You are on page 1of 18

Ella Maule, s1423558 CAE3

CAD Systems And Their Use In


Automotive Part Design
Technical Report
Ella Maule, s1423558

24 th November 2016

Introduction
Computer design technologies have a wide range of uses in engineering. From aiding in the
production of 2-dimensional engineering drawings to flexible, high precision and robust 3-
dimensional design and analysis. Computer aided engineering (CAE) is used in all engineering
disciplines from original design, through lifecycle modelling and simulation, to design
development.

In this report, the use of systems, such as Solid Edge ST9, in automotive part design is
discussed. To illustrate this the design for a steering wheel and quick release mechanism will
be used.

Parametric Design
Solid Edge ST9, like many CAE modelling systems, allow the user to build 3D models from 2D
sketches. There are a number of features that ease this for the user including; extruding and
revolving from 2D sketches and adding cut outs and holes. These are in the form of integrated
work flows (figures 1-2) which take the user through the formation step-by-step.

Figure 1: Sketching tools Figure 1: 3D solids tools

Parts are built in a number of stages. First, a basic solid part using the extrude, revolve, sweep
or loft function is created. From here, material can be removed using the cut, revolve cut,
hole and extrude functions. Finally, the part can be smoothed and rounded using the round
function This allows complex parts to be built up stage by stage. Each addition/subtraction of
material is stored in the history tree.
Ella Maule, s1423558 CAE3

Figure 2: 2D sketch of the shape to be revolved to form the spindle.

Figure 3 shows a sketch to be revolved around a defined rotation axis. The initial sketch was
drawn directly on top of an existing 2D engineering drawing. This tool would allow a whole
range of parts, which currently only exist on record in 2D drawings to be converted to 3D
computer models relatively easily.

Figure 3: (a) Revolved sketch (b) with the addition of the patterned ridge extrusions.

After the sketch in figure 3 was dimensioned (parametric modelling) it was revolved to give a
basic part (figure 4a). Finally, the part had ridges patterned and extruded and corners rounded
to give the desired finish (figure 4b). A similar approach was taken for producing the boss and the
pull for the quick release.
Ella Maule, s1423558 CAE3

These tools (see figure1 & 2) reduce the design time required for modelling parts. Parts are
defined by their shape, dimensions, reference planes, face relationships, explicit
dependencies and their creation direction. This method of assigning dimensions allows the
user to make changes easily and allows part optimisation and analysis to be carried out
confidently on robust models. In parametric (history based) modelling, these changes
propagate through the design by history tree regeneration.

Figure 4: History Tree.

The constraints (figure 6) used to define the part limit the effect the propagation can have on
the overall geometry. This method of storing geometry chronologically allows parts to be
altered/ re-dimensioned fast and therefore speeds up the design time required.

Figure 5: Constraint tools. Includes: Connect,


Parallel, Equal, Horizontal/vertical, Perpendicular.

Parametric (history) modelling is particularly useful for creating part families. Groups of parts
with similar core design but with variations to basic dimensions and/or individual
combinations of features (e.g holes and chamfers). Within automotive part design this is a
particularly useful tool. Multiple variations of a part can be made to fit vehicle editions and
customised vehicles.
Ella Maule, s1423558 CAE3

For Large vehicle assemblies, families of parts for repeatable pieces can be produced. This
eases part management and reduced time spent unnecessarily designing parts. A simple part
family was made to create the pin connections for a attaching the multiple assemblies.

