Optimization and Robust Design with modeFRONTIER - Ukintpress ...
Optimization and Robust Design with modeFRONTIER - Ukintpress ...
Optimization and Robust Design with modeFRONTIER - Ukintpress ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Optimization</strong> <strong>and</strong> <strong>Robust</strong> <strong>Design</strong> <strong>with</strong> <strong>modeFRONTIER</strong><br />
Application in automotive Industry<br />
FIAT GROUP AUTOMOBILES – CHASSIS & VEHICLE DYNAMICS<br />
Marco Spinelli<br />
Chassis & Vehicle Dynamics<br />
Virtual Analysis Manager<br />
Francesco Linares<br />
EnginSoft GmbH<br />
Technical Manager SW Solutions<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis
Introduction : Scenario & Targets<br />
<br />
Examples of DOE, ROBUST DESIGN <strong>and</strong><br />
OPTIMIZATION in Chassis & Vehicle Dynamics<br />
<br />
Summary & Conclusion<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
2
Scenario<br />
Automotive industry competition<br />
Fiat Group Automobiles is a reference for product development time<br />
Analisi<br />
scenari<br />
one<br />
Impostazi<br />
Sviluppo tecnico<br />
e tecnologico<br />
Verifica<br />
di<br />
processo<br />
Preserie<br />
Salita<br />
a<br />
produttiv<br />
2001 STILO<br />
briefing Scheda “0”<br />
Delibera<br />
job 1<br />
contrattuale<br />
tecnica<br />
-22<br />
-12,5<br />
0<br />
Lancio<br />
commerciale<br />
2006 BRAVO<br />
2007 500<br />
-18<br />
-6.5<br />
2008 Delta -16<br />
-4<br />
2008 MiTo -16<br />
-4<br />
+ Quality<br />
+ Innovation<br />
Key success factor :<br />
Virtual Analysis<br />
- Costs<br />
- Time to market<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
3
Virtual Analysis in chassis & vehicle dynamics<br />
Vehicle : H<strong>and</strong>ling & ride comfort performances must be take in<br />
account at the same time <strong>and</strong> reach the Vehicle Tecnical<br />
Specification coming from targetsetting of Customer Car Profile<br />
<br />
Subsystems : Performances contribution at Subsystems level<br />
must be also take in account (PTmounts(<br />
& Suspension System)<br />
<br />
To improve QUALITY <strong>and</strong> reduce TIME a better approach could<br />
be used from the beginning in virtual analysis of vehicle<br />
dynamics :<br />
DOE – OPTIMIZATION – ROBUST DESIGN<br />
<br />
Very high potential of that approach for h<strong>and</strong>ling & ride comfort<br />
analisys based on integration of :<br />
MULTIBODY SIMULATION + OPTIMIZATION TOOLS<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
4
Multibody : FIAT Customization of MSC.ADAMS/Car<br />
Car<br />
Multi-body<br />
Models<br />
Assembling vehicle<br />
subsystems<br />
Analysis:<br />
• K&C SUSPENSION<br />
• HANDLING (+driver controls)<br />
• RIDE COMFORT<br />
• FATIGUE TEST<br />
• DRIVEABILITY<br />
Post-processing:<br />
• Automatic plotting of graphs<br />
• Calculation of single maneuver<br />
synthesis parameters<br />
Assemblies/ Testrig<br />
Full-vehicle H<strong>and</strong>ling<br />
Full-vehicle Comfort<br />
Suspension K&C<br />
Suspension NVH<br />
Suspension fatigue<br />
Engine<br />
…<br />
Subsystems<br />
Front Suspension<br />
Rear Suspension<br />
Steering system<br />
Brakes system<br />
Conceptual Driveline<br />
Anti-roll bar<br />
Engine<br />
Tires<br />
Body<br />
• Calculation of H<strong>and</strong>ling/Comfort/<br />
Steering/Braking Subjective Quality<br />
Indexes<br />
Suspension analysis<br />
H<strong>and</strong>ling maneuvers<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
5
Multiobjective <strong>Optimization</strong> tool : ModeFrontier<br />
Input variables:<br />
Spring Stiffness <strong>and</strong> preload<br />
Bumpstop clearance <strong>and</strong> characteristics<br />
