automatic shape variations for optimization and innovation · delphi corporation, technical center...

2nd IFIP Working Conference on

Computer Aided Innovation

October 8 - 9, 2007

Delphi Corporation, Technical Center Brighton

Product Development IT - Chrysler LLC

Noel León, José Cueva, César Villarreal, Sergio Huitrón, Germán Campero

Automatic shape variations for optimization and

innovation Shape Optimization of Cylinderhead Gasket using CFD

Tecnologico de Monterrey, CIDYT-CIII

Ave. Eugenio Garza Sada # 2501

Col. Tecnológico, Monterrey, NL, CP. 64839, Mexico

http://cidyt.mty.itesm.mx


Product Development IT, Chrysler LLC

2nd IFIP Working Conference on Computer Aided Innovation

October 8-9, 2007

Automatic shape variations for optimization and innovation:

Shape Optimization of Cylinderhead Gasket using CFD

N. León, J. Cueva, C. Villarreal, S. Huitrón, G. Campero

1. Introduction

2. Splinization and Mesh Morphing

2.1. Mesh morphing

3. Case Study: Shape Optimization of CylinderHeadGasket using CFD

3.1. Optimization of the engine cooling process

3.2. Preliminary results

4. Future work

5. Conclusions

2

Agenda




October 8-9, 2007




• The available CAD-CAE tools and methods are notalways adapted to satisfy design requirements that areincreasingly more focused on potential creativityenhancement.

• This paper focused on the innovation of shape andtopology using proven and mature methods that has beenused successfully for parametric optimization, extendingits application to design automation and Computer AidedInnovation.

1. Introduction

3




October 8-9, 2007




• Commonly, performance improvement is first obtainedthrough quantitative changes in parametric design insearch for optimization.

• When the resulting improvements reach the limits ofperformance improvements, new searches must be carriedon through qualitative changes or paradigm shifts.

1. Introduction (2)

4




October 8-9, 2007




• Qualitative changes and paradigm shifts lead toinnovation.

• Optimization looks for the most advantageous state ofequilibrium before a contradictory situation is faced:

The improvement of one or several performancecharacteristics, deteriorates others.

• In order to get out of this deceptive goal, innovativeconcepts are needed by changing not only parametricvalues, but also shapes, topologies and physicalprinciples.

5

1. Introduction (3)




October 8-9, 2007




• Parametric CAD and CAE systems were originally

conceived for facilitating only parametric variations on

modeled parts.

• In recent times, topology optimization methods have been

also introduced in meshing environments.

1. Introduction (4)

6




October 8-9, 2007




• Topological optimizationfunctions are currently usedto find optimum topologiesand shapes for given partsunder prearranged conditions.

• Shapes and topologiesobtained this way are notmodels with a CAD structure,and require manual postprocessing or even acomplete redesign

1. Introduction (5)

7




October 8-9, 2007




• Topological optimization:

1. Introduction (6)




October 8-9, 2007




• The splinization concept was

presented during the 1st IFIP

Working Conference on Computer

Aided Innovation in Ulm, Germany

in November 2005 for

implementing methods that may be

integrated into conventional

CAD/CAE systems for executing

shape and topology changes that

transcend parametric values.

1. Introduction (7)

9




October 8-9, 2007




• Performance enhancements are searched with the aid of

genetic algorithms and shape and topology variations at both

CAD and mesh models that may be converted in both

directions.

• Now a new way of combining the splinization approach with

the meshing software, for integrating the shape variation

method to the CAE analysis method is presented.

• This method has special advantages as it allows shape

variations at mesh level that may be later translated into direct

modification for a previously “splinized” CAD feature.


10




October 8-9, 2007




11

• In this case mesh morphing in existing commercial finite

element meshing software was used.


• The way the mesh

surrounding the shape

is modified depends

on a bias value (so

called a “handle”) that

is equivalent to the

tension of a Bezier

curve.




October 8-9, 2007




• The use of the “morphing” became advantageous because:

– It allows shape variations to a model without remeshing it.

– The interpolation of previous results can save computational

time enormously in FEM simulations.

– Previously converged simulation results (or interpolated new

results) can be used as an initial guess and speed up the

simulation process.

– It takes advantage of converged solution

– It is good to have a mesh model that remains consistent


12




October 8-9, 2007




• The final goal is to integrate in an automatic manner.

– CAD/mesh shape optimization process through splinization

– CAE Thermal simulations

– Genetic algorithms

• The splinization approach and mesh morphing are

connected by positioning the morphing handles at the

control points that define the spline.


13




October 8-9, 2007




• Because it is sometimes desirable to modify only a certain

feature within a mesh model, manipulating node coordinates

must be done in a very controlled manner to maintain model

consistency and quality of the mesh elements.

• The best way to modify a certain shape at mesh level is to alter

the nodes defining it.

• Smoothing the node displacements allows that the perturbation

of the elements is minimized and mesh topology is preserved.

• This effect is known as mesh morphing, which is a tool

included in some meshing software.


14

2.1. Mesh morphing




October 8-9, 2007




• For each modifiable shape, edge domains are formed in the shape that

defines the mesh boundaries.

• Several stepped regions allow defining a bigger region than would be

allowed by mesh quality criteria when performing the variation in a big

region.

• Within a domain, handles allow the actual node manipulation.


15

2.1. Mesh morphing

• Even though this kind of

variations has major

limitations because of

mesh quality constraints,

it is very practical when

only small changes in

shapes need to be made.




October 8-9, 2007




• Simulations for the performance enhancement of the cooling water jacket of

cylinder engines involves:

– Geometry – Meshing tools – Computational Fluid Dynamics (CFD).

• It is a highly time consuming process and is usually performed manually

modifying the gasket holes dimensions based on the results of CFD simulations

subject to manufacturing, packaging and material constraints.

