Upload
hadung
View
217
Download
3
Embed Size (px)
Citation preview
3
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount
Shape Design Optimization
4
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Agenda Overview
Bushing-style Elastomeric Engine Mounts
Data Matching Optimization Problem Statement
Problem Setup
Results
Summary
5
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Bushing-style Elastomeric Engine Mounts
The powerplant is the largest concentrated
mass in the vehicle and if not properly constrained will
cause vibration and shocks to be transmitted into the
vehicle.
The proper design of the rubber
mounts may be the most
effective engineering approach
to improve the ride.
Background
6
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Bushing-style Elastomeric Engine Mounts
Engine mount (bushing) is made of rubber.
This bushing is encased in a steel sleeve, attached to
the frame of the car.
The steel center cylinder is attached to the engine.
Basic Geometry
7
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Bushing-style Elastomeric Engine Mounts
Different stiffnesses desired in each principal direction. Compromise between isolation (i.e. soft rates) and gross motion control
(i.e. stiff rates) characteristics
Desired Response
8
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Bushing-style Elastomeric Engine Mounts
Analysis is achieved in following stages: Thermal shrinkage/cool-down of the molded rubber journal
Diametrical radial swage of outer metal
Displacement of inner tube to determine tangential static rates in principal
directions
Typical Analysis Objectives: Model key aspects of physical manufacturing process
Predict static rates in key principal directions
Prevent detrimental residual stresses (strains)
Predict dynamic rates
Typical Analysis Goals
9
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Loadings
Cool-down: Mount is cooled from 160 C to 20 C to
simulate mold injection and cool-down
As the mold-bonded bushing cools residual stresses due to
thermal shrinkage puts the rubber in tension, which greatly
reduces durability
10
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Loadings
Swaging: Reduction of the outer sleeve diameter
The effect of proper swage is to convert the negative-valued
pressures to positive compressive stresses. This improves
durability.
11
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Loadings
Vertical Compression: Determine Primary Direction
Static Stiffness
Displace inner cylinder in large-void direction
12
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Loadings
Vertical Tension: Determine Secondary Direction
Static Stiffness
Displace inner cylinder in small-void direction
13
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Loadings
Lateral/Radial: Determine Third In-Plane Direction
Static Stiffness
Displace inner cylinder in no-void direction
14
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
The original rubber bushing was designed with a Shore A
hardness value of 60.
The Purchasing department has acquired a rubber with a
Shore A hardness value of 70. The initial Elastic Modulus difference is ~40%!
You have been requested to re-design the rubber
bushing geometry such that the principal-direction static
bushing stiffness rates are equivalent while maintaining
the peak maximum principal strains to be less than 45%.
Problem Statement
15
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Isight Simflow
Problem Setup
16
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Problem Setup - Abaqus
Geometry – 14 independent, parameterized geometry variables
OD: 101.6 mm
ID: 25.4 mm
17
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Problem Setup - Abaqus
Mesh
Hybrid, reduced-Integration, first-order plane strain
elements (CPE4RH)
Local refinement seeds along radii of voids
18
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Problem Setup - Abaqus
Material
Mooney-Rivlin material model for a 70 Shore A rubber with an effective Poisson’s ratio of 0.4995
Solver
Abaqus/Standard run in a quasi-static fashion
Contact
Self-Contact Pairs with a Coulomb friction coefficient of 0.2
Procedure
Seven analysis stages (steps) to represent manufacturing and desired load-unloading sequences
Computational Resources
Using 4-cores on a 3.0 GHz Quad-Core 64-bit Windows machine, each Abaqus analysis completes in ~ 6 minutes.
19
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Problem Setup – Data Matching
Data Match components are used to compare Simulation and Target curves based on
Absolute area difference between simulation and experiment load-deflection curves
Minimizing the “absolute area difference” objective function brings all the simulation data points in proximity to the experimental data points
Maximum difference between the simulation and experiment curves.
Focuses on reducing the local maximum differences between the two data sets.
When used in combination, these objective functions provide a robust solution to the data matching problem.
20
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Problem Setup – Optimization
Optimization Technique
Hooke-Jeeves with a max of 100 evaluations
H-J is well suited for non-linear design spaces and does not require the calculation of gradients
100 evaluations completes in ~10 wall clock hours
Design Variables
14 geometrical variables
Constraints
Peak Max Principal Strain cannot exceed 45%
Objectives
Minimize Data Match parameters for each loading
21
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Results
A quality fit is found within 50 evaluations
22
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Does the Preload affect the Final Optimal Shape?
Data Match to Target with No Preloading
Cooldown and Swaging not included
A significant shape change is observed as compared
to preloaded optimal shape.
23
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Does the Preload affect the Final Optimal Shape?
“No Preloading” Optimal Geometry Utilized with Preloading
A stiffer-than-expected response would be found for this geometry.
24
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
A significant shape change is
observed between the analyses
A stiffer-than-expected response
would be found for the “No-Preload”
optimal geometry.
Bottom-line:
Include preload effects in
shape optimization analyses
The importance of including preloading effects
Elastomeric Engine Mount Optimization
Preload Excluded
Preload Included
25
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Abaqus Token Reduction Scheme with Isight
Abaqus jobs “launched” via Isight sim-flows use progressively fewer Abaqus
tokens than nominal stand-alone Abaqus job execution
Depends on total number of analyses, number of cores per analysis and
the number of simultaneous jobs
For typical Isight studies up to 60% fewer Abaqus tokens consumed per job
vs. nominal stand-alone execution
Implementation subject to several limitations affecting availability of token
reduction
For this 100-analysis problem, we used 4-cores with one simultaneous job.
The token usage is as follows:Job Number 1 2 .. 6 .. 16 .. 41 .. 100
Abaqus Tokens 8 6 5 4 3 3
26
© D
assa
ult
Sys
tèm
es Ι
Con
fiden
tial
Info
rmat
ion
Elastomeric Engine Mount Optimization
Summary
We have learned the following in this study:
Bushing-style elastomeric engine mounts require complex and
competing structural requirements.
Abaqus is able to analyze this class of problem that includes
nonlinear geometry, nonlinear materials and contact.
Preloading effects should be included in this class of problem such
that the optimized shape can meet the design requirements.
Isight automates the steps of a single analysis workflow and then
can systematically evaluate the design based on the input variable
ranges, objectives and constraints.
Abaqus Token Reduction Scheme with Isight can greatly reduce the
cost of Design Exploration studies.