Upload
trinhtuong
View
214
Download
0
Embed Size (px)
Citation preview
334
Int. J. Engg. Res. & Sci. & Tech. 2015 xxxxxxxxxxxxxxxxxxxxxxxx, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
ESTIMATING THE EFFECT OF CAD
MODEL SIMPLIFICATION TECHNIQUES FOR
FEA SIMULATIONS
Subash karan R1* and Ravichandran R1
At present, simulation is at the heart of product development cycle. The simulation is currentlyused indesign and dimensioning of the mechanical parts. We focusing on the dynamics of themicroscopic world and on realizing simulations that capture maximum complicity using minimumcomputation time and cost. FEM becomes the most useful approach in simulation of mechanicalbehaviour. For that it requires initial CAD model. It is not easy to analyse complex models. So, itis necessary to simplify the models before simulation. In this paper presents an automatic CADmodel simplification approaches. First part of this paper deals with a Rule based approach tosimplify the features for that the rules are generated with help of expert’s knowledge andsimplification has obtained in Knowledgeware a special tool available in CATIA V5. Second partpresents a Hybrid method approach to simplify CAD model. The implementation of algorithm onOpen Cascade Platform is also presented. Finally estimate the effect of using this approach onCAD model. It shows that computational time is reduced with minimal changing the exactitudeof results.
Keywords: Simulation, CAD Model, FEM, Rule Based Approach, Hybrid Approach, Simplification
*Corresponding Author: Subash karan R � [email protected]
1 Department of Mechanical Engineering, Arunai Engineering College, Thiruvannamalai, Tamilnadu, India.
Int. J. Engg. Res. & Sci. & Tech. 2015
Research Paper
INTRODUCTION
Computer simulations play a vital role in design,
research, development, prototyping and validation
of mechanical products. It indeed to reduce the
risk and inefficient in product testing. For example:
UNO III vehicle design, in that when we began
the UNO III redesign, for rider safety it was clear
that manually tuning controllers and testing on
the actual vehicle would be inefficient and risky.
Instead they used computer simulation to model
and simulate the UNO III mechanical systems.
During real-world test, things move so fast it
isimpossible to understand everything that
ishappening. In simulations however, we can
freezetime and inspect every aspect of the model
to get aclear picture of how mechanics are
behaving. Wefocus on the dynamics of the
microscopic worldand on realizing simulations
that capture maximumcomplicity using minimum
computation time andcost.
Engineering models today are mainly
developed with a 3D computer Aided Design
ISSN 2319-5991 www.ijerst.com
Special Issue, Vol. 1, No. 2, April 2015
2nd National Conference on ‘‘Recent Advances in Science,
Engineering&Technologies’’ RASET-2015
© 2015 IJERST. All Rights Reserved
335
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
Software [2]. If we run a Finite Element Analysis(FEA)on a part with hundreds of small feature asoftenleads to computational time will be very largeandpoor quality mesh and hence may lead toaffectanalysis result. Then, it automaticallyincreased thecost of labour [3].In order to getaccurate result in a timelymanner, one must utilizesimplified models. Thesimplified models are usedin many simulationdomains: dynamic analysis ofbogie for forestrymachines [4]; modeling andsimulations of dicycle[5]; Thermo mechanicalsimulation of aerospacestructures [6]; virtualmodeling and simulation ofmilitary vehicles [7];Optimization of vehiclestructure [8], etc.
RELATED WORKAtul Thakur et al. [9] had studied that theexistingmodel simplification techniques that areuseful forthe physical based simulations and it canbeclassified them broadly into four maincategoriesbased upon the type of simplificationoperatorsused in respective techniques. ie, surfaceentity,volumetric entity, explicit feature anddimensionalreduction.Hamdi mounir et al. [10] had stated thattheFinite element method (FEM) becomes mostlyusedapproach in simulation of mechanicalbehaviour.The adaptation steps consists insimplification ofCAD model geometry byeliminating details holes,chamfers, fillets basedon a combination ofeliminating details andmerging faces .H.Zhu and C.H.Menq [11] hadreviewed on BRepmodel simplification byautomatic fillet/roundsuppression. The majorshape of the primaryfeatures may not be affectfillet and rounds cangreatly change the geometricand topologicalpatterns of the primary features.
R. Ferrandes et al. [12] had proposedanevaluation method of simplification detailsforfinite element modal preparation. If a shapedetailremoved during the shape simplificationprocessproves to be influence on mechanicalbehaviour, itcan be reinserted on simplifiedmodel.Shuming Gao et al. [13] had proposed afeaturesuppression based framework to cadmodelsimplification. CAD mesh model issegmented intoregions using the improvedwatershed algorithmand form features arerecognized based on regionlevel representationusing graph based featurerecognition method.
