Documente Academic
Documente Profesional
Documente Cultură
Vehicle Dynamics
Masters Thesis
Modeling of a World Rally Championship
Car Damper and Experimental Testing of
Its Components
Author:
Richard Chahine
rchahine@kth.se
Examiner:
Mentor:
Project Commissioner:
Version 6
Stockholm, June 2011
Richard Chahine
Masters Thesis
Please note that most figures in this report are not drawn to scale or with high precision. They are
only used as an aid to make the text easier to understand.
ii
Richard Chahine
Masters Thesis
Abstract
Rally cars are driven on many different types of surfaces. Each type of surface demands a special type
of damper setup. In order to achieve optimum performance on the snow covered and icy Swedish
roads, the gravel of the Spanish rallies and the smooth tarmac of the German rallies, a large flexibility
in the possible damper settings is required. Prodrive, a British motorsport group, has been racing two
Mini Countryman as factory team cars for BMW Mini as of Rally DItalia in Sardinia in May 2011 and
has requested that hlins Racing AB equips these cars with dampers. hlins Racing AB has been
developing a damper for rally applications called the TPX. This damper is equipped with an Active
Rebound Control system (ARC). The ARC allows for high levels of grip to be achieved together with
good chassis control.
The TPX damper with its ARC system is quite complex in structure. As there are many parts in the
damper which can be altered, optimizing the damper would require a very large number of tests. A
physical model of the TPX damper with its ARC system would reduce the time spent in the lab and
help speed up the development of the damper. Prodrive would also like to a have a model of the
damper that they can use in their model of the entire car which they use to setup the cars for races
and to develop the car.
The goal of this Masters Thesis was therefore to create a model based in MatLab Simulink that
qualitatively but not necessarily quantitatively replicated the dampers behavioral trends.
Components which are very difficult to model, such as shim stacks, needed not be modeled. Their
characteristics could be measured in the lab.
During this Masters Thesis project a model for the TPX damper was created using Simulink to model
most of the physical parts of the damper. The rest of the model including its inputs and control were
taken care of by a GUI. The model functions so simulations can be performed. Plotting the results of
the simulations together with data from experimental tests was also made easy by the GUI. The
results from comparisons between the simulated damper and the real object indicate that
refinements need to be made to the model before it can be put to use as a tool for helping in
optimizing the TPX dampers construction. Hysteresis in the form of friction as well as damper
flexibility does not seem to be negligible. The variation of the oils compressibility and the dynamic
behavior of the check valves also need to be looked into. The graphs from the simulations seem to
replicate the real dampers performance trends as intended. The numerical magnitudes of the data
produced by the simulation are however not accurate. Overall the model produced during this
masters thesis seems to be a good step forward on the path to producing a useful model. Some
suggestions for the next steps in improving the model are provided.
iii
Richard Chahine
Masters Thesis
iv
Richard Chahine
Masters Thesis
Table of Contents
1
Introduction ..................................................................................................................................... 1
3.1
3.2
3.3
PDS........................................................................................................................................... 8
3.4
ARC .......................................................................................................................................... 8
3.5
3.6
4.2
4.3
4.4
4.5
6.1
6.2
6.3
6.3.1
Controls ......................................................................................................................... 32
6.3.2
Inputs ............................................................................................................................. 32
6.3.3
6.3.4
6.3.5
6.3.6
6.3.7
Starting a simulation...................................................................................................... 34
6.3.8
6.3.9
7.2
7.3
Richard Chahine
8
Masters Thesis
Test 1: P vs. Q for the shim stack as a function of ARC chamber pressure ......................... 45
8.2
Test 2: ARC chamber stiffness: flow into the chamber versus change in ARC pressure. ...... 46
8.3
Conclusions.................................................................................................................................... 49
10
11
Acknowledgements ................................................................................................................... 52
12
References ................................................................................................................................. 53
vi
Richard Chahine
Masters Thesis
1 Introduction
Rally cars are driven on many different types of surfaces. Each type of surface demands a special type
of damper setup. In order to achieve optimum performance on the snow covered and icy Swedish
roads, the gravel of the Spanish rallies and the smooth tarmac of the German rallies, a large flexibility
in the possible damper settings is required. This results in the need for dampers with many variable
parameters. In order to know how the damper behaves when each parameter is varied a large
number of tests is required.
hlins Racing AB has been developing a damper for rally applications called the TPX. Prodrive, a
British motorsport group, has been racing two Mini Countryman as factory team cars for BMW Mini
as of Rally DItalia in Sardinia in May 2011. hlins has received the order to equip these cars with
dampers. The TPX dampers have been chosen for this application. The front dampers on these cars
are equipped with Active Rebound Control (ARC); a system which causes the damping force on the
rebound stroke to be increased when the rebound chamber pressure is high for a long period of
time. This allows for high levels of grip to be achieved by having low rebound damping at frequencies
around 16 Hz. It is the motion of wheel in the 16 Hz region that has the largest effect on the cars grip.
At the same time larger rebound damping forces can be used for motional frequencies of
approximately 1.5 Hz. The cars motion in the 1.5 Hz region has the greatest affect on how the driver
perceives the car. Low rebound damping in this region would result in a feeling of the car wobbling
about. A high damping in this region would make the car feel more stable, responsive and
predictable for the driver thereby improving its chassis control. In this way the ARC system allows for
both high grip and high chassis control to be achieved.