Figure 6: Family of Parts. Offspring table (L) for alterations to Master Part (R)

Assemblies
Very rarely are parts designed to be used in isolation but are one of many required to function
together. Solid Edge allows users to compile hundreds, even thousands, of parts in an
assembly document which represents a functioning product. The ability to share parts and
assembly documents allows internal and externally designed parts to be consolidated into
one. Pre-made/off the shelf parts, from external manufacturers, can be placed alongside
internally designed parts to be trial tested within an assembly. Multiple parts are consolidated
by setting rules and relationships (similar to those used in parametric modelling with the
addition of the function mate) between planes, axis and faces. Appropriate constraining is
particularly important when requiring an animation of the assembly. By assigning
relationships, spatial presence and material, assembly movement can be illustrated.
Ella Maule, s1423558 CAE3

Figure 7: Assembly Explosions of the Quick Release

The assembly suite allows the user to monitor the presence and visibility (active/inactive) of
the parts in the assembly. This is particularly helpful when design in-situ is required. This
method of part design allows the user to directly reference the position of existing
components. This is particularly useful when engine modifications are desired. By limiting the
available space according to the existing geometry of the assembly ad-ons and modifications
can be designed quickly and easily.

Large assemblies can be broken down into sub-assemblies, this was used for the final quick
release and steering wheel assembly and can be seen in figure 8.

Figure 8: Quick Release and Steering wheel assembly tree build from multiple sub assemblies.
Ella Maule, s1423558 CAE3

The use of sub-assemblies allows the interaction of fewer parts to be analysed specifically,
requiring less computing power. This is particularly useful when analysing the strain on the
drive shaft due to the applied motor torque. The effect on the entire car may not be important
and therefore isolating the important components allows faster, more specific data to be
gathered. This kind of analysis is done by the method of Finite Element (FE) analysis. Kinematic
Analysis is also possible within assembly, this is discussed in Finite Element Analysis and
Optimisation.

In a similar way to part families, assembly families can be produced. This allows product
variations to be kept track of and altered through the master document if changes are
necessary. It also allows multiple parts to be tested within an assembly to choose the most
appropriate.

Drafting
SolidEdge includes an engineering drawing/ drafting tool. Drawing style options make it easier
to create drawings that automatically conform to your(company) standards. E.g 1 st or 3rd
angle projections, the presence of an auxiliary view or not, whether dimensions are labelled
in a chained or stacked formation, page size, cut out views and details.

Finite Element Analysis and Optimisation


Finite Element Analysis (FEA) is the applied use of the finite element method to
mathematically model and solve complex structural, fluid and energy problems. Within
SolidEdge, FEA can be used to solve linear and non-linear statics, dynamic responses, buckling,
heat transfer and thermal stress. Understanding this behaviour is vital within the automotive
part design sector.

FAE works by dividing the part into a finite number of tetrahedral elements within a mesh.
SolidEdge has a tetrahedral shaped mesh as default however FEA can be carried out using a
number of 2D and 3D element shapes depending on the problem.

FAE works by dividing the part into a finite number of elements within a mesh; 1D beam
elements, 2D shell elements or 3D solid elements. The choice of mesh and density of elements
depends on the property being explored or accuracy required. FAE requires large computing
power, particularly where high node density is required. It is therefore essential that the user
can recognise where, within the part, the process can be simplified. e.g. simplifying the part
geometry, increasing the node density in areas of particular interest only, while ensuring the
density is appropriately low in other areas. The results of FEA analysis can be presented in a
number of ways and can be programmed to default to the company standards. Graphical
plots, static images and animations can all be produced. This is discussed in Rendering and
Animation.
Ella Maule, s1423558 CAE3

An FEA analysis was carried out on the spindle of the quick release. Fixing the rear face, a
torque of 30Nm was applied across the ridges (figure 9). A material choice of stainless steel
was chosen and s mesh size of 1.44mm was applied before it was solved using the finite
element analysis for Von Mitel Stress. The results, including an exaggerated (10% of part)
deformation, are shown in figure 10.

Figure 10: Applied torque of 30Nm to spindle.

Figure 9: Results of the FEA on the quick release spindle.

The results of the FEA show that the maximum Von Mitel Stress in the part (with applied
torque of 30Nm) is 94.8 MPa, a third of the yield stress of stainless steel. This would give a
good safety factor of 3.
Ella Maule, s1423558 CAE3

SE also includes the tools to carry out part optimisation studies. Variable optimization is
where a particular variable value is found that optimises a property subject to certain
condition. For example, a part dimension might be found that reduces its mass but ensures it
can still withstand a required strain before fracture. More advanced optimisation software
can produce genetic algorithm and simulated annealing global optimisation approximations.