Anti-roll bar diameter<br />
Damper characteristics<br />
Bushing characteristics<br />
Parameter values Solver <strong>Design</strong> performance<br />
StdA<br />
P.C.<br />
• Process Integration: integrates any CAE software<br />
• <strong>Design</strong> Automation: automates design process<br />
• <strong>Design</strong> <strong>Optimization</strong>: algorithms drive design to optimum<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
6
Multiobjective optimization & robust design :<br />
Adams/Car<br />
linked <strong>with</strong> modeFrontier<br />
<strong>modeFRONTIER</strong><br />
MSC.ADAMS/Car<br />
Model modifications<br />
Input variables<br />
Results simulation<br />
Output <strong>and</strong> comparison<br />
objectives <strong>and</strong> constraints<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
7
REAR_MV<br />
muletto validato CRF<br />
ISO SMORZANTE 223<br />
Minimo realizzabile per 263 con<br />
ammortizzatore attuale<br />
MT104<br />
4000<br />
3500<br />
3000<br />
2500<br />
2000<br />
1500<br />
1000<br />
500<br />
0<br />
-3000 -2500 -2000 -1500 -1000 -500 0 500 1000 1500 2000 2500 3000<br />
-500<br />
-1000<br />
-1500<br />
-2000<br />
velocità [mm/s^2]<br />
Input Variables, Objectives <strong>and</strong> Constraints<br />
Input variables:<br />
Objectives:<br />
Constraints:<br />
Suspension Vertical Stiffness <strong>and</strong><br />
Damping<br />
(spring, shock absorber)<br />
Suspension Longitudinal<br />
Stiffness <strong>and</strong> Damping (bushings,<br />
shock absorber)<br />
Suspension Rolling stiffness<br />
(spring, anti-roll-bar)<br />
Suspension Elasto-Kinematic<br />
Input variables:<br />
Spring Stiffness <strong>and</strong> preload<br />
Bumpstop clearance <strong>and</strong> characteristics<br />
Anti-roll bar diameter<br />
Damper characteristics<br />
Bushing characteristics<br />
Key synthesis H<strong>and</strong>ling<br />
parameters<br />
(understeer, sideslip curve, yaw, rolling -<br />
gains, time delays)<br />
AY/DVOL - gain (g/deg)<br />
Key synthesis Comfort<br />
parameters (peak accelerations, time<br />
dissipations, RMS/RMF)<br />
0.0180<br />
0.0160<br />
0.0140<br />
0.0120<br />
0.0100<br />
0.0080<br />
0.0060<br />
0.0040<br />
0.0020<br />
0.0000<br />
frequency response - lateral acceleration / steering angle<br />
80 km/h 0.45g<br />
263 Muletto c alc olo pneumMule (195 "Stilo")<br />
223 s perim pneumMule (195 "Stilo")<br />
263 target<br />
263 7q rev 4 pneumMule (195 "Stilo")<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
0.0100<br />
frequency (Hz)<br />
BETA/DVOL - gain (deg/deg)<br />
0.0700<br />
0.0600<br />
0.0500<br />
0.0400<br />
0.0300<br />
0.0200<br />
0.500<br />
0.450<br />
0.300<br />
0.250<br />
0.200<br />
0.150<br />
0.100<br />
0.050<br />
0.000<br />
frequency response - yaw rate / steering angle<br />
80 km/h 0.45g<br />
0.400<br />
frequency response - side slip angle / steering angle<br />
0.350 100 km/h - 0.4 g<br />
PSIP/DVOL - gain (1/s)<br />
263 Muletto c alc olo pneumMule (195 "Stilo")<br />
223 s perim pneumMule (195 "Stilo")<br />
263 target<br />
263 7q rev 4 pneumMule (195 "Stilo")<br />
Option 1<br />
Option 2<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
frequency (Hz)<br />
Ride heights in various load conditions<br />
Feasibility of the components for<br />
.ex. rate between axial <strong>and</strong> radial bushing<br />
stiffness, damper characteristics, bumpstop<br />
length <strong>and</strong> characteristics etc.<br />
Performance constraints<br />
BETA/DVOL - gain (deg/deg)<br />
0.0600<br />
0.0500<br />
StdA<br />
0.0400<br />
0.0300<br />
0.0200<br />
0.0100<br />
0.0000<br />
frequency response - side slip angle / steering angle<br />