The dampers have been tested together with the team drivers on several occasions and the results
were good. The drivers liked the new Active Rebound Control system. As there are many parameters
in the Active Rebound Control unit and the damper that can be altered, optimizing a damper with the
ARC system would require a very large number of tests. A model of the damper with the ARC would
reduce the amount of tests needed by making it possible to judge what effects certain changes to the
damper settings have. A model could therefore save considerable time in the optimization process.
This Masters Thesis was therefore commissioned by hlins Racing to create such a model that could
be used to study the dampers performance trends when it settings are altered. When a good setting
is found experimental tests can be performed with similar settings to find the best one. This way the
number of tests can be reduced.
Erik Lnnqvist, who is the mentor for this Masters Thesis, has worked on a Simulink model for a TT44
damper as his Masters Thesis dated year 2000 [1]. This report was used together with the hlins
manual for the TT44 damper [2] to understand how the model was constructed. Although the TT44
and the TPX dampers are very different both in their construction and their working principles the
basic approach to modeling the damper could be used as a start off point for building a model for the
TPX.
Richard Chahine
Masters Thesis
Richard Chahine
Masters Thesis
The secondary goal of this Masters Thesis is to provide a model that Prodrive, a British rally team,
can use in their model of the entire rally car.
Richard Chahine
Masters Thesis
Richard Chahine
Masters Thesis
Compression
check valve
located here
PDS
Middle piping
(inside rod)
Compression
chamber
Compression
piston
Middle
chamber
Compression
shim stack
located here
ARC
Rebound
shim stack
Rebound
piston
Rebound
chamber
Rebound
check valve
located here
Richard Chahine
Masters Thesis
Richard Chahine
Masters Thesis
Richard Chahine
Masters Thesis
High Speed
If high pressures are reached in the compression chamber during a compression stroke, the
compression shim stack will open allowing the damper oil to move to the middle chamber. If the
middle chamber pressure then rises, the rebound check valve will open sending oil into the rebound
chamber. This occurs at higher damper speeds in parallel to the low speed flow which continues to
take place at high speeds. The direction of the oil flow in such a case is shown in Figure 5.
3.3 PDS
The Positional Damping System (PDS) provides extra damping when the damper is nearing its
maximum compressed position. As this system does not have an impact on how the damper
functions under normal conditions it can be disregarded in this damper model.
3.4 ARC
The ARC system, shown in Figure 6, provides high damping forces at low motional frequencies
together with low damping forces at higher frequencies of motion. The ARC chamber is a separate
chamber located within the middle chamber. A tight canal is present between the ARC chamber and
the rebound chamber. As rebound pressure increases during a rebound stroke the pressure
difference between the ARC chamber and the rebound chamber causes damper oil to flow from the
rebound chamber to the ARC chamber. A shim with a hole in it, located between the canal and the
ARC chamber, serves as the main flow restrictor. The ARC chamber has flexible walls made of shims
and o-rings. As the pressure in the chamber builds up these flex outwards. One of the shims is
connected to a disk which in turn rests on the rebound shim stack. So as the pressure in the rebound
chamber builds up, the pressure in the ARC chamber increases causing its walls to flex. One of the
ARC chambers walls in turn applies a force on to the rebound shim stack thereby increasing the
pressure difference between the middle chamber and the rebound chamber required to open the
rebound shim stack.
8
Richard Chahine
Masters Thesis
ARC chamber
Figure removed at the
request of hlins Racing AB
Rebound
shim stack
Gas Reservoir
Separator
Piston
Oil Reservoir
Richard Chahine
Masters Thesis
10
Richard Chahine
Masters Thesis
4 Model Principles
Mathematical models are created for the flow restrictors, shim stacks, chamber characteristics and
the gas reservoir. Each of these mathematical models is used in several places within the Simulink
model. The data specific to each damper part is used together with a combination of these four basic
mathematical models to create the blocks that model the part.
11
Richard Chahine
Masters Thesis
ALL 5
2.5
Comp shims
Reb Shims ARC
Comp CV
Reb CV
Reservoir Shims
1.5
0.5
10
20
30
40
50
Flow [l/min]
60
70
80
90
Figure 8: The flow past the shim stacks as a function of the pressure drop over the shim stacks for 5 of
the shim stacks present in the damper.
As one can see in Figure 9, the same pressure drop over the shim stack results in two different flows
past the shim stack depending on whether the flow is being increased or decreased.
Rebound Shimstack ARC
2.5
1.5
0.5
10
20
30
40
50
Flow [l/min]
60
70
80
90
Figure 9: The flow past the rebound shim stack as a function of the pressure drop over it.
The data from the flow bench was therefore divided into two portions, one for when the flow is
increasing and one for when the flow is decreasing. The data was filtered and the red and the green
lines attained. This is shown in Figures 10 and 11, where Figure 11 is an enlarged picture of the
central portion of Figure 10. In a case in which the flow increases for a short time and then drops
again the two curves might be closer to one another. This is however not taken into consideration in
the model.
12
Richard Chahine
Masters Thesis
2.5
x 10
Blue line:
Linear model
1.5
0.5
0.5
1.5
Flow [m3/s]
-3
x 10
Figure 10: The flow past the rebound shim stack as a function of the pressure drop over it with the
filtered data and the linear model added. The filtered data is too close to the original data to be seen
in this figure. Therefore, an enlargement is provided in Figure 11 below.
6
x 10
Measured
data
1.6
1.55
Filtered data:
Green line
1.5
1.45
Blue line:
Linear model
1.4
Filtered data:
Red line
1.35
1.3
Measured
data
1.25
7.1
7.2
7.3
3
7.4
7.5
-4
Flow [m /s]
x 10
Figure 11: An enlargement of a portion of Figure 10, showing the green, the blue, and the red lines.