The use of FEA and optimisation software (particularly within a CAD system) has a number of
benefits to the automotive part design process. The ability to predict part performance under
expected condition allows improvements to be made quickly and easily and allows the design
engineer to evaluate different geometries and materials of the part by comparing behaviour.
With this software integrated into the CAD system it limits the need for expensive prototype
production and testing.

FEA has its limitations. This means that obtaining valid results requires an extensive
knowledge of the assumptions made within the analysis and the ability to make robust models
with valid simplifications. For example, within the SE environment, parts are created which
include internal and external right angle corners which, in reality, do not exist. These tight
corner right angles (particularly internal corners) often form stress singularities which result
in (divergent) limitless increases in stress as the element size decreases (Grieve, 2006). This
error can be reduced by modelling the part out of a material which has a level of plastic
deformation. If stresses are not important (i.e when calculating modal frequencies) these
sharp corners have no effect and therefore the simplification will help the FEA. Another
limitation of FEA is its assumption that the part material is homogenous, which when working
close to maximum stress/strain can result in false positive confirmations of part validity. To
ensure these limitations dont increase the error in the analysis to the point where it becomes
invalid, steps can be taken to confirm the analysis. By plotting results against decreasing
element size, the data should converge. Any divergence suggests the model is not robust
enough to withstand the FEA (Siemens, 2016).
Ella Maule, s1423558 CAE3

A graph, figure 11, of maximum stress over a range of mesh sizes was plotted for the analysis
of the spindle (shown in figures 9,10). This was to test for divergence of the analysis. Mesh
size ranged from 0.41mm to 1.84mm. Ideally a larger range of mesh sizes should be tested
for however the model of the spindle used resulted in an unsolvable FEA analysis at mesh
sizes out of this range. The results of the graph suggest the stress value is diverging which
implies there is stress singularity within the geometry. This suggests the model geometry is
too complicated and required some valid simplification before a confident result can be
drawn from the analysis.

FEA Analysis - Maximum Vin Mises Stress Over A


Range Of Mesh Sizes
140
120
Max. VM Stress (MPa)

100
80
60
40
20
0
0 0.5 1 1.5 2
Mesh size (mm)

Figure 11: Graph of Maximum Von Mises Stress at different mesh sizes

Curves and Surfaces


The production of cures and surfaces in a 3D CAD environment is required to create aesthetic,
ergonomic designs which is particularly important for automotive parts. Curves ad surfaces,
however are considerably more complex to model than straight lines as they are described
by a set of high order polynomial parametric equations:

() = + + 2 + 3 + + + [1]
Ella Maule, s1423558 CAE3

High order curves are calculated using a set of design points either by the method of spline
interpolation or approximation. Spline by interpolation forces the curve through all of the
user defined control points whereas approximation splines give each control point a pull value
changing the shape of the curve. The curve doesnt go through any of the control points but
is instead influenced by their pull value (Admin, 2012).

Interpolation Splines Approximation Splines

Control Point End Point

Figure 12: Spline Methods.

There are two approaches to spline interpolation; Lagrangian and Hemite. Both fit a curve of
m degrees through m+1 control points, however, the Hermite approach requires the control
points to also describe the tangent of the curve at that particular point. This extra constraint
to the curve gives the user more local control over the shape.

Lagrangian Approach Hermite Approach

Figure 13: Interpolated Splines - Lagrangian and Hermite approach.

There are also multiple approaches to spline approximation; Bezier, B-Spline, rational splines
and Non-Uniform Rational B-Spline (NURBS). Bezier curves interpolate the end points but
approximate all intermediate points according to the weighting each control point has. Bezier
curves are most simply described by an equation using a blending function which blends the
effect of the control points on each section of the curve (Mill, 2016):


() = , () where , = ( ) (1 ) [2]

=0
Ella Maule, s1423558 CAE3

It is possible create high order Bezier curves to deal with complex shapes however this can be
computationally expensive. To account for this, Bezier curves are often constructed of a
number of lower order curves. The continuity (geometric and parametric) of the resultant
curve is maintained by ensuring the last/first two control points (one is shared in space) of
each segment are evenly spaced along a straight line. B-splines are a special case of Bezier
curves. They offer more control and flexibility but in return require more computation. This
is done by overlapping curve segments (knots) and instead of interpolating the end points
(Andersson, 2003) they are shared. The more shared points the higher the degree of continuity.

