and Direction
Reference Frame
Absolute
Velocity Magnitude Maximum Flow [m/s]
3.18
Velocity Magnitude Medium Flow [m/s]
2.02
Coordinate System
Cartesian (X, Y, Z)
X-Component of Flow Direction
08624 (Constant)
Y-Component of Flow Direction
0.3014 (Constant)
Z-Component of Flow Direction
0.4067 (Constant)
Temperature Maximum Flow [K]
278.25
Temperature Medium Flow [K]
279.11
Turbulence Specification Method
K and Epsilon
Turbulent Kinetic Energy [m2/s2]
1
Turbulent Dissipation Rate [m2/s3]
1
Internal Emissivity
1
Participates in Solar Tracing
Disabled

63

Inlet Boundary Condition Front Middle Left


Type
Velocity Inlet
Velocity Specification Method
Magnitude and Direction
Reference Frame
Absolute
Velocity Magnitude Maximum Flow [m/s]
3.7
Velocity Magnitude Medium Flow [m/s]
2.34
Coordinate System
Cartesian (X, Y, Z)
X-Component of Flow Direction
0.847 (Constant)
Y-Component of Flow Direction
-0.352 (Constant)
Z-Component of Flow Direction
0.3979 (Constant)
Temperature Maximum Flow [K]
278.05
Temperature Medium Flow [K]
279.16
Turbulence Specification Method
K and Epsilon
Turbulent Kinetic Energy [m2/s2]
1
Turbulent Dissipation Rate [m2/s3]
1
Internal Emissivity
1
Participates in Solar Tracing
Disabled
Inlet Boundary Condition Middle Right
Type
Velocity Inlet
Velocity Specification Method
Magnitude and Direction
Reference Frame
Absolute
Velocity Magnitude Maximum Flow [m/s]
3.18
Velocity Magnitude Medium Flow [m/s]
2.02
Coordinate System
Cartesian (X, Y, Z)
X-Component of Flow Direction
08624 (Constant)
Y-Component of Flow Direction
0.3014 (Constant)
Z-Component of Flow Direction
0.4067 (Constant)
Temperature Maximum Flow [K]
278.25
Temperature Medium Flow [K]
279.11
Turbulence Specification Method
K and Epsilon
Turbulent Kinetic Energy [m2/s2]
1
Turbulent Dissipation Rate [m2/s3]
1
Internal Emissivity
1
Participates in Solar Tracing
Disabled

64

Outlet Boundary Condition REC-mode


Outlet
Pressure Outlet
Gauge Pressure
0 (Constant)
Radial Equilibrium Pressure Distribution
Disabled
Backflow Total Temperature [K]
300 (Constant)
Turbulence Specification Method
K and Epsilon
Backflow Turbulent Kinetic Energy [m2/s2]
1
Backflow Turbulent Dissipation Rate [m2/s3]
1
External Black Body Temp Method
Boundary Temperature
Internal Emissivity
1
Target Mass-flow-rate
Disabled
Participates in Solar Tracing
Disabled
Wall Boundary Conditions
Type
Wall
Thermal
Convective
Heat-Flux
Coupled
Momentum
Wall Motion
Stationary Wall
Shear Condition
No Slip
Motion
Relative to Adjacent Cell Zone Enabled
Table ##: Settings used in FLUENT for all steady state Comfort simulations in the thesis work.

The settings for the Cooling simulations were the same except for solver settings and inlet
boundary condition settings. The difference was a non-steady setting in Time Solver Setting
and for all Inlet Boundary Conditions Temperature was set to the specific UDF variable
instead of a constant value.