P.C.<br />
80 km/h 0.45g<br />
263 Muletto c alc olo pneumMule 101 (195 "Stilo")<br />
223 s perim pneumMule (195 "Stilo")<br />
100<br />
263 target<br />
99<br />
263 7q rev 4 pneumMule (195 "Stilo")<br />
Efficienza ammortizzatore [%]<br />
Tarature ammortizzatore posteriore 263<br />
confronto con tarature X250<br />
98<br />
97<br />
96<br />
95<br />
94<br />
93<br />
92<br />
0 200 4.5 400 600 800 1000 1200<br />
0 0.5 1 1.5 2 2.5 3 3.5 4<br />
frequency (Hz)<br />
Rigidezza verticale [N]<br />
0.2<br />
0.15<br />
0.0000<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
frequency (Hz)<br />
0<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
-0.02<br />
Forza [N]<br />
-0.05<br />
0<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
AY/DVOL - phase (s)<br />
0.1<br />
0.05<br />
0<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
-0.05<br />
-0.1<br />
-0.15<br />
PSIP/DVOL - phase (s)<br />
-0.04<br />
-0.06<br />
-0.08<br />
-0.1<br />
BETA/DVOL - phase (s)<br />
-0.1<br />
-0.15<br />
-0.2<br />
-0.25<br />
-0.3<br />
-0.2<br />
-0.12<br />
-0.35<br />
-0.25<br />
-0.14<br />
-0.4<br />
frequency (Hz)<br />
frequency (Hz)<br />
frequency (Hz)<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
8
DOE Study <strong>and</strong> <strong>Optimization</strong> Method<br />
Preliminary Study :<br />
DoE Study (Sobol/Reduced-factorial distribution) =><br />
Selection of principal variables / constraints / objectives<br />
Main Study :<br />
DoE Study or DoE Study + <strong>Optimization</strong> of a<br />
limited number of input variables & objectives<br />
using detailed model or response surfaces<br />
Pareto FRONTIER => Selection of “optimum”<br />
solutions of vehicle target setting<br />
<strong>Robust</strong>ness evaluation of Pareto optimal<br />
solution<br />
Results of DOE study:<br />
Main Goal DOE / <strong>Optimization</strong>:<br />
Linear <strong>and</strong> non-linear<br />
correlations<br />
Main Influences<br />
(Inputs <strong>and</strong> Outputs)<br />
Target feasibility & optimization<br />
Target feasibility<br />
Sensitivity <strong>and</strong> correlation<br />
direct/inverse of design<br />
variables vs targets &<br />
between each design variables<br />
Principal Heavy costraints<br />
(boundary condition)<br />
<strong>Optimization</strong> of targets<br />
Trade-off scenarios<br />
<strong>Robust</strong>ness of solutions<br />
-16% -12% -17% -14% -30%<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
9
Introduction<br />
<br />
Examples of DOE, ROBUST DESIGN <strong>and</strong><br />
OPTIMIZATION in Chassis & Vehicle Dynamics<br />
<br />
Summary & Conclusion<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
10
Applications – some examples<br />
<br />
H<strong>and</strong>ling & Ride Comfort<br />
<br />
<br />
K&C Suspension<br />
PT Mounts System<br />
<br />
<br />
Suspension NVH - H<strong>and</strong>ling<br />
Active Systems (ARB)<br />
<br />
<br />
Focus on Timing & optimization results<br />
Method comparison<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
11
0.0250<br />
0.0200<br />
0.0150<br />
0.0100<br />
0.0050<br />
0.0000<br />
0.1<br />
0.05<br />
0<br />
-0.05<br />
-0.1<br />
-0.15<br />
-0.2<br />
-0.25<br />
-0.3<br />
-0.35<br />
-0.4<br />
frequency response - lateral acceleration / steering angle<br />
100 km/h - 0.5g<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
frequency (Hz )<br />
Sol 125<br />
BASE Progetto 263<br />
Sol 105<br />
Sol 162 (90)<br />
Sol 162<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
frequency (Hz)<br />
0.600<br />
0.500<br />
0.400<br />
0.300<br />
0.200<br />
0.100<br />
0.000<br />
0<br />
-0.02<br />
-0.04<br />
-0.06<br />
-0.08<br />
-0.1<br />
-0.12<br />
