CAD-CAE Integration - Department of Engineering Design

paraderollΤεχνίτη Νοημοσύνη και Ρομποτική

17 Νοε 2013 (πριν από 3 χρόνια και 7 μήνες)

74 εμφανίσεις

CAD
-
CAE Integration

Yogesh Kulkarni


Software Development Manager, Autodesk, Pune, India

Speaker’s Bio
-
data


Currently, Software
Development Manager at
Autodesk, Pune,
India


Before
Autodesk
,
worked
in companies like
PTC, SDRC, UGS, with total CAD software
development experience of about 14
-
15 years.


Bachelors
and
Master’s
degree in Mechanical
Engineering.

Agenda


Introduction


CAD
-
CAE interop


Idealization/Model Simplification


Can features help?


Feature info export and issues


References/Videos



Introduction


CAD: Computer Aided Design


CAE: Computer Aided Engineering


(FEA : Finite Element Analysis)


CAD
-
>CAE happens “over the wall”.


Once CAD design is done, designer passes
model “over the wall” to CAE analyst.


After reviewing results, Analyst may suggest
design changes.


Model is sent back again “over the wall”.


This process may happen multiple times
(iterative)


Typical Workflow

Model Pre
-
Processing



Healing


Tweaking Geometry


Removing Geometric Features


Regularizing Geometry


Finding Surface Overlap


Validating Geometry


Dimension Reduction


CAD
-
CAE
Interop


Its different !!


CAD
-
CAD
interop between system aims at
preserving Geometry
-
Topology
-
Intent as
intact as possible.



Whereas
, CAD to CAE
interop:


Can
Involve
change in dimension of entities


Will
Usually Employ Differing Amounts of
Detail


Most
Often Requires
Idealization (using Engineer’s
Judgment)

The
difference


Idealization!!


CAD models are detailed mainly for manufacturing
requirement


Whereas CAE typically does not like minute details


Apart from removing small details, it may take help of
symmetry of part and even dimension reduction to reduce
complexity so as to have quicker, well
-
formed analysis
results

CAD centric or CAE centric


In the CAD
-
centric process, the design is done in
CAD system and then
in an
iterative design
process, where design changes and analysis is
repeatedly performed
, is used to improve or
refine the
design.


In the CAE
-
centric process, engineering analyses
are performed
initially to define and refine a
design concept using idealized analysis models
before establishing
the CAD product model
.


Model
Simplification
discussed in this
presentation is of CAD
-
Centric approach
.

Idealization in CAD model


To safeguard proprietary knowledge many CAD
modelers do not expose
feature information
.


Also
,
data
capture methods like laser scanning,
neutral data
translators
give
CAD models which
do not have feature information.


Idealization on such ‘dumb’ CAD models is

classified
into two types
:


Tessellated :
faceted geometry, typically
triangles,
generated
from point cloud data


Exact:
either
Boundary Representation [B
-
rep] or
Constructive Solid Geometry [CSG]

CAD model idealization approaches


Surfaces

:
low pass
filtering
, face cluster
based
simplification and
size based entity
decimation
.


Volumes

:
voxel based and
effective
volume
based techniques. It
is advantageous
for
multi
-
resolution modeling and capable of LOD
(Level of
Details) simplification
of
features


Dimension Reduction:
Medial axis, Mid
-
surface generation, Skeletal shape

Surfaces : Low Pass Filtering


Electrical signals often carry
noises which are
equivalent
to the
small features we wish to
remove.


Noises
are smoothed out by discarding the
extra
frequency terms
after representing the signal by
sinusoidal
functions


Challenge is to define Shape as function which is
combination of Sinusoidal Functions.


Each
vertex location is changed to the weighted
average of neighboring vertex locations.
Averaging has the effect of low
-
pass
filtering

Face Clusters


Clusters(groups) are regions of interest


Faces are grouped based on certain geometric
properties, say, distance or angles.


Based on Error criterion, Face groups are
further grouped by collapsing edges in
between.



Size based


Works typically on cellular topology. Shapes
have explicit volumetric representation called
cells.


Edges in cells are ranked by geometric error
introduced if they are removed. Edges are
then contracted starting with lowest cost.


Volumes below certain threshold are
suppressed.

Voxel based


Octree is generated.


Shape is divided into cubes which are further
classified into Inside (Black),Outside (White),
Border (Grey). Greys are further divided up to
a set level.


Surface is passed through Grey nodes.



Dimension Reduction


Reducing the
dimensions of
CAD models is also
beneficial
in some
application
areas.


For
example, if long slender round bar
is modeled
as a
1D beam , there is negligible
effect on the
accuracy of
the analysis but the
computational time
will reduce
dramatically.


T
echniques are:


volume
-
based, where the input is iteratively
thinned until
a
final
skeleton is derived.


Boundary based, boundaries
are extracted and skeletons
are
generated directly
from the boundary data.

1D Lines, Curves



2D Surfaces



3D Volumes

Beams, Trusses

Axisymmetric Shells


Plates & Shells



3D Solid Elasticity

Creating Engineering Analysis Models

Idealized Type

Element

Type

Medial Axis Transform (MAT)


Can be used
to obtain the skeleton of a shape,
and
provides radius
value which represents
the distance from
a medial
axis point to its
nearest boundary points,
or equivalently
, to
where the medial axis ball
touches the
boundary.


Mid
-
surface Abstraction


C
andidate
surface
pairs
are identified
on the
solid model which
represent thin
walls on
the part. A medial
-
surface is then
constructed
between
each surface pair, and
the
resulting
surfaces
are trimmed and
extended to form
a
consistent
model.