B-Splines are defined in a similar way to Bezier curves however their blending functions are
different:

() = , () where [4]
=0

+ [5]
, () = ,1 () + ()
+1 + +1 +1,1

+1 ,1 () = 1 ,1 () = 0 [6]

Uniform Rational B-splines are a special case of B-splines. They are defined by the same
equation (eqn 4) as B-splines except it is presented using a homogeneous coordinate system.
This has the added benefit of now being able to model conics, which standard b-splines
cannot.

The mathematics behind a Non-Uniform Rational B-Spline is a combination of all of these. It


doesnt integrate any points but instead uses knots and knot multiplicity to encourage the
curve through the point. Non-uniform knot vectors allow any spacing of the knots, including
multiple knots (adjacent knots with the same value) (Dodgson, 2000). NURBS offer the highest
degree of control and flexibility when it comes to plotting curves however they are
considerably more expensive and require large computational power.

Surfaces are built up by mapping each point of one spline over a second through interpolation
or approximation. The simplest way of constructing a Bezier surface is the tensor product
which produces a Bezier Patch. The tensor product produces a surface of the order d+1,
defined by (d+1)2 control points. This is given by (Dodgson, 2000):


(, ) = ( ) (1 )1 ( ) (1 ) , [7]

=0 =0

Continuity between patches is only guaranteed if each of the curves across the join obey the curve
continuity conditions.
Ella Maule, s1423558 CAE3

B-Spline patches are made in a similar way as the Bezier patches. A tensor product of the
equation is taken to give (Dodgson, 2000):
+1 +1

(, ) = , , (), (), +1 , +1 [8]


=1 =1

In SolidEdge, predominantly, the Bezier approximation spline method is used for mapping
curves and surfaces, however, the mathematics is hidden behind workflows leaving a user
accessible curve/ surface forming tools. These workflows include Sweep, Loft, Round/Blend
and BlueSurface.

For Example, during the production of the steering wheel the tools sweep, loft and round
where used. The sweep function was used to create the basic handle:

Figure 14: Sweep function to produce steering wheel handle.

The loft function was used to connect the handle to the body:

Figure 15: Loft function to connect handle.


Ella Maule, s1423558 CAE3

And Finally the rounding function was used to smooth the geometry:

Figure 16: Rounding function.

There are multiple roots to creating a completely smooth path, some of which make
smoothing in place down the line more difficult or even impossible. The method that was
approached initially involved sharp cornered lofts being introduced and curves being added
from there. This, however, resulted in a highly faceted, uneven surface where some corners
could not be smoothed or became undefined. The second approach to smoothing the part,
where the lofting joins were chamfered, resulted in a final part that was significantly
smoother and took fewer smoothing functions.

To better visualise any curved surfaces, the Zebra Stripe function can be used. Here, new faces
are illustrated by an offset in the stipe pattern at the boundary. From figure 17 it can clearly
be seen that the first method resulted in lower quality curves and surfaces compared to the
second. This discrepancy is why experience and practice is particularly important when it
comes to creating robust, marketable models of parts.

First Approach Second Approach

Figure 17: Zebra stripe analysis of first approach (L) and second approach (R).
Ella Maule, s1423558 CAE3

Where more advanced surface modelling is required, systems such as Rhino (Rhino, N.d) or
Alias cam be used. Similar modelling software (Parasolids) means these externally produced
surfaces can be exported for use in SolidEdge.