-0.14<br />
-0.16<br />
-0.18<br />
frequency response - yaw rate / steering angle<br />
100 km/h - 0.5g<br />
Sol 125<br />
BASE Progetto 263<br />
Sol 105<br />
Sol 162 (90)<br />
Sol 162<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
frequency (Hz)<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
frequency (Hz)<br />
0.0900<br />
0.0800<br />
0.0700<br />
0.0600<br />
0.0500<br />
0.0400<br />
0.0300<br />
0.0200<br />
0.0100<br />
0.0000<br />
0<br />
-0.05<br />
-0.1<br />
-0.15<br />
-0.2<br />
-0.25<br />
-0.3<br />
-0.35<br />
-0.4<br />
-0.45<br />
frequency response - side slip angle / steering angle<br />
100 km/h - 0.5g<br />
Sol 125<br />
BASE Progetto 263<br />
Sol 105<br />
Sol 162 (90)<br />
Sol 162<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
frequency (Hz)<br />
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5<br />
frequency (Hz)<br />
H<strong>and</strong>ling&Comfort<br />
DOE & <strong>Optimization</strong> h<strong>and</strong>ling&comfort performances<br />
Input Variables Constraints Objectives<br />
- Spring stiffness <strong>and</strong> preload - Vehicle Heighs (different load conditions) H<strong>and</strong>ling&Comfort synthesis parameters<br />
- Bumpstop clearance <strong>and</strong> stiffnesses - components feasibility - Roll,Pitch,yaw velocity,sideslip angles<br />
- Bushing stiffnesses X,Y,Z - time delays,pk/gain frequency<br />
- Shock-absorber characteristics - Peak/peak obstacle passing<br />
- ARB diameter<br />
K&C models<br />
H<strong>and</strong>ling model Comfort model<br />
AY/DVOL - gain (g/deg)<br />
Angolo d’Assetto<br />
PSIP/DVOL - gain (1/s)<br />
Frequency Response<br />
BETA/DVOL - gain (deg/deg)<br />
Velocità d’Imbardata<br />
AY/DVOL - phase (s)<br />
PSIP/DVOL - phase (s)<br />
BETA/DVOL - phase (s)<br />
Simulated manoeuvers :<br />
K&C : Elastokinematics<br />
H<strong>and</strong>ling : Sweep, Step Steer<br />
Comfort : Obstacle passing,<br />
Motorway, Pavè<br />
Results:<br />
Performance target<br />
optimization<br />
9 input variables<br />
6 objectives<br />
12 constraints<br />
7 maneuvers<br />
350 designs:<br />
70 Sobol x 5 MOGA-II iterations<br />
alt. 350 Sobol<br />
3.5 days<br />
Accelerazione Long. Guida Sedile<br />
Accelerazione Vert. Guida Sedile<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
12
Ride Comfort (H<strong>and</strong>ling Constraints)<br />
DOE & optimization comfort performances satisfying h<strong>and</strong>ling targets<br />
SF<br />
Input Variables Constraints Objectives<br />
- Spring stiffness <strong>and</strong> preload - Vehicle Heighs - Seats acceleration (RMF, amplitude)<br />
- Bumpstop clearance <strong>and</strong> stiffnesses - components feasibility - Peak/peak obstacle passing<br />
- Bushing stiffnesses X,Y,Z H<strong>and</strong>ling parameters : - body motion (velocity&acceleration)<br />
- Shock-absorber characteristics - gain ay,Theta,Psip,Beta<br />
SF<br />
- time delays Psip/Ay,…<br />
Run_RMS_PHI Run_RMS_PHIP RMS_ZZP_4_12Hz HW_RMS_ZZG AZ_F AZ_R EN_RMS_ZZG EN_RMS_ZZP<br />
Improved Ride-comfort<br />
SF<br />
SF<br />
! = 25.5% ! = 19% ! = 18.8% ! = 0.06 g ! = 0.11 g ! = 16 % ! = 10 %<br />
Simulated<br />
manoeuvers :<br />
K&C : Elastokin.<br />
H<strong>and</strong>ling : Sweep,<br />
Step Steer<br />
Comfort : Ostacle,<br />
Hole,<br />
Pavè track<br />
Motorway<br />
Results:<br />
Max performance allowed under<br />
constraints<br />
0.05<br />
Driver characteristics<br />
0.00<br />
Gain [1/s]<br />
0.50<br />
0.45<br />
0.40<br />
0.35<br />
0.30<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
ID0 ID65 ID112 ID318<br />
ISO-H<strong>and</strong>ling<br />
0 1 2 3 4<br />
Freq [Hz]<br />
7 input variables<br />
9 objectives<br />
9 constraints<br />
