Optimal vehicle suspension characteristics for increased structural fatigue life Braham Breytenbach Public defence 28 Mei 2010
Optimal vehicle suspension characteristics for increased structural fatigue life
Braham Breytenbach
Public defence
28 Mei 2010
Problem statement
� Highly competitive market for vehicles
� Reduce vehicle mass, increase payload
� Road loads on structure must be reduced!!!
Problem statement
Questions:
1. Suspension characteristics for structural life?
2. Are the optimal characteristics sensitive to payload?
Approach
Mathematical modelling
�Experimentally validated
�Computationally efficient
Suspension optimisation for life
�Dynamic-Q algorithm
�Optimal spring and damper characteristic
�Different load cases considered
Test vehicle
� Land Rover Defender 110 -> fully instrumented
� Hydro-pneumatic 4S4 suspension system
Instrumentation
� Suspension force load cell
� Strain gauges => suspension mounting
Field tests
Discrete obstacles:
Ride comfort mode Handling mode
Field tests
Rough road / Random terrain:
Ride comfort mode Handling mode
Field tests
Data repeatability:
0.5 1 1.5 2 2.5 3-10
0
10Body Vertical Accelerations
LR
Acc.[
m/s
2]
0.5 1 1.5 2 2.5 3-10
0
10
RR
Acc.
[m/s
2]
0.5 1 1.5 2 2.5 3-20
0
20
LF
Acc.
[m/s
2]
Time [s]
Run 16Run 20Run 22
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 30
2000
4000
6000
8000
Load C
ell
Forc
e [
N]
Left Rear Suspension Forces
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 30
2000
4000
6000
8000
Calc
ula
ted L
R F
orc
e [
N]
Time [s]
Run 16
Run 20
Run 22
Mathematical model
� Linear models inadequate
� 7-DOF model with non-linear force characteristics
� Stresses predicted by quasi-static approach
� Damage estimated by Miner’s rule Z
Friction characterisation
Experimental Setup
Friction characterisation
Test input
100 110 120 130 140 150 160 170 180 190 200
-0.1
-0.05
0
0.05
0.1
Displacement 1
Dis
pla
cem
ent [m
]
250 300 350 400 4500
0.05
0.1
Displacement 2
Dis
pla
cem
ent [m
]
240 260 280 300 320 340 360 380 400 4200
0.05
0.1
Displacement 3
Time [s]
Dis
pla
cem
ent [m
]
Spring
Friction
Tester
Friction characterisation
Test results
-0.1 -0.05 0 0.05 0.13500
4000
4500
5000
5500Force Displacement
Displacement [m]
Fo
rce [N
]
Measured
Model
-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2-500
-400
-300
-200
-100
0
100
200
Experimental Friction Characteristics
Velocity [m/s]
Frictio
n F
orc
e [N
]
500 kPa
1500 kPa
2000 kPa
2500 kPa
3000 kPa
Friction characterisation
Test results
300 320 340 360 380 400 420 440 460 480 500-500
-400
-300
-200
-100
0
100
200
300
Time [s]
Frictio
n F
orc
e [N
]
Friction Force for Displacement 3
Measured
Model
Model correlation
Discrete obstacles: body vertical accelerations
0.5 1 1.5 2 2.5 3 3.5-10
-8
-6
-4
-2
0
2
4
6
8
10
Time [s]
Re
ar
Ac
cele
ratio
n [m
/s2]
Measured
Linear Pitch Bounce Model
0.5 1 1.5 2 2.5 3 3.5-10
-8
-6
-4
-2
0
2
4
6
8
10
Time [s]
Rear
Vert
ical A
ccele
ratio
n [m
/s2]
Measured
Non-linear Full Vehicle Model
Linear model 7-DOF non-linear model
Model correlation
Discrete obstacles: suspension forces
Linear model 7-DOF non-linear model
0.5 1 1.5 2 2.5 3 3.50
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Time [s]
Rear
Suspe
nsio
n F
orc
e [N
]
Measured
Linear Pitch Bounce Model
0.5 1 1.5 2 2.5 3 3.50
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Time [s]
Re
ar
Su
sp
en
sio
n F
orc
es
[N
]
Measured
Non-linear Full Vehicle Model
Model correlation
Rough road / Random terrain:
� Low speed correlation poor -> tyre and friction models!!!
� High speed dynamic correlation excellent (error < 5%)
� Damage correlation acceptable (error < 30%)
� Better correlation in ride comfort mode
Mathematical optimisation
� Objective function: structural damage over rough terrain
� Design variables: � Static gas volume (pneumatic spring stiffness)
� Damper scale factor
� Constraint functions:� Loss of wheel contact < 10% of total time
� Bump stop contact unacceptable
� Two load cases considered:� Unladen Land Rover 2.2 ton
� Fully laden Land Rover 4.5 ton
Mathematical optimisation
Cost function visualisation:
Unladen 2.2 ton Fully laden 4.5 ton
Mathematical optimisation
Monte Carlo simulation:
Unladen 2.2 ton Fully laden 4.5 ton
Mathematical optimisation
Results:
Unladen vehicle
Static Gas Volume
DamperScale Factor
Damage as % of baseline
Fatigue damage 0.4-0.8ℓ 0.4 29%
Ride comfort 0.5ℓ 0.3 -
Handling 0.1ℓ 3 -
Fully Laden vehicle
Static Gas Volume
DamperScale Factor
Damage as % of baseline
Fatigue damage 0.5-0.8ℓ 0.7 86%
Mathematical optimisation
Robust optima:
Step into the feasible design space:
[ ]
( )
( )xg
xgu
uxx
xxxrobust
∇
∇=
⋅×
∆+
∆+=
***
1
Mathematical optimisation
Robust optima:
Spring Static Volume [l]
Dam
per
Sca
le F
acto
r
Unladen
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0.5
1
1.5
2
2.5
3
3.5
4
Cost Function
Feasible
Bump Stop Constraint Active
Optimum X*=[0.49; 0.34], F*=22.3%, Std. Dev. =1.2536%
Spring Static Volume [l]
Dam
per
Sca
le F
acto
r
Fully Laden
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0.5
1
1.5
2
2.5
3
3.5
4
Cost Function
Feasible
Bump Stop Constraint Active
Optimum X*=[0.38; 0.36], F*=47.7%, Std. Dev. =3.448%
Mathematical optimisation
Multi-variable optimisation:
0 0.5 1 1.5 2 2.5 3 3.5 4
-0.5
0
0.5
Strut Force Response
Displacement [m]
Velocity [m/s]
0 0.5 1 1.5 2 2.5 3 3.5 4
4000
6000
Forc
e [N
]
Spring Force
0 0.5 1 1.5 2 2.5 3 3.5 4
-2000
0
2000
Forc
e [
N]
0 0.5 1 1.5 2 2.5 3 3.5 4
2000
3000
4000
5000
6000
7000
Forc
e [
N]
Time [s]
Symmetric Damper Scale Factor
Assymmetric Damper Scale Factor
Suspension Force - Symmetric DSF
Suspension Force - Assymmetric DSF
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1-3000
-2500
-2000
-1500
-1000
-500
0
500
1000Damper Characteristics
Velocity [m/s]
Dam
pe
r F
orc
e [N
]
Unladen
Fully Laden
Heavy Load
Extreme Load
Conclusions
� Minima in low damping, low stiffness region
� Cost function is insensitive to static gas volume -> load levelling pneumatic suspension
� Optima are constrained by bump stop constraint
� Damper characteristic is sensitive to payload change
Recommendations
� Variable damping suspension, rather than 4S4
� 4S4 should be considered for handling
� Improved tyre and friction model!!!