Direct Modelling
History based modelling does have its limitations. Up to 100 elements it is a fast and effective
way of managing a part. However much more than 100 elements, results in significantly
slower regenerations. It can also be difficult for 3rd party users to make changes when
unknown rules are limiting the change. Finally, because history modelling stores relationships
in the order they are create, it is important to create relationships in an order that doesnt
limit future rules. The alternative, which provides solution to some of these problems, is the
recent introduction of synchronous (direct) modelling. Instead of storing rules as history,
changes are made directly to the part and are not bound by rigid predefined model definitions.
The pros and cons to each often limit their use exclusively. Advanced users of CAD systems
(such as Solid Edge ST9) integrate both synchronous and history modelling when designing
complex parts.

Standard Parts
The term standard parts refers to any premade part accessible to the user. However, these
parts can be defined globally (e.g ISO 4017 M8 bolt or can be manufacturer or company
specific. Access to libraries of standard parts is very useful to part designers as it reduces the
design time required for part and assembly production. CAD programmes, including SolidEdge,
often contain built in standard part wizards. Parts include; fasteners, bearings, bolts and pipe
fittings. These are usually recognised industry standards stored parametrically to allow
dimensions to be changed.

Figure 18: Examples of a few of the


engineering standard reference parts in
SE.
Ella Maule, s1423558 CAE3

Standard parts can also be sourced from specialist exchange agents, for example
traceparts.com. These companies specialise in part model publishing and management. Many
specialists store hundreds of millions of parts from millions of part vendors
(www.traceparts.com) over a wide range of disciplines and specialties. Lastly standard parts
can be found on open source exchange forums, such as GrabCad, where independent
designers can publish their 3D models for other users to access and include I their own work.
The model of the formula student steering wheel contains an open sourced microcontroller
board and screen from GrabCad (Troy, 2015).
User designed
Open Sourced Standard part

Steering Wheel
Assembly

Figure 19: The use of an open sourced part, a standard part and user made part to produce a steering wheel assembly.

It is important to note that open sourced parts are unmonitored, the reliability of each part
should therefore be questioned and steps should be taken to validate its robustness and
suitability. Integrated wizard and specialised agent sourced parts are highly monitored and
can therefore be used confidently.

The use of standard parts can drastically reduce the time required to model a product. It
allows the user to focus more on the creative design and less on unnecessary time consuming
modelling. As industry standards for file sharing are slowly being introduces, most recently
ISO 14306:2012 JT, file sharing and part sourcing is becoming easier.
Ella Maule, s1423558 CAE3

Part Management
It is important, however, to have a well understood management system for part storage and
internally used standard parts. This is because any changes made to standard or reused parts
propagates through all assemblies it is active in. To prevent this, and other, mistakes
happening a thorough part management system is essential.

Within part management systems, particularly in larger companies, it is essential that;


everyone can work from the same data, therefore centralisation is important; part revision
history is retained; file locking is used to allow people to work simultaneously safely and
previous iterations of designs are accessible for referral or reinstatement. The chosen
management system must be optimised to ensure file control, accessibility and sharing can
be possible without mistakes causing file change errors.

There are multiple approaches to managing standard parts. Self-management is the most
basic approach. This usually involves tracking parametric models through part/ assembly
families or by classifying them (within CAD) as adjustable through a complex file and folder
structure. When a couple of people are working with a couple hundred parts and up to a
hundred assemblies and iterations this can be a cheap way of keeping track of parts and
projects. However, when numbers get any greater than that, this system leaves a lot of room
for error; file changes break assemblies, multiple changes by different users overwrite each
other and backups delete important history. Although self-management can be cheap to set
up, errors can be costly and often require significant time input.

The solution to this is the use of a product data management (PDM) system. There are a
number on the market (Synergis software, Onshape) which can be tailored to the needs of
the company and the CAD software used. Recently, PDM software has transitioned to a wholly
cloud based system.

Siemens has developed a SolidEdge specific design management system- - SolidEdge SP.
Based on a world leading collaborative platform it provides a scalable solution for company
specific needs and because it was developed for SolidEdge specifically they can both be
integrated into company systems easily.

Figure 20: Basic Part Management tools within SE.