• Each of the iterations takes about 1/2 week (mesh generation and analysis).

3. Case Study:

Shape Optimization of CylinderHead Gasket using CFD

16

• Under these conditions the

possibilities of variations of

shapes and topologies are

very limited and therefore

only changes of the holes

diameters are commonly

introduced.




October 8-9, 2007




• Water jackets are designed for peak power

condition that is representative of the worst case

scenario with respect to heat rejection to

coolant with the flow rate specified at peak

pump output.

– Predominant design variables

• Number of gasket holes.

• Dimension of each gasket hole.

• Location of each gasket hole.

– Restrictions

• Gasket holes should reside inside of intersection

of block and head water jackets.

• Flow-distribution near critical areas such as

exhaust valve bridge,

• Spark plug and peak velocity zones that are prone

to metal erosion.

17

3. Case Study:





October 8-9, 2007




• The first goal is to significantly reduce the time for finishing a design, byproviding a methodology that integrates the engineering concept and thegeometry restricted to manufacturing, packaging and material constraints.

• Additionally performance increments beyond those achieved bytraditional manual methods may be expected as not only changes indiameter values but also shape and topology variations.

• Thermal analyses are performed to predict heat transfer coefficients(HTC).

• Thermal analyses provide the coolant flow rate from the water pumpperformance, heat fluxes corresponding to the peak power condition indifferent regions of the water jacket.

18

3. Case Study:





October 8-9, 2007




• Holes in a cylinder head gasket are the only available passage for cooling

fluid between the cylinder head and cylinder block.

19

3. Case Study:

Shape Optimization of CylinderHead Gasket using CFD3.1. Optimization of the engine cooling process




October 8-9, 2007




• The pattern of gasket holes defines the flow conditions

• Changes in their pattern can positively impact heat transfer coefficients

for the cylinders, thus improving engine cooling.

• In this case the objective of gasket holes optimization is to maximize heat

transfer coefficient as well as to minimize the difference among heat

transfer coefficients among cylinders.

• With the help of the mesh morphing process explained earlier, in the DOE

step re-meshing is avoided while boundary conditions are changed.

• The optimization step uses mesh morphing to achieve a new mesh in a

reasonable time.

3. Case Study:


20





October 8-9, 2007




• Overall procedure of DOE and optimization process of gasket holes.

3. Case Study:


21





October 8-9, 2007




• The main benefit is the reduction of man-

hours required for set-up and post

processing by eliminating routine tasks

that are now implemented automatically.

• The implemented procedure allows closing

this shape optimization process by feeding

the shape variations of the mesh morphing

controlled by the handles to the CAD

environment as control points of a spline.

3. Case Study:


22





October 8-9, 2007




• The investigation has been conducted in a simplified model, in which six

holes perform a scale factor, without shape and/or topology variations yet.

• The process integration for performance optimization in the software

applications has been already achieved.

3. Case Study:


23


1Inlet/Outlet

2 4

35

6

Top View




October 8-9, 2007




• The objective is to maximize the minimum heat transfer coefficient

(HTC) as well as to minimize the maximum difference of HTC

among cylinders.

• To perform the DOE analysis the objective function is defined as

follows:

• f = HTC1+HTC2+HTC3-3(|HTC1- HTC2|+|HTC1-TC3|+|HTC2-

HTC3|)

3. Case Study:


24





October 8-9, 2007




• Results of the DOE analysis

• The tendency is having bigger holes while they are farther from the inlet.

25

3. Case Study:

Shape Optimization of CylinderHead Gasket using CFD3.2. Preliminary results




October 8-9, 2007




• The results of the DOE analysis

26

3. Case Study:

Shape Optimization of CylinderHead Gasket using CFD3.2. Preliminary results




October 8-9, 2007




• Taking advantage of the automatic shape variation presented in this

paper, it is being working in applying this technique looking for

innovative shapes of Savonius tye rotors for wind generators.

4. Future work

27

• The Savonius wind turbine

– Drag type vertical axis wind turbine.

– Developed in the early 20’s of last century.

– Has been undermined because of its low efficiency

in comparison with traditional wind turbines of

horizontal axis.




October 8-9, 2007




• Research work has been done for finding new shapes of the

Savonius rotor making subtle changes to its profile shape, but the

traditional geometrical features such as arcs and lines have remain

unchanged.

4. Future work

28

• By using splines in the rotors

design it is possible to create

forms never explored before,

which may be analyzed trough the

techniques described in this paper

for the Shape Optimization of

CylinderHead Gasket using CFD.




October 8-9, 2007




4. Future work

29

Banesh (1994)

Patent 1996

Rahai (2005)

Patent App. 2007

Traditional Rotor (1922)

Patent 1924

• Known rotor designs:




October 8-9, 2007




4. Future work

30

• Possible rotor designs variations:




October 8-9, 2007




• CAI may benefits from mature optimization methods for

performing shape and topologic variations in CAD/CAE

environments leading to innovative shapes and topologies that

allow achieving higher performance when parametric optimization

faces technical contradictions.

• Using “splinization” techniques together with mesh morphing and

genetic algorithms facilitates the search for higher performance

exploring innovative shapes that avoid expensive manual trial and

search methods.

31

5. Conclusions




October 8-9, 2007




• The authors acknowledge the support received from Tecnológico de

Monterrey through Grant number CAT043 to carry out the research

reported in this paper.

• Authors acknowledge collaboration effort with Vamshi Korivi,

Surendra Gaikwad and Su Cho working from Aerothermal Center

of Competence and Product Development IT, Chrysler LLC.

32

6. Acknowledgments

automatic shape variations for optimization and innovation · delphi corporation, technical center...

Documents