A linear model, adapted to the data by the author, is represented by the blue line positioned in
between the red and the green ones as shown in Figures 10 and 11. When the model operator
selects, shim stacks measured in the GUI control panel either the red or the green lines are used to
decide the flow past a shim stack for a certain pressure drop over the shim stack depending on
whether the flow is increasing or decreasing. If shim stacks modeled is selected the blue line is
used to decide the flow past a shim stack for a certain pressure drop.
13
Richard Chahine
Masters Thesis
This is done for all the shim stacks and check valves in the model. In the case of the check valves
which are also shim stacks only the linear model is used; this being the red line shown in Figure 12.
Both the rebound and compression check valves, not only the rebound check valve shown in Figure
12, displayed what seems to be a very dynamic behavior. This was suspected by hlins engineers.
The exact data from the flow bench measurement for the check valves was therefore not used but
the linear model shown in Figure 12 was used instead. The reservoir check valve could not be placed
in the flow bench without the construction of a special adapter. The reservoir check valve data was
set to be the slope of the line for the rebound check valve together with half the preload of the
compression check valve. This should be an acceptable temporary solution.
5
x 10
3.5
Linear model:
Red line
2.5
Measured
data
2
1.5
1
0.5
0
0.5
1
Flow [m3/s]
1.5
-3
x 10
Figure 12: The flow past the rebound check valve as a function of the pressure drop over the check
valve.
14
Richard Chahine
Masters Thesis
is the change in chambers volume
Richard Chahine
Masters Thesis
The initial gas pressure can be used together with the oil reservoir pressure to calculate the force on
the piston which separates the gas and the oil according to equation 6.
(6)
where:
is the force on the separator piston
is the oil reservoir pressure
is the pressure in the gas chamber
is the area of the separator piston
The separator pistons velocity and position can then be calculated by integrating its acceleration
which can be calculated from the force on the separator piston according to equation 7. The initial
conditions for both the separator pistons velocity and the separator pistons position are considered
to be their values at the beginning of the simulation when the damper is standing still and can
therefore be set to zero.
(7)
where:
is the separator pistons mass
is the separator pistons acceleration
The separator pistons velocity can then be used in equation 8 to calculate the rate of change of oil
reservoir volume. This rate of change of volume is the equivalent of a flow into or out of the oil
reservoir.
(8)
where:
is the oil flow in or out of the oil reservoir
is the separator pistons velocity.
The separator pistons position can then be used to calculate the change in gas reservoir volume
which in turn can be used with equation 5 to calculate the new instantaneous gas pressure in the gas
reservoir. This process is then continuously looped to calculate the gas pressure at each point in the
simulation timeline.
Richard Chahine
Masters Thesis
5 Model Construction
The model of the TPX damper is built in Simulink. The different levels in the model are described
below. Figure 13 shows a tree diagram containing only the major subsystems of the model. The input
to the damper model is the damper velocity curve that the model is to follow. The main output from
the model is the damping force that the damper gives. The highest level in the simulink model is
shown in Figure 14. Figures 15 to 34 show the contents of the subsystem that is discussed in the
corresponding section.
Model
ARC Chamber
Compression
Chamber
Compression
check valve
GUI input
control
Volume
Calculations
Compontents
Pressure sum
on main
piston =>
Damping
force
Compression
low speed
Main
Compression
Shim Stacks
Main
Rebound Shim
Stacks
Gas Reservoir
Oil Resrvoir
Reservoir
Shim Stacks
and Reservoir
Pipes
Gas Reservoir
Middle Piping
Rebound
Chamber
Rebound
check valve
Separator
Piston
Figure 13: A tree diagram showing the major subsystems of the model.
damping_force
To Workspace2
Damping force
Damper velocity curve
Damper velocity
Piston position
Results
position
To Workspace1
Damper_velocity
To Workspace
Figure 14: Overall model view, the highest level of the model.
17
Rebound low
speed
Richard Chahine
A
Masters Thesis
This subsystem contains the blocks necessary to make it possible for the GUI to control the damper
velocity input signal to the Simulink model.
GUI_input_signal
+ve sinus
1
Damper velocity curve
-ve sinus
+ve step
Multiport
Switch
-ve step
TPX damper
This subsystem contains the blocks that make up the physical model of the damper. It contains three
of its own subsystems: Volume Calculations, Components subsystem, and Force on Main Pistons.
Clock
1
Damper velocity
Simulation time
Cormpression volume
Compression volume dot
Rebound volume
Rebound volume dot
Piston position [m]
Volume calculations
Cormpression volume
Compression volume dot
Rebound volume
Rebound volume dot
Components
1
Damping force
18
Richard Chahine
Masters Thesis
5
Piston position [m]
-C-
1/s
1
Damper velocity=
piston velocity
-1
Piston Position
Initial Compression
chamber volume
-K-
Piston_area_compression
Integrator
Limited
Add
-K-
Piston Velocity
1
Cormpression volume
2
Compression volume dot
Piston_area_compression1
-CInitial Rebound
chamber volume
-KPiston_area_rebound
-1
Add1
-KPiston_area_rebound1
3
Rebound volume
4
Rebound volume dot
B.B Components
This is the main subsystem at this level. Its contents represent the dampers actual components and
include: rebound check valve, compression check valve, rebound low speed, compression low speed,
main shim stacks, compression chamber, rebound chamber, middle piping, gas and oil reservoirs, as
well as the ARC chamber.