14 maneuvers<br />
300 Sobol alt.<br />
80 Sobol +<br />
MOGA-II 4<br />
iterations => 320<br />
designs<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
13
H<strong>and</strong>ling & Comfort <strong>Robust</strong> <strong>Design</strong><br />
DOE & <strong>Robust</strong> design of h<strong>and</strong>ling & ride comfort performances<br />
Input Variables (noise variables) Constraints Objectives<br />
- Spring stiffness <strong>and</strong> preload - Vehicle Heighs (different load conditions) H<strong>and</strong>ling&Comfort synthesis parameters<br />
- Bumpstop clearance <strong>and</strong> stiffnesses - components feasibility - Roll,Pitch,yaw velocity,sideslip angles<br />
- Shock-absorber characteristics - time delays,pk/gain frequency<br />
- STD DEV Parameters - Peak/peak obstacle passing<br />
Input Variables (optimization var.)<br />
- Bushing stiffnesses X,Y,Z<br />
Approach :<br />
1. DOE + optimization to find Pareto solutions. –><strong>Robust</strong> <strong>Design</strong> of Pareto Solutions<br />
including STDEV as parameter<br />
2. DOE & optimization including STDEV for all interesting inputs from the beginning<br />
Best compromise time - accuracy :<br />
1. Large DOE <strong>with</strong> evaluation main parameters, correlation inputs-outputs,correlation<br />
between objectives<br />
2. Reduction of number of input variables & objectives using direct / inverse correlation<br />
3. New DOE + optimization to find Pareto solutions<br />
4. Evaluation of Pareto solutions including stdev – check max/min performances<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
14
H<strong>and</strong>ling & Comfort <strong>Robust</strong> <strong>Design</strong><br />
Statistic distribution<br />
input variables<br />
<strong>Optimization</strong> of MEAN<br />
<strong>and</strong> STDEV<br />
25 Sobol x 26 LHS +<br />
MOGA-II 5 iterations<br />
=> 3300 designs<br />
Full-vehicle analysis<br />
<strong>Optimization</strong> of nominal<br />
performance<br />
One of Pareto<br />
FRONTIER optimum<br />
set-ups<br />
before <strong>Robust</strong> Study<br />
Optimum set-up<br />
chosen after <strong>Robust</strong><br />
Study<br />
4 input variables<br />
6 objectives<br />
2 constraints<br />
7 maneuvers<br />
Results:<br />
<strong>Robust</strong> design evaluation of h<strong>and</strong>ling & ride<br />
comfort performances<br />
H<strong>and</strong>ling Output<br />
Nominal value<br />
50 Sobol + MOGA-II 5<br />
iterations => 250<br />
designs<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
15
K&C Suspension<br />
DOE & <strong>Optimization</strong> K&C targets new suspension layout<br />
Input Variables Constraints Objectives<br />
Suspension attachments : Layout bushing feasib. Elastokinematics parameters :<br />
- geometry X,Y,Z Durability - toe & camber gain<br />
- Bushing characteristics X,Y,Z - Wheelbase & track var. vs Fx,Fy,Fz<br />
PT10<br />
Strictly correlated output<br />
parameters<br />
K&C curve<br />
PT2<br />
PT1<br />
Camber<br />
Toe<br />
PT1<br />
Camber<br />
Toe<br />
NP<br />
Simulated manoeuvers<br />
(ride height StdA & 2P) :<br />
Parallel travel<br />
Lateral load 2W parallel<br />
Braking 1W<br />
Results:<br />
Correlation & Sensitivity input/output<br />
variables<br />
Correlation design variables vs targets<br />
<strong>Optimization</strong> of K&C targets<br />
16 input variables<br />
8 objectives<br />
5 constraints<br />
4 analysis<br />
1000 designs:<br />
200 Sobol x 5 MOGA-II iterations<br />
17 hours<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
16
K&C <strong>Robust</strong> <strong>Design</strong><br />
DOE & <strong>Robust</strong> design of suspension characteristics <strong>and</strong> tolerances analysis<br />