Ella Maule, s1423558 CAE3

Rendering and Animation


Benefit to creating digital 3D part models and assemblies is that illustrations and animations
can be produced relatively easily. From engineering drawings (discussed in DRAFTS) to
photorealistic views, CAD systems often include a number of imaging tools.

Product analysis tools include FE analysis (discussed earlier) and Kinematic analysis, where
the mechanism of an assembly can be checked to ensure parts do not collide (interference
tool). Kinematic analysis requires assemblies to have been carefully constructed to ensure
motion constraints (& degrees of freedom) can be inferred directly. Animations are created
using by applying equations (differential and algebraic) to the parts motion. These equations
ae solved for each step of the motion. SolidEdge (like other CAD software) included a number
of integrated solvers which allow the user to balance speed of solving to accuracy.

FE Analysis results can also be animated, for example deflection, vibrations and heat flows.
To make deflections clearer, animations often amplify the result by some factor related to the
part size. An animation of the loading in figures 9 and 10 can be found in the CAE3f drive.

SolidEdge also included the tools to produce illustrations and animations of rendered parts
and assemblies. Exploded views of assemblies (useful for assembly instructions), animations
of FE analysis and fly-arounds are easily produced in SolidEdge. The rendering package
included in SE is relatively basic, however, the solid modeller in SE (Parasolids) allows the
geometry to be exported to 3rd party specialist rendering software such as Keyshot. Keyshot
is an embedded client of SolidEdge and is included in many SolidEdge packages. Here, more
advanced rendering can be applied and photorealistic images produced.

Figure 21: Steering Wheel rendering.

These tools are particularly useful for low budget marketing and for presenting design
concepts to those with less technical knowledge.

Conclusion
Although SolidEdge has its limitations, the benefits are present throughout the product
lifecyle. They give manufacturing technicians more convenient access to part designs so input
Ella Maule, s1423558 CAE3

to the manufacturability of parts can be given, they reduce the time spent and cost producing
prototypes as ParaSolids data can easily be converted for CNC manufacture or 3D rapid
prototyping and the production interactive assemblies and animations allows design intent
of parts to be better understood to those less technically informed.

If used by a well-trained, confident engineers, the business-wide benefits, particularly for


automotive part design and manufacturing, make SolidEdge a valuable tool.

Glossary of terms
In-situ in place

Splines (mathematical) - A spline is a numeric function that is piecewise-defined by polynomial


functions.

Simulated annealing A method for solving unconstrained and bound-constrained optimisation


problems. The approach models the process of heating material and cooling slowly to decrease
defects and system energy (mathworks).

Genetic algorithm - A method for solving unconstrained and constrained optimisation problems. The
approach models the process of natural selection and evolution by repeatedly modifying a
population of individual solutions until an optimum (maxima) solution is found (mathworks).

References
Admin. (2012, October 8). Interpolation and approximation spline. Retrieved from More Process:
http://www.moreprocess.com/computer-graphics/interpolation-and-approximation-spline

Andersson, F. (2003). Bezier and B-spline Technology.

Dodgson, N. (2000, September 25). B-Splines. Retrieved from Some Mathematical Elements of
Graphics:
https://www.cl.cam.ac.uk/teaching/2000/AGraphHCI/SMEG/node4.html#SECTION00041500
000000000000

Grieve, D. J. (2006, January 20). Errors Arising in FEA. Retrieved from Tech.Plym:
http://www.tech.plym.ac.uk/sme/mech335/feaerrors.htm

Mill, F. (2016). Curves. Edinburgh.

Rhino. (N.d). Advanced Surface Modeling for Solid Edge. Retrieved from NURBS modeling for
Windows: http://www.ndar.com/software-
cd/softwares/mcneel/rhino/texte/bfts/rhino_edge_a4.pdf

Siemens. (2016). FEA / Finite Element Analysis. Retrieved from PLM Software: FEA / Finite Element
Analysis

Troy. (2015, March 23). KS0108 128 x 64 LCD. Retrieved from GrabCad:
https://grabcad.com/library/ks0108-128-x-64-lcd-1

You might also like