19
20
Middle Piping
To Workspace2
Middle_chamber_pressure
ARC Chamber
3
Rebound volume
4
Rebound volume dot
Flow to ARC
ARC Pressure
1
Cormpression volume
2
Compression volume dot
Rebound Chamber
2
Rebound chamber
oil pressure
1
Compression chamber
oil pressure
To Workspace1
Rebound_chamber_pressure
To Workspace
Compression_chamber_pressure
Compression chamber
Compression check valve opens during rebound and rebound check valve open during compression!
Richard Chahine
Masters Thesis
Richard Chahine
Masters Thesis
-K1/k compression cv
1
Middle piping pressure
-C0
rebound cv preload
1
Flow to rebound chamber
from middle chamber
via check valve
1
Compression
Chamber pressure
-K-
2
Middle piping pressure
1/k compression cv
-C0
compression cv preload
1
Flow to compression chamber
from middle chamber
via check valve
21
Richard Chahine
Masters Thesis
2
Rebound Chamber pressure
|u|
Add
1
Middle piping pressure
-K-
Abs
Gain
Divide
Sqrt
-K-
-K-
1
Flow to middle piping
from rebound chamber
A_LS_rebound
Cd_rebound
rho_oil
rho_oil
-1
Representation of
Q=Cd*A*sqrt(2*delta_p/rho_oil)
Add
|u|
Abs
Gain
u
Divide
Sqrt
-K-
-K-
Cd_compression
1
Flow to middle piping
from compression chamber
A_LS_compression
rho_oil
rho_oil
Representation of
Q=Cd*A*sqrt(2*delta_p/rho_oil)
-1
Allows the flow to be -ve (ie reverse direction) if delta_p is negative
even though sqrt and abs are used when calculating the flow
22
Richard Chahine
Masters Thesis
Shimstack_measured_modeled
Lookup Table2 +
du/dt
Multiport
Switch1
Switch3
Derivative
Scope2
1
Middle chamber pressure
3
Rebound chamber
oil pressure
Add3
-C-
Divide1
Switch1
-C-
rebound preload
ARC_on_off
2
ARC pressure
1
Flow past rebound shim stacks
Multiport
Switch2
Lookup
Table (2-D)
1
Compression chamber
oil pressure
2
Middle chamber oil pressure
Add2
Add
-Ccompression preload
Shimstack_measured_modeled
Divide
-C-
Switch
Multiport
Switch
du/dt
Lookup Table +
1
Flow past compression shim stacks
Switch2
Derivative1
Lookup Table1 -
23
Richard Chahine
Masters Thesis
4
Compression volume dot
2
Flow to middle piping from compression chamber
delta_pc/dt
5
flow past compression shim stack
1
Flow to compression chamber from middle chamber via check valve
Add
3
Compression volume
Divide
1
x1/s
o
Compression chamber
Integrator
oil pressure
p_gas_initial
p gas initial
beta_oil
beta_oil
delta_p/dt
Add
1
xos
1
Rebound chamber
oil pressure
Integrator
Divide
p_gas_initial
p gas initial
3
Rebound volume
beta_oil
beta_oil
24
Richard Chahine
Masters Thesis
delta_p_m/dt
Divide
4
Flow past compression shim stacks to middle chamber
p_gas_initial
middle_piping_volume
Integrator
1
Pressure in
middle piping
p_gas_initial
1
xo s
beta_oil
beta_oil
7
Flow past rebound check valve
Oil_reservoir_pressure
To Workspace
Oil Reservoir Pressure
Oil Reservoir
1
Pressure in middle piping
1
Oil flow into reservoir
Figure 28: The contents of the Gas and Oil Reservoirs subsystem.
B.B.J.A
The gas reservoir subsystem makes use of the equations presented in section 4.4 to calculate the
change in gas pressure and the rate of change of the oil reservoir volume.
25
Richard Chahine
Masters Thesis
1
Oil reservoir pressure
gas pressure
-K-
Separator piston
Gas reservoir
-C-
Separator piston
velocity
-K-
Separator piston
position
-1
separator piston
area2
2
Oil reservoir volume dot
Gas_reservoir_pressure
To Workspace
-Kseparator piston area1
1
Gas Reservoir volume
Add1
-CInitial oil volume
Figure 29: The contents of the Gas Reservoir and Separator Piston subsystem.
B.B.J.A.A
Gas reservoir
This subsystem contains equation 5 which depicts the change in gas pressure as a function of the
displacement of the separator piston.
2
Initial gas volume
u
Divide
1
Gas reservoir volume
-C-
Math
Function
Product
1
Gas presssure
B.B.J.A.B
Separator piston
This subsystem contains the equations describing the motion of the separator piston as discussed in
section 4.4.
1
Net force on separator piston
separator piston
acceleration 1/s
Divide
Integrator
-C-
1
xo s
Integrator1
26
1
Separator piston
velocity
2
Separator piston
position
Richard Chahine
Masters Thesis
B.B.J.B
Oil reservoir
The oil reservoir subsystem houses the chamber characteristics described in section 4.3. As discussed
earlier these are used to calculate the pressure in the reservoir.