65

Appendix 4: Journal Files


Five different journal files were used to set up the simulations.
1) General Settings
The first file is where all the general settings were set, e.g. activating the solving of all
equations, specifying turbulence model and so forth.
;------------------------------------------------------------------;
GENERAL SETTINGS FOR COOLING SIMULATION LOOP1
;
;------------------------------------------------------------------;
;----------------------------------; DEFINING UNSTEADY STATE SOLUTION ;
;----------------------------------;
/define/models/unsteady-1st-order
y
;
;--------------------------------; TURNING ON THE ENERGY EQUATION ;
;--------------------------------;
/define/models/energy
y
y
y
y
y
;
;------------------; ENABLING GRAVITY ;
;------------------;
/define/operating-conditions/gravity
Y
0
0
-9.81
;
;----------------------------; SETTINGS FOR VISCOUS MODEL ;
;----------------------------;
; TURBULENCE MODEL USED IS THE K-E REALIZABLE
define/models/viscous/ke-realizable
y
; INCLUDING THE BUOYANCY EFFECTS IN THE K-E MODEL
/define/models/viscous/buoyancy-effects
y
; INCLUDING ENHANCED WALL TREATMENT
/define/models/viscous/near-wall-treatment/enhanced-wall-treatment
y
; SETTING THE THERMAL EFFECTS IN THE K-E MODEL
/define/models/viscous/near-wall-treatment/wf-thermal-effects
y
;

66

;------------------; SOLVING SETTINGS ;


;------------------;
; ENABLING THE SOLVING OF ALL EQUATIONS
/solve/set/equations/flow
y
/solve/set/equations/ke
y
/solve/set/equations/temperature
y
; MOMENTUM DISCRETIZATION IS SET TO A FIRST ORDER SCHEME
solve/set/discretization-scheme/mom
1
; UNDER-RELAXATION FACTORS
/solve/set/under-relaxation/pressure
0.5
/solve/set/under-relaxation/mom
0.2
; DISABLING SECONDARY GRADIENTS
(rpsetvar 'temperature/secondary-gradient? #f)
(rpsetvar 'temperature/shell-secondary-gradient? #f)
; DISABLE THE CONVERGENCE CHECK
/solve/monitors/residual/check-convergence
y
y
y
y
y
y
y
; MAKING SURE ALL IS MONITORED
/solve/monitors/residual/monitor
y
y
y
y
y
y
y

67

2) Materials
The second journal file is where all the materials and their individual properties are set:
;-------------------------------------; SETTING THE MATERIALS USED IN LOOP1 ;
;
(INCLUDING THERMAL EXPANSION)
;
;-------------------------------------;
/define/materials/change-create
air
air
y
boussinesq
1.225
n
n
n
n
n
n
y
0.003
n
n
/define/materials/change-create
aluminum
steel
y
8030
y
constant
502.48
y
constant
16.27
n
n
/define/materials/change-create
aluminum
canvas
y
157
y
constant
1873
y
constant
0.4
n
n
/define/materials/change-create
aluminum
plywood
y
590
y
constant
2500
y

68

constant
0.109
n
n
/define/materials/change-create
aluminum
glass
y
2700
y
constant
840
y
constant
0.78
n
n
/define/materials/change-create
aluminum
foamrubber
y
80
y
constant
1800
y
constant
0.3
n
n
/define/materials/change-create
aluminum
carpet
y
1600
y
constant
1465
y
constant
0.3
n
n
; SOLID CELL MATERIAL IN THE SEATS FROM ZENITHAS LOOP3_1 MATERIAL
/define/materials/change-create
aluminum
seatfoam
y
500
y
constant
1075
y
constant
50
n

69

3) Changing Fluid Zones to Solid Zones


The third journal file is used to set the fluid zones in the seats to solid zones:
;-------------------------------------------------------------; CHANGING THE FLUID CELLS INSIDE THE SEATS INTO SOLIDS CELLS ;
;-------------------------------------------------------------;
/define/boundary-conditions/zone-type
fluid-seat-back
solid
/define/boundary-conditions/zone-name
fluid-seat-back
solid-seat-back
/define/boundary-conditions/zone-type
fluid-seat-front-dr
solid
/define/boundary-conditions/zone-name
fluid-seat-front-dr
solid-seat-front-dr
/define/boundary-conditions/zone-type
fluid-seat-front-pas
solid
/define/boundary-conditions/zone-name
fluid-seat-front-pas
solid-seat-front-pas