Feature based


Ready access to features helps in de
-
featuring


Process:


If there are no ready features do feature
recognition first


Identify features to remove


Once faces are removed, heal so as to fill the
volume removed








What is feature?



Features are defined as geometric and
topological patterns of interest in a
part model
and which represent high level entities useful in
part analysis
.”


Henderson, 1990


Examples:


functional feature: for example, a pivot
,


design feature
: Extrude, revolve


manufacturing feature: a turned cylinder, a milled
slot


application specific feature: these could be any
combination of topological,
geometric, metric
,
color
and texture attributes or non
-
visual features

Feature based reordering


Decomposition : Easy to
de
-
feature


Reordering
of
Features



Summary of Simplification techniques

Problems in adopting feature based
approach


Most of
CAD
-
CAE integrations
are without
design
intent i.e. features.


Features
are
typically not fully exported
by
CAD software to protect intellectual property
captured in design intent
.


But, there ARE ways in which feature
information can be transferred…

Need ready feature info


Designer intent is captured in CAD data model
mainly in form of features.
Better integration
with
downstream applications
is
possible if this
intent
gets
translated as much as possible, but
that's
not
the typical case.


To correct the loss of feature data, wherever
available, Feature
Recognition (FR
) is carried
out.


FR
is a
difficult
problem and
even with
heuristic
algorithms, its far from being useful on complex
models.

Direct Modelers


Off
-
late some Direct Modelers
like Space
-
claim, Autodesk Fusion
do provide edit
-
ability to model
even though feature tree did not
get exchanged.


These, typically, do not do true
feature
-
editing
-
update
operations but just minor
modifications/tweaking
, giving
a
feel of feature editing.


Feature Export


Neutral

File formats supporting features
: say,
STEP. Need to mold your feature info into
neutral format which may not be possible.


API

: Using programming. Limited by what has
been exposed by APIs.


Macros
: Log files sometimes do reveal feature
recipes but again to a level they have been
allowed to expose.

STEP


International Organization for Standardization
(ISO) Technical Committee
184 has
been working
to cover parameterization of models, geometrical
constraints commonly
used into product shape
modeling.


This
can transfer parametric information and
constraints in
2D sketches only.


Construction
history of the shape
configurations
generated
by extruding
or revolving 2D sketches
can only be transferred.

Application Programming Interfaces
(APIs)


Some modelers
provide APIs to
customize their
application as well
as building 3
rd

party
sofwtare

on
top.


These APIs can be
used as neutral
-
command
-
map

Macro Parametric approach


A macro
file
that
records modeling
command sequence
is exchanged.


Even if
some CAD
modelers do log
operations, they
may not be revealing
enough
to
reconstruct
model
exactly.

Dual Model approach


The primary model
defines
the construction
history. It has an associated
secondary model
of the B
-
rep type.


The
secondary model can be used in the
receiving
system to
check the validity of model
transfer.


The
model is reconstructed in
receiving
system
by transferring primary model.

Neutral Commands, XML approach


To transfer parametric information including
design history, a set of
standard commands
is
defined
and used as a neutral format.


Neutral
format can
be extensible
mark up
language (XML) technology to express a set of
standard modeling
commands extracted from a
CAD
model.


This
approach transfers
model history
, however it
can transfer design history based only on limited
set
of modeling
commands.

Problems : Persistent naming


In Feature based modeling system, feature is
defined
referring to entities (
faces, edges
, vertices) in the
explicit (B
-
rep) geometry. Each feature, however,
introduces a modification
in the model which is
regenerated when a feature is added to
the tree
.


In
many cases
the entities
that have been used to dene
a feature may be
split
-
ed

or deleted by
the feature
itself or by subsequent features. For this reason
references to entities
used during
the design process
can be re
-
evaluated in an erroneous way, resulting
in
unexpected
results.

What are others doing?


Ansys

supports bidirectional connections with the CAD environments like
CATIA V5, Siemens
NX

etc.
The associative capabilities of these geometry
interfaces allow users to change a model’s geometry without having to
reapply
loads. Not sure if full or localized re
-
meshing happens as well as
about Model Simplification.



Automatic
associative import allows to
transfer
models from a
Catia

V5
session to an
Abaqus
/CAE
session.
A
ny
features you create in
Abaqus
, such
as partitions, loads, boundary conditions, sets and surfaces, are regenerated
each time you import the modified
Catia

model into
Abaqus
.



Autodesk
Project Centaur:
Uses
Inventor parameter
table driven
optimization in cloud to come up with best solution.
Not sure if it does
optimized re
-
meshing and model simplification.


References
:



“Closing the CAD/CAE gap”


Scientific Computing


“Software offers full associativity with Design changes”


An Analyst’s View : STEP
-
Enabled CAD
-
CAE Integration


"A Survey of CAD Model Simplification Techniques for Physics
-
based
Simulation Applications",
Atul

Thakur,
Ashis

Gopal

Banerjee, and
Satyandra

K. Gupta


An approach to B
-
rep model simplification based on region
suppression


“Geometry Simplification”, Carlos
Andujar



"Medial Object Extraction
-

A State of the Art"; Yogesh Kulkarni, Dr.
Shailesh Deshpande


“Bridging the Gap Between CAD and CAE”; David A.
Selliman
, James
Ackron
,


“Feature
-
based
Techniques for
Handling Geometric Models”, Dinesh
Shikhare


Exchange of CAD Part Models Based on the Macro
-
Parametric
Approach, Choi et al.

Q & A

Yogesh.Kulkarni@autodesk.com