2
Oil reservoir volume dot
delta_p_res/dt
3
Oil flow into reservoir
Add
1
Oil reservoir volume
1
xos
1
Oil Reservoir
Pressure
Integrator
Divide
p_gas_initial
p_gas_initial
beta_oil
beta_oil
B.B.J.C
The subsystem named Reservoir shim stacks and reservoir pipes contains a model of the reservoir
shim stacks. These shim stacks are modeled in the same way as the compression and rebound shim
stacks mentioned earlier. The reservoir low speed flow was calculated using the restriction equation
in a similar way to the compression low speed flow and the rebound low speed flow.
Shimstacks
Shimstack_measured_modeled
-K-C-
reservoir preload
Constant
Switch
slope control
1
Oil reservoir pressure
Multiport
Switch
delta_p
Lookup Table +
du/dt
Switch3
Derivative1
Scope1
2
Pressure in middle piping
u
Divide
Sqrt
-KCd_reservoir
-K-
A_LS_reservoir
rho_oil
rho_oil
-1
Allows the flow to be -ve (ie reverse direction) if delta_p is negative
even though sqrt and abs are used when calculating the flow
Check valve
both r plus since it
-1
-Cworks in the opposite direction
Reservoir cv preload
0
Constant1
-KSwitch1
1/k_reservoir_cv
Figure 33: The contents of the Reservoir Shim stacks and Reservoir pipes subsystem.
27
1
Oil flow into reservoir
Richard Chahine
Masters Thesis
-KPiston_area_compression
-KPiston_area_rebound
Add
-KShaft Area
28
1
Force
Richard Chahine
Masters Thesis
6 GUI Control
A MatLab GUI (graphical user interface) has been built in order to facilitate use of the model and its
initiation files as well as offer the possibility to provide additional on demand calculations for the
model. Nineteen files are used to run the model. One file is the Simulink model. Six of the files that
are used contain shim stack data. Two of the files are used to process the shim stack data. Seven of
the files process data to enable plotting of certain figures. One file contains all the dampers structural
data along with some other data needed to describe the damper. Then there are the two GUI files.
One is just a figure of the GUI and the other is a large file that is used to control all the files as well as
produce certain graphs, perform the calculations for the reservoir and rebound controls and process
other initiation data and inputs from the user.
Getting the different files of the model to interact with one another is a non-trivial task as there are
three workspaces to consider as well as the adjustment of the data vectors for plots, reading
experimental data, adjusting experimental data etc. The Simulink model usually makes use of its own
workspace. A command line integrated in the GUI but executed in the MatLab workspace makes the
Simulink model run in the BASE workspace. The GUI also makes use of its own workspace. Some of
the GUIs code is run in its own private workspace while other parts are sent to the base workspace
for execution. The data that is to enter the Simulink model is, however, always sent to the base
workspace. Simulink will also report results to the base workspace which the GUI then reads to
produce graphs of the results as well as provide result data.
Two functions that the GUI handles on its own are the Reservoir Needle Control and the Low Speed
Rebound Control. These are run in the GUIs own workspace. The GUIs control panel is shown in
Figure 38.
Richard Chahine
Masters Thesis
panel the user of the model can type in the number of clicks from the fully closed position that the
reservoir needle has been placed in. The number of clicks is then used by the GUI to calculate the
opening area between the needle and the hole in which it is placed. Thereby the area that restricts
the low speed flow in and out of the oil reservoir can be calculated in the following way.
R1
R2
20
Richard Chahine
Masters Thesis
seen in Figure 37. The area restriction for the low speed rebound damping as a function of the
number of clicks can be calculated in the following way.
The vertical movement, x, of the cylinder blocking the holes can be calculated from the number of
clicks that have been chosen by using equation 11.
(11)
The size of the area of the holes that is not covered by the cylinder needs to be determined. This is
done by regarding one hole at a time. How the open area of the hole varies with the cylinders vertical
position is then calculated using equations 12, 13 and 14 together with Figure 38.
Figure 38: One of the three holes that controls the low speed rebound flow.
(12)
where:
is the vertical distance at which the hole that is currently being regarded is positioned
from the cylinder when the cylinder is in its fully closed position.
is the vertical distance between the holes upper edge and the edge of the cylinder that is
covering the rest of the hole.
31
Richard Chahine
Masters Thesis
(13)
where:
is the angle shown in Figure 38.
is the radius of the hole shown in Figure 38.
(14)
where:
is the holes area.
The area calculation can then be done for all three holes to give the total opening area of the flow
restrictor between the middle chamber and the rebound chamber.
Explanation
A heave sine signal that starts at zero and goes
in the positive direction
A heave sine signal that starts at zero and goes
in the negative direction
A step signal that starts at zero and goes in the
positive direction
A step signal that starts at zero and goes in the
negative direction
A recorded damper velocity file, note: this
feature is not complete as there has been no use
for it during the projects proceedings.
Richard Chahine
Masters Thesis
Richard Chahine
Table 2: Selectable output plots
As a function of simulation time
Damping Force
Compression Chamber Pressure
Rebound Chamber Pressure
Middle Chamber Pressure
Oil Reservoir Pressure
Gas Reservoir Pressure
ARC Chamber Pressure
Damper Position
Damper Velocity
Masters Thesis
Richard Chahine
Masters Thesis
Simulation 1
30
30
Sinus +ve
0.1
1
2
OFF
Simulation 2
30
30
Sinus +ve
0.25
1
2
OFF
The difference in the results generated by the two simulations is clearest when comparing the
damping force curves. As one can see in the graphs in Figure 39 the difference in the damping force is
greatest during the rebound stroke at the lower damper velocity of 0.1 m/s. At the higher damper
velocity of 0.25 m/s the difference shrinks.