4) Setting up The Solar Load Model


The fourth journal file was used to set all parameters used in the solar load model:
;----------------------------------------------------------------;
SOLAR LOAD AND S2S SETTINGS FOR LOOP2_3
;
;
(SOLAR VECTOR 15 degrees)
;
;----------------------------------------------------------------;
; ENABLING THE SOLAR LOAD MODEL AND DEFINING THE SUNS POSITION
/define/models/radiation/solar
y
/define/models/radiation/solar-parameters/sun-direction-vector
0
-0.25882
0.96593
/define/models/radiation/solar-parameters/sol-camera-pos
; SETTING ILLUMINATION PARAMETERS
/define/models/radiation/solar-parameters/illumination-parameters
constant
1000
constant
0
0.5
; ENABLING THE S2S MODEL
/define/models/radiation/s2s
y
; SETTING THE PARAMETERS FOR THE SURFACE TO SURFACE MODEL
/define/models/radiation/s2s-parameters/set-vf-parameters
1000
0
0
1
5
10

70

10
; SETTING THE TEMPERATURE FOR PARTIAL ENCLOSURES (as the seats)
/define/models/radiation/s2s-parameters/partial-enclosure-temperature
318

5) Setting all Boundary Conditions


The fifth and last journal file is used to set all the boundary conditions:
;-----------------------------------------------------------------------;
Boundary Conditions for the Steady State Comfort Simulation
;
;
Vent REC VALIDATION LOOP1
;
;-----------------------------------------------------------------------;
;--------------------------------------; SETTING THE SOLID CELLS IN THE SEATS ;
;--------------------------------------/define/boundary-conditions/solid
solid-seat-back
y
seatfoam
n
n
y
0
0
0
0
0
1
n
/define/boundary-conditions/solid
solid-seat-front-dr
y
seatfoam
n
n
y
0
0
0
0
0
1
n
/define/boundary-conditions/solid
solid-seat-front-pas
y
seatfoam
n
n
y
0
0
0
0
0
1
n

71

;-------------------------------; SETTING THE OULET USED IN REC ;


;-------------------------------;
/define/boundary-conditions/pressure-outlet/outlet_evakfan
n
0
n
n
300
n
y
y
n
1
n
1
y
n
1
n
n
;-----------------------------------; SETTING THE INLETS WITH MASS FLOW ;
;-----------------------------------;
/define/boundary-conditions/velocity-inlet/inlet_left_nossle
y
y
n
2.59
y
n
0.8644
n
0.2994
n
0.4039
n
279.9
y
n
1
n
1
y
n
1
n
define/boundary-conditions/velocity-inlet/inlet_right_nossle
y
y
n
2.38
y
n
0.8667
n
-0.2973
n
0.4005

72

n
280.45
y
n
1
n
1
y
n
1
n
define/boundary-conditions/velocity-inlet/inlet_mitt_left
y
y
n
3.70
y
n
0.8470
n
-0.3520
n
0.3979
n
278.05
y
n
1
n
1
y
n
1
n
define/boundary-conditions/velocity-inlet/inlet_mitt_right
y
y
n
3.18
y
n
0.8624
n
0.3014
n
0.4067
n
278.25
y
n
1
n
1
y
n
1
n
define/boundary-conditions/velocity-inlet/inlet_b_left
y
y
n

73

2.82
y
n
0.6924
n
0.6298
n
0.3518
n
284.8
y
n
1
n
1
y
n
1
n
define/boundary-conditions/velocity-inlet/inlet_b_right
y
y
n
3.67
y
n
0.6925
n
-0.6307
n
0.3504
n
285.33
y
n
1
n
1
y
n
1
n
;--------------------------------------------------------------------; SETTING THE INLETS AND OUTLETS NOT USED AND THUS MODELLED AS WALLS ;
;--------------------------------------------------------------------;
/define/boundary-conditions/modify-zone/zone-type/outlet_rear
wall
define/boundary-conditions/wall/outlet_rear
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n

74

n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/modify-zone/zonetype/inlet_side_defroster_right
wall
define/boundary-conditions/wall/inlet_side_defroster_right
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/modify-zone/zone-type/inlet_side_defroster_left
wall
define/boundary-conditions/wall/inlet_side_defroster_left
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8