Input Variables Constraints Objectives<br />
- Geometric Tolerances of suspension attach. - K&C<br />
- H<strong>and</strong>ling VTS<br />
Influences from input variables<br />
on h<strong>and</strong>ling output parameter<br />
TPsipAy 0.5Hz<br />
0.127 – 0.137s<br />
KDVOL<br />
Influences from input variables<br />
on K&C output parameter<br />
68 – 75 deg/g<br />
Results:<br />
Influence of geometric tolerances on K&C <strong>and</strong> h<strong>and</strong>ling VTS<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
17
PT Mounts System<br />
<strong>Optimization</strong> of PTMounts System<br />
Input Variables Constraints Objectives<br />
- Geometry (Elastic center) - components feasibility - Force trasmission,working points<br />
- Mounts stiffnesses characteristics (X,Y,Z ) - geometry layout - RMS/RMF seat acceleration<br />
- peak acceleration<br />
Targets Force trasmission& geometry<br />
Targets Force trasmission& stifnesses<br />
K t<br />
K lin<br />
K t<br />
x4<br />
x2<br />
x1<br />
x3<br />
Ride comfort<br />
Filtering AV3<br />
Simulated manoeuvers:<br />
Static analysis (loads fatigue &<br />
misuse)<br />
Dynamic analyses (idle,WOT)<br />
Ride comfort pavè & motorway<br />
Results:<br />
Optimize engine suspension characteristics in multiple mission<br />
Trade off in performances ride comfort , idle & acoustic<br />
12 input variables<br />
9 objectives<br />
4 maneuvers<br />
500 designs:<br />
100 Sobol x 5<br />
MOGA-II iterations<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
18
Suspension NVH - H<strong>and</strong>ling<br />
DOE & <strong>Optimization</strong> forces trasmitted from suspension<br />
Input Variables Constraints Objectives<br />
- Bushing stifnesses X,Y,Z H<strong>and</strong>ling performance : Ride performance :<br />
- toe vs Fy, toe vs Mz - Peak/peak obstacle passing<br />
- camber vs Fy - Med./High frequency force trasmission<br />
1 2 3 4<br />
Mod 1 Mod 2<br />
3 Models<br />
• K&C<br />
• Comfort<br />
• State Space<br />
Matrix<br />
Results:<br />
NVH-Ride <strong>Optimization</strong> determining<br />
optimum vehicle set-ups considering<br />
H<strong>and</strong>ling constraints. Response<br />
surfaces used for FEM model<br />
analysis <strong>and</strong> final optimization.<br />
30 60 120 160<br />
8 input variables<br />
5 objectives<br />
3 constraints<br />
4 analysis/maneuvers<br />
500 designs:<br />
100 Sobol x 5 MOGA-II iterations<br />
1.5 days<br />
Mod 3 Mod 4<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
19
modeFrontier - ADAMS - Matlab co-simulation<br />
Active Systems : DoE h<strong>and</strong>ling <strong>and</strong> comfort parameters<br />
Input Variables Constraints Objectives<br />
- Spring stiffness <strong>and</strong> preload - specific h<strong>and</strong>ling parameters - H<strong>and</strong>ling & Comfort VTS<br />
- Shock-absorber characteristics<br />
- Bushing stiffnesses X,Y,Z<br />
New front active ARB<br />
Matlab/Simulink<br />
MSC.ADAMS<br />
<strong>modeFRONTIER</strong><br />
Input<br />
Variables<br />
Outputs,<br />
Objectives &<br />
Costraints<br />
Matlab/Simulink<br />
MSC.ADAMS<br />
Results:<br />
- H<strong>and</strong>ling <strong>and</strong> comfort targets,<br />
- Correlation & sensitivity input/output variables<br />
- Study of controls in all vehicle condition<br />
16 input variables<br />
12 objectives<br />
6 constraints<br />
5 analysis<br />
250 designs: 250 Sobol<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
20
DOE-<strong>Optimization</strong> Study results<br />
Simulation Time %Perf. Improv.<br />
H<strong>and</strong>ling&Comfort 15min/design, 350designs => 3.5ggr 15%<br />
H<strong>and</strong>ling / Comfort <strong>Robust</strong> <strong>Design</strong> 8min/design, 3300 designs => 20ggr 10%<br />