Damping Force
1000
800
600
400
200
-400
-600
-800
0.2
0.4
0.6
0.8
1
1.2
Simulation Time [s]
1.4
1.6
1.8
Figure 39: Damping force as a function of simulation time for Simulations 1 and 2.
The reason for this can be seen in Figure 10 in section 4.2 where the line corresponding to the shim
stacks modeled function crosses the Y axis at a higher point than the one for the shim stacks
35
Richard Chahine
Masters Thesis
measured. A higher pressure difference between the two chambers involved is therefore required in
order for the shim stacks to begin allowing a flow past them in the case when the shim stacks
modeled function is selected in comparison to the case in which the shim stacks measured function is
selected. This occurs in the region when the shim stacks just began opening. The difference between
the damping forces at higher damper velocities is due to that the linear model and the measured
model lines in Figure 10 dont exactly overlap.
Compression
chamber pressure
sensor
Test rig
Rebound chamber
pressure sensor
TPX damper
Middle
chamber
pressure
sensor
Oil reservoir
pressure
sensor
Figure 40: The experimental setup used in the D14 test cell at hlins Racing AB.
A special TPX damper with holes drilled for pressure sensors was used. Four pressure sensors were
positioned so that one measured the compression chamber pressure, another the rebound chamber
pressure, a third the middle chamber pressure and a fourth the oil reservoir pressure. Additional data
that could be gathered from this test rig included the damping force, the damper position and the
time elapsed during the test. Later, when files from the test cell were loaded into the GUI the
damper position was derived to obtain the damper velocity.
36
Richard Chahine
Masters Thesis
Amplitude
[mm]
30
3.18
1.06
7.95
2.65
15.91
5.31
26.53
7.96
26.53
7.96
Reservoir
Needle clicks
30
30
30
30
30
30
30
15
15
41*
41*
Rebound Low
Speed clicks
30
30
30
30
30
30
30
15
15
60*
60*
Simulation: shim
stacks modeled
500
400
300
200
Simulation: shim
stacks measured
100
0
Test 1
-100
-200
-300
-400
0.2
0.4
0.6
0.8
1
1.2
1.4
Simulation Time [s]
1.6
1.8
Richard Chahine
Masters Thesis
From the damper position curves in Figure 42 one can see that the algorithm in the GUI code that
adjusts the data gathered from the test so that both the simulation curve and the experimental test
curve have the same starting time is not perfect.
Damper position
0.03
Test 1
0.025
0.02
Simulation:
shim stacks
modeled and
measured
(overlap)
0.015
0.01
0.005
-0.005
0.2
0.4
0.6
0.8
1
1.2
Simulation Time [s]
1.4
1.6
1.8
38
Richard Chahine
Masters Thesis
x 10
Simulation: shim
stacks measured
1.8
1.7
Simulation: shim
stacks modeled
1.6
1.5
1.4
Test 1
1.3
1.2
0
0.5
1
1.5
Simulation Time [s]
Figure 43: Rebound chamber pressure as a function of simulation time for test 1.
6
x 10
1.55
Simulation: shim
stacks measured
1.5
1.45
1.4
Simulation: shim
stacks modeled
1.35
1.3
Test 1
1.25
1.2
1.15
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Simulation Time [s]
1.6
1.8
Figure 44: Compression chamber pressure as a function of simulation time for test 1.
The higher system pressure can be observed in Figures 45 and 46 which shown the oil reservoir and
middle chamber pressures respectively. The lower oil reservoir pressure in the experimental case
could be due to the inaccurate way of filling the damper with gas. So the exact pressure of the gas
the damper is filled with is a little uncertain.
39
Richard Chahine
Masters Thesis
6
x 10
Simulation: shim
stacks modeled
and measured
(overlap)
1.4
1.35
1.3
Test 1
1.25
1.2
0.2
0.4
0.6
0.8
1
1.2
1.4
Simulation Time [s]
1.6
1.8
2.2
Figure 45: Oil reservoir pressure as a function of simulation time for test 1.
6
x 10
Simulation: shim
stacks measured
Simulation: shim
stacks modeled
1.4
1.35
1.3
Test 1
1.25
1.2
0.2
0.4
0.6
0.8
1
1.2
1.4
Simulation Time [s]
1.6
1.8
Figure 46: Middle chamber pressure as a function of simulation time for test 1.
When studying the damping force, compression chamber pressure, rebound chamber pressure,
middle chamber pressure, and oil reservoir pressure as a function of damper velocity as shown in
Figures 47 to 51 respectively it is clear that the curves for the experimental data are set much wider
apart than those for the simulation data. The reason for this could be flexibility in the damper and
friction between the pistons and the cylinder walls. Another source of this could be the shim stacks
surface flexibility. The shim stack surface may take different shapes during different accelerations of
the flow past the shim stack. The shim stack may also flex out and slightly without opening when the
pressure on its underside builds up.
40
Richard Chahine
Masters Thesis
Test 1
Simulation: shim
stacks modeled
600
400
200
Simulation: shim
stacks measured
-200
-400
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02 0.04
Damper Velocity [m/s]
0.06
0.08
0.1
1.6
x 10
Simulation: shim
stacks measured
1.55
1.5
Simulation: shim
stacks modeled
1.45
1.4
1.35
Test 1
1.3
1.25
1.2
1.15
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02 0.04
Damper Velocity [m/s]
0.06
0.08
0.1
Figure 48: Compression chamber pressure as a function of damper velocity for test 1.