75

/define/boundary-conditions/modify-zone/zone-type/inlet_main_def_right
wall
define/boundary-conditions/wall/inlet_main_def_right
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/modify-zone/zone-type/inlet_main_def_left
wall
define/boundary-conditions/wall/inlet_main_def_left
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/modify-zone/zone-type/inlet_front_floor_right
wall
define/boundary-conditions/wall/inlet_front_floor_right
0
n
0
y
aluminum
y
heat-flux
n

76

0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/modify-zone/zone-type/inlet_front_floor_left
wall
define/boundary-conditions/wall/inlet_front_floor_left
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/modify-zone/zone-type/inlet_floor_rear_right
wall
define/boundary-conditions/wall/inlet_floor_rear_right
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n

77

0.8
n
0.8
;-------------------------------------------------------------; COUPLED BOUNDARIES (SO THEIR SHADOWS WON'T GET OVER WRITTEN);
;-------------------------------------------------------------;
/define/boundary-conditions/wall/wall_seats-shadow
0
n
0
y
aluminum
y
coupled
n
n
n
/define/boundary-conditions/wall/wall_seats:037-shadow
0
n
0
y
aluminum
y
coupled
n
n
n
/define/boundary-conditions/wall/wall_seats:038-shadow
0
n
0
y
aluminum
y
coupled
n
n
n
/define/boundary-conditions/wall/wall_doors
0
n
0
y
foamrubber
n
y
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_doors:034
0
n
0
y

78

foamrubber
n
y
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_doors:035
0
n
0
y
foamrubber
n
n
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_doors:036
0
n
0
y
foamrubber
n
n
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_roof:040-shadow
0
n
0
y
steel
n
n
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_floor:042
0.01

79

n
0
y
foamrubber
n
y
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_floor:042-shadow
0.01
n
0
y
foamrubber
n
y
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_floor:043
0.01
n
0
y
foamrubber
n
y
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_floor:043-shadow
0.01
n
0
y
foamrubber
n
y
n
n
n
1
n
n
1

80

n
/define/boundary-conditions/wall/wall_floor:044
0.01
n
0
y
foamrubber
n
y
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_floor:044-shadow
0.01
n
0
y
foamrubber
n
y
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_floor:045
0.01
n
0
y
foamrubber
n
y
n
n
n
1
n
n
1
n
/define/boundary-conditions/wall/wall_floor:045-shadow
0.01
n
0
y
foamrubber
n
y
n
n
n
1

81

n
n
1
n
;----------------------------------------------------; SETTING THE BC:S INCLUDED IN THE SOLAR RAY TRACING ;
;----------------------------------------------------;------------; CONVECTIVE ;
;------------;
/define/boundary-conditions/wall/wall_a-pillars
0.0007
n
0
y
steel
y
convection
n
95
n
316
n
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_b-pillar
0.0007
n
0
y
steel
y
convection
n
95
n
316
n
n
n
n
1
n
n
1
y
n
0.8
n
0.8

82

/define/boundary-conditions/wall/wall-c-pillar
0.0007
n
0
y
steel
y
convection
n
95
n
316
n
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_doors_outside
0.0007
n
0
y
steel
y
convection
n
95
n
316
y
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_doors_outside:001
0.0007
n
0
y
steel
y
convection
n
95
n

83

316
y
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_doors_outside:002
0.0007
n
0
y
steel
y
convection
n
95
n
316
y
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_doors_outside:003
0.0007
n
0
y
steel
y
convection
n
95
n
316
y
n
n
n
1
n
n
1
y
n

84

0.8
n
0.8
/define/boundary-conditions/wall/wall_roof_outside
0.0007
n
0
y
steel
y
convection
n
95
n
316
y
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_floor_outside
0.0015
n
0
y
steel
y
convection
n
95
n
316
y
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_floor_outside:048
0.0015
n
0
y
steel
y
convection

85

n
95
n
316
y
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_floor_outside:049
0.0015
n
0
y
steel
y
convection
n
95
n
316
y
n
n
n
1
n
n
1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_floor_outside:050
0.0015
n
0
y
steel
y
convection
n
95
n
316
y
n
n
n
1
n
n