K&C Suspension 1min/design, 1000designs => 17h 25%<br />
PT Mounts System 4min/design, 500designs => 1.3ggr 30%<br />
Co-simulation MSC.ADAMS-Matlab<br />
Active Control System<br />
Suspension NVH<br />
70min/design, 250designs => 10ggr<br />
4min/design, 500designs => 1.5ggr<br />
+10% of feasible<br />
design<br />
+25% of feasible<br />
design<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
21
Methods Comparison – a test case<br />
H<strong>and</strong>ling <strong>and</strong> Ride-Comfort Performances<br />
Comparison between traditional & multi-objective optimization approach<br />
H<strong>and</strong>ling/Comfort simulation -<br />
test case<br />
Needs<br />
TRADITIONAL APPROACH<br />
Knowledge <strong>and</strong> experiences about<br />
vehicle dynamics & simulation<br />
MULTIOBJECTIVE OPTIMIZATION<br />
APPROACH<br />
Knowledge <strong>and</strong> experiences about<br />
vehicle dynamics & simulation . Basic<br />
knoledge of Doe, optimization<br />
techniques<br />
Method ‘Trial <strong>and</strong> error’ Multiobjective & statistical<br />
Human time dedicated<br />
(pre/post)<br />
8 days 3 days<br />
Calculation time 1 day 5 days<br />
Total time 2 weeks < 2 weeks<br />
Human activities<br />
Numbers of solutions /<br />
scenarios evaluated :<br />
Optimal performance<br />
reached<br />
80% for model modification & run ,<br />
20% postprocessing<br />
X<br />
20% for model modification & run ,<br />
80% postprocessing <strong>and</strong> scenarios<br />
analysis<br />
10X – 100X<br />
70-80% 100%<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
22
Introduction<br />
<br />
Examples of DOE, ROBUST DESIGN <strong>and</strong><br />
OPTIMIZATION in Chassis & Vehicle Dynamics<br />
<br />
Summary & Conclusion<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
23
Summary<br />
Fiat’s latest success products, were developed using intensive use of virtual analysis<br />
in only 16-18 months.<br />
New target for future projects : increment of virtual analysis to reduce cost & time<br />
development <strong>and</strong> to allow more robust experimental testing<br />
DOE, <strong>Robust</strong> design & multiobjective optimization are success key factor<br />
The use of <strong>modeFRONTIER</strong> in Chassis Virtual Analysis as a DOE or/<strong>and</strong> optimization<br />
tool enables:<br />
To Reduce time <strong>and</strong> calculation loops<br />
To Gain a deeper underst<strong>and</strong>ing of the system <strong>and</strong> the correlation between input variables<br />
<strong>and</strong> output parameters<br />
To increase possibility to explore best solutions & scenarios<br />
To evaluate robust <strong>and</strong> stable solutions<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
24
Conclusion & next steps…<br />
Advantages in extended application :<br />
REDUCTION OF DEVELOPMENT TIME<br />
VEHICLE TESTING & EXPERIMENTAL TUNING QUALITY ORIENTED<br />
APPROACH THE OPTIMUM SOLUTION ALREADY FROM THE BEGINNING<br />
TARGET FEASIBILITY & MAIN COSTRAINTS ARE KNOWN FROM EARLY PHASE<br />
CLEAR & OPTIMIZED TRADE OFF OF PERFORMANCES<br />
EXPLORE BEST SCENARIOS AND GIVE MORE CHOISES TO MANAGEMENT<br />
Don’t forget to improve…<br />
Model complexity & detail of physical systems <strong>and</strong> components<br />
Statistical approach in model correlation : numerical model & measurements statistically<br />
correlated<br />
... Words to take in mind...<br />
“DOE - OPTIMIZATION - ROBUST DESIGN”<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
25
Thanks !<br />
Thank you for your attention<br />
Stuttgart, 24.06.2010<br />
FGA - Engineering & <strong>Design</strong> – Chassis & Vehicle Dynamics – Virtual Analysis<br />
26