41
Richard Chahine
Masters Thesis
6
1.9
Simulation: shim
stacks measured
1.8
Simulation: shim
stacks modeled
x 10
1.7
1.6
1.5
1.4
Test 1
1.3
1.2
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02 0.04
Damper Velocity [m/s]
0.06
0.08
0.1
Figure 49: Rebound chamber pressure as a function of damper velocity for test 1.
6
1.45
x 10
Simulation: shim
stacks modeled
and measured
(overlap)
1.4
1.35
1.3
Test 1
1.25
1.2
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02 0.04
Damper Velocity [m/s]
0.06
0.08
0.1
Figure 50: Middle chamber pressure as a function of damper velocity for test 1.
6
1.45
x 10
Simulation: shim
stacks modeled
and measured
(overlap)
1.4
1.35
1.3
Test 1
1.25
1.2
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02 0.04
Damper Velocity [m/s]
0.06
0.08
0.1
Figure 51: Oil reservoir pressure as a function of damper velocity for test 1.
42
Richard Chahine
Masters Thesis
An example of the graph produced when the mean function for the experimental data is selected in
the GUI is shown in Figure 52. One should still look at the original data before looking at this graph to
make sure there arent too large cycle to cycle variations which could be caused by changes in
damper temperature for example.
Damping force vs Time MEAN
800
Test 1
600
Simulation: shim
stacks modeled
400
200
Simulation: shim
stacks measured
-200
-400
0.2
0.4
0.6
0.8
1
1.2
Time [s]
1.4
1.6
1.8
Figure 52: Damping force as a function of damper velocity for test 1 with the mean function selected
for the experimental test data when loading it into the GUI.
43
Richard Chahine
Masters Thesis
44
Richard Chahine
Masters Thesis
8 ARC Implementation
The ARC system is a complex one and is therefore very difficult to model. The ARC chamber has
flexible walls. The flexibility of each of these walls varies from the others. In order to get the
performance of the ARC system into the model two experimental tests were devised. The aim of the
first test is to obtain the flow past the rebound shim stack, with the ARC system connected to it, as a
function of the ARC pressure as well as the pressure drop over the shim stack, i.e. between the
rebound chamber and the middle chamber. The goal of the second test is to attain a stiffness for the
ARC chamber in terms of flow into the chamber versus change in ARC pressure.
8.1 Test 1: P vs. Q for the shim stack as a function of ARC chamber
pressure
Figure 53 below shows the experimental setup for this first test. The rebound piston is placed on a
piston shaft together with the rebound shim stack and the ARC system. A gas pressurized oil
reservoir is connected to the cavity in the shaft and pressurizes it.
Gas reservoir
Separator piston
Shaft
Direction
of Oil flow
Rebound piston
45
Richard Chahine
Masters Thesis
Flexible
walls
O-rings
Figure 54: An enlargement of the ARC chamber showing the hole drilled between the shaft cavity and
the ARC chamber.
The oil reservoir has to be large in order to compensate for slight oil leakage out of the ARC chamber.
This setup can then be run in the flow bench with the flow bench running backwards so that the oil
flows in the direction shown in Figure 53. The pressure on the upper side of the piston should be
equivalent to the rebound chamber pressure and the pressure on the lower side representative of
the middle chamber pressure. Using this setup the flow past the rebound shim stack as a function of
the ARC pressure as well as the pressure drop over the shim stack should be obtained.
8.2 Test 2: ARC chamber stiffness: flow into the chamber versus change in
ARC pressure.
In this test the construction used in test 1 can be reused. The oil reservoir connection is replaced by a
cylindrical plunger as shown in Figure 55.
Cylindrical
plunger
Pressure
sensor
Figure 55: The experimental setup for Test 2: A plunger replaces the oil reservoir.
The construction with the hole drilled between the cavity in the shaft and the ARC chamber as
shown in Figure 54 is maintained. A pressure sensor that measures the pressure inside the shaft is
placed on the piston shaft. The plunger can then be moved up and down while the pressure as well
as the plungers position is recorded. Using the plungers position as a function of time one can then
determine the flow into the ARC chamber. The pressure sensor will then give the pressure inside the
46
Richard Chahine
Masters Thesis
shaft which should be almost equal to the pressure inside the ARC chamber if the canal joining the
two is large enough. Using the pressure as a function of the flow into the ARC chamber, a type of
stiffness for the ARC chamber represented by equation 15 can be determined.
(15)
where:
is a stiffness
is the change in pressure in the ARC chamber
is the flow into or out of the ARC chamber
By having the plunger operated using a regulator that aims to keep the pressure in the shaft constant
the oil leakage out of the ARC chamber can be determined.
8.3
Model Implementation
The ARC system has been implemented in the model by using two blocks. The first block is the ARC
chamber subsystem whose contents are shown in Figure 56.
OBS!! controllera hur det r tnkt att look up tablen ska handskas med negativa flden, dvs fld
1
Rebound Chamber pressure
ARC pressure
|u|
Add
Representation of
Q=Cd*A*sqrt(2*delta_p/rho_oil)
Abs
u
Gain
Divide
Sqrt
-K-
-K-
Cd_compression
A_LS_compression
rho_oil
rho_oil
Flow in to ARC
chamber
Pressure in ARC
1
xos
Integrator1
Lookup Table
-1
1
ARC Pressure
ARC_on_off
2
Flow to ARC
0
Constant
temporary to run without ARC1
Multiport
Switch1
flow to ARC
Figure
56: ARC
chamber
contents.