86

1
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_kaross
0.001
n
0
y
steel
y
convection
n
95
n
316
y
n
n
n
1
n
n
1
y
n
0.8
n
0.8
;-------------------------------------------------------; BOUNDARIES INCLUDED IN BOTH SOLAR RAY TRACING AND S2S ;
;-------------------------------------------------------;---------; COUPLED ;
;---------;
/define/boundary-conditions/wall/wall_doors-shadow
0.001
n
0
y
foamrubber
n
y
n
n
n
1
n
y
1000
y
/define/boundary-conditions/wall/wall_doors:034-shadow
0.001
n
0
y
foamrubber
n

87

y
n
n
n
1
n
y
1000
y
/define/boundary-conditions/wall/wall_doors:035-shadow
0.001
n
0
y
foamrubber
n
y
n
n
n
1
n
y
1000
y
/define/boundary-conditions/wall/wall_doors:036-shadow
0.001
n
0
y
foamrubber
n
y
n
n
n
1
n
y
1000
y
/define/boundary-conditions/wall/wall_roof:040
0.0007
n
0
y
steel
n
y
n
n
n
1
n
y
1000
y
/define/boundary-conditions/wall/wall_seats
0
n
0

88

y
aluminum
y
coupled
n
n
n
n
1
n
y
1000
y
/define/boundary-conditions/wall/wall_seats:037
0
n
0
y
aluminum
y
coupled
n
n
n
n
1
n
y
1000
y
/define/boundary-conditions/wall/wall_seats:038
0
n
0
y
aluminum
y
coupled
n
n
n
n
1
n
y
1000
y
;------------; CONVECTIVE ;
;------------;
/define/boundary-conditions/wall/wall_ip
0.005
n
0
y
foamrubber
y
convection
n
50

89

n
318
y
n
n
n
1
n
y
1000
y
n
0.8
n
0.8
define/boundary-conditions/wall/wall_rooftop
0.0015
n
0
y
steel
y
convection
n
50
n
323
y
n
n
n
1
n
y
1000
y
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_tunnel
0.0015
n
0
y
steel
y
convection
n
50
n
333
n
n
n
n
1
n
y
1000
y

90

n
0.8
n
0.8
; WINDOWS
/define/boundary-conditions/wall/wall_windows
0.005
n
0
y
glass
y
convection
n
95
n
316
y
n
n
n
1
y
semi-transparent
y
1000
y
n
0.8
n
0.8
n
0.1
n
0.1
n
0.1
n
0.8
;------------; HEAT-FLUX ;
;------------/define/boundary-conditions/wall/wall_doors:033
0.001
n
0
y
foamrubber
y
heat-flux
n
0
y
n
n
n
1
n
y
1000
y

91

n
0.8
n
0.8
define/boundary-conditions/wall/wall-ratt
0.01
n
0
y
foamrubber
y
heat-flux
n
0
y
n
n
n
1
n
y
1000
y
n
0.8
n
0.8
;-----------------------; CONVECTIVE BOUNDARIES ;
;-----------------------;
/define/boundary-conditions/wall/wall_warm
0.0015
n
0
y
steel
y
convection
n
50
n
343
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_seatback_back
0.001
n
0
y

92

steel
y
convection
n
50
n
323
y
n
n
n
0.8
n
0.8
;----------------------; HEAT-FLUX BOUNDARIES ;
;----------------------;
/define/boundary-conditions/wall/wall_floor
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_floor:041
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n

93

0.8
n
0.8
/define/boundary-conditions/wall/wall_floor:046
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_floor:047
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
; FRAMES OF THE DOORS ON INSIDE
define/boundary-conditions/wall/wall_doors_frame
0
n
0
y
aluminum
y
heat-flux
n
0
n

94

n
n
n
1
n
n
1
n
n
0.8
n
0.8
define/boundary-conditions/wall/wall_doors_frame:004
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
define/boundary-conditions/wall/wall_doors_frame:031
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
define/boundary-conditions/wall/wall_doors_frame:032
0
n

95

0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_roof
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1
n
n
0.8
n
0.8
/define/boundary-conditions/wall/wall_roof:039
0
n
0
y
aluminum
y
heat-flux
n
0
n
n
n
n
1
n
n
1

96

n
n
0.8
n
0.8

97

98

You might also like