MIddle chamber
pressure on
backside of upper Osubsystem
ring
The restriction model discussed
in section 4.1 is used to calculate the flow into the ARC chamber
Behvs den hr? kommer trycket i melan kammaren trycka p vre shimstacken med oringen i ARC kammaren?
Can i measure pressure in ARC chamber and then link that directrly to the shimstack opening via experimental tests?
or
pressure
in ARCthe
chamberARC
directly to damping
force (probably
not but could
try to find a multiplication
or addition
from difference in pressure between
chamber
and
theprobably
rebound
chamber.
A look up table
factor that is proporitonal to the ARC pressure that can be put in to get a shift in damping curve from that given by a test of a TPX without ARC)
containing the data from Test 2 in section 8.2 is used to obtain the change in ARC chamber pressure
from the flow into the ARC chamber. This change in pressure is then integrated, with the initial
pressure in the ARC chamber set to the initial gas reservoir pressure. The new ARC chamber pressure
can then be used to calculate the new instantaneous flow through the restriction. The flow to the
ARC chamber is then sent to the rebound chamber block shown in Figure 26 where it is added to the
flow out of the rebound chamber. The second portion of the ARC implementation is found in the
Main Rebound Shim Stacks subsystem shown in Figure 23. This block receives the ARC pressure
signal from the ARC chamber subsystem and uses it as input to a look up table containing the data
47
Const
temporary to run
Richard Chahine
Masters Thesis
from Test 1 in section 8.1. The flow past the rebound shim stacks can then be obtained as the ARC
chamber pressure and the pressure difference over the rebound shim stacks are known.
48
Richard Chahine
Masters Thesis
9 Conclusions
During this Masters Thesis project a model for the TPX damper was created using Simulink blocks to
represent the physical parts of the damper. The rest of the model including its inputs and control
were taken care of by a GUI. The model functions so simulations can be performed. Plotting the
results of the simulations together with data from experimental tests was also made easy by the GUI.
The results from comparisons of the simulated damper to the real object indicate that refinements
need to be made to the model before it can be put to use as a tool for helping in optimizing the TPX
dampers construction. Hysteresis in the form of friction as well as damper flexibility does not seem
to be negligible. The variation of the oils compressibility and the dynamic behavior of the check
valves also need to be analyzed in more detail. The surface flexibility of the shim stacks should also
be studied further. The graphs from the simulations seem to replicate the real dampers performance
trends as intended. The numerical magnitudes of the data produced by the simulation are however
not accurate. It looks almost like there is a scaling factor problem but that might not necessarily be
the case as no such fault has yet been found in the code. Overall the model produced during this
Masters Thesis seems to be a good step forward on the path to producing a very useful model. In
section 10 below some suggestions for the next steps in improving the model are provided.
49
Richard Chahine
Masters Thesis
50
Richard Chahine
Masters Thesis
Produce the fixtures required to perform Tests 1 and 2 described in section 8, and fill the ARC
subsystem in the model with the data from the results of these tests.
Perform a larger number of tests with the experimental damper. Consider performing tests
with a kind of step input in order to help judge sources of error. This could be useful when
analyzing at shim stack and check valve motion.
Perform tests with the shim stacks in the flow bench with varying degrees of acceleration of
the flow and study how the shim stack characteristics vary when the flow increases and then
decreases. Do the lines move closer together than the ones shown in Figure 10?
Create a fixture or find a way to place the reservoir check valve in the flow bench in order to
obtain the
versus Q curves for it.
Find a better way to model the check valves behavior as their dynamic behavior most likely
needs to be taken into account.
Study the interaction between shim stacks.
Study the causes of the hysteresis in the damping curves, could it be due to:
o The separator piston friction.
o Flexibility in the damper construction.
o Main piston friction, what happens if the sealing ring type is changed? Do the test
curves change?
o Surface flexibility of the shim stacks.
Measure how the oil compressibility varies as a function of temperature and pressure.
After all the above have been considered, one can look again into the ARC system and
compare its performance to the data produced by the model. Run the damper in the test rig
with the ARC system in place and without it. Do the same in the model and then compare the
results.
Flow bench tests can also be performed with varied number of preload clicks for the
reservoir shim stack.
51
Richard Chahine
Masters Thesis
11 Acknowledgements
The author would like to thank in particular Magnus Danek, Erik Lnnqvist, Nils Gran Nygren,
Fredrik Pettersson, and David Bolander as well as the rest of the Car Division at hlins Racing AB for
their support and guidance during this masters thesis project. Thanks are also orderly due to Docent
Lars Drugge at the Department of Vehicle Dynamics at KTH for his advice concerning the thesiss
general structure.
52
Richard Chahine
Masters Thesis
12 References
[1] Lnnqvist, Erik. Masters Thesis: A Physical Model and an Experimental Comparison of Characteristics of a
High Performance Damper. Thesis. Royal Institute of Technology, 2000. Stockholm, 2000. Print.
[2] Burness B. and Nygren N-G., Inside the hlins TT44 manual, hlins Racing AB, 1999.
[3] hlins database: drawings and CAD model of the TPX44 damper, last accessed May 2, 2011.
[4] Karlsson, Arne. Formelsamling i Strmningsmekanik. Upplaga 2.11. Mekanik, KTH, 29 August
2007. Print.
[5] Young, H. D., and R. A. Freedman. Sears and Zemansky's University Physics: with Modern Physics.
11th ed. San Francisco: Pearson, Addison Wesley, 2004. 418. Print.
53