R.J.Donaghy and C.G.Amstrong [14] usethemedial axis transform (MAT) to carry outtheadaptation and the idealization of B-Repgeometry.The MAT method builds the skeleton of
Figure 1: Simulation ofUNO III vehicle redesign
Figure 2: Virtual model and Simulationof military vehicle
336
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
ageometrical representation in order to obtainthemedial axis.
M.Hamdi et al[15] CAD/CAE interoperability,anautomatic generation of analysis model basedonthe simplification of CAD geometryisimplemented by using CASCADE platform.Inorder to be independent of CAD/CAM system,theproposed algorithm relies on a neutral file(STEP)to recuperate the data of the part to besimplified touse various tools of simulation; thesimplified cadgeometry will be also stored in theSTEP format.
Oussama jaider et al. [16] had described anewapproach to eliminate the undesirableininterpretations of features, accordingtomanufacturing rules and metal removalprincipals.Automatic feature Recognition (AFR)has played acrucial role linking CAD activities andcomputeraided process planning (CAPP).
Bob Evans et al. [17] had stated thattheinterference of machine learning as decisiontreeinduction to derive process control rulesandparticipation of experts knowledge to do thebest.They recently applied a machine learningstrategyknown as decision tree induction to derivea set ofrules.
Jiawei Han and Micheline kamber [18] looksatrule based classifiers, where the learned modelisrepresented as a set of IF-THEN rules. Theythenstudied the ways in which they can begenerated;either from decision tree or directlyfrom thetraining data using a sequential coveringalgorithm.
A.Sheffer [19] reported that, face clustering isatechnique to cluster the faces in the inputmodel.The clusters thus formed represents theregion ofintersects that may be considered forsimplification.There are three main steps followed
in thisapproach: face clustering, finding thecollapsiblefaces and simplification.
Satoshi kanai et al. [20] had described thattheCAD systems and assembly models tend tohave ahuge number of parts and very complexinnerstructures. For achieving the light weightandstrengthened parts, the inner structures ofhousingsuch as rib or bosses have verycomplexgeometries. In order to simplify thiscomplexgeometry: Remove invisible technique isused. In this paper utilizes two differentapproaches to simplify the cad model. Thefirstapproach is knowledge based. In thisapproach theknowledge is gain from expertsdemo’s andencoded that data as simplificationrule. It is alsocalled as a rule based approaches.The rules showsthat whether the feature /part thatshould beconsidered for suppression or not.
The second approach is using hybridmethodof simplification. In this approach consistsoffollowing phases. The phase A consists of stepofpre-treatment of CAD model. The phaseBconsists in identifying details for detailsforsuppression. The phase C consists ofallowssimplification and reconstruction ofsimplifiedmodel. For this approaches the opencascadeplatform is utilized.
METHODOLOGYIn this session has been divided into twomainphases. First phase consists of knowledgebasedapproach and second phase leads tosuppressionby hybrid approach. The overview ofthisapproaches are discussed below.
Overview of knowledge based approachThe proposed approach is to design anautomatedtool to identifying the features capableforsuppression. The overall approach forgenerating
337
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
a CAD simplification using rule basedapproachas shown below.
whethermodel is suppressed or not. Smalldimensions areleads to suppress.
Rule 3: Instead of absolute dimensions, therelativedimensions of features are consideredassuppressed or not.
Rule 4: Boundary and load conditionsareimportant factors that also taken to beconsiderationfor suppression or not.
Rule 5: Feature lies close to the loadapplicationcannot to be suppressed.
Rule 6: concentrate on feature formed byoffset,mirrored and used as a reference, of otherfeatures.Because it leads to deleting of overallcomponentor access denied or redefined overother features.
3.1.2 Inductive Decision TreeInductive decision trees have proven to beasuccessful tool within the numerous fieldsandthese are proven to be advantageous inimprovingthe accuracy or efficiency of the system
The above is an example of arepresentativedecision tree for suppressing a holefeatureconsidering attributes diameter, minimumdistance to the boundary condition feature, andthegeometric type. The left node is the node
Figure 3: Overview of Rule Based Approach
First, the experts knowledge is capturedfromdemo and suppression rules are generatedformstatistical induction learning techniques[21].Thegenerated CAD model is converted intoneutralformat (STEP) and then rules generatedareexpressed as a binary decision tree.
The decision to suppress or not, on afeaturedepends up on the analysis contentandapplication. Using that knowledge gained fromtheexpert and the decision rules are generatedforsuppressing the CAD model for FEAsimulations.For suppression we consideredfollowing.
3.1.1 Machine Learning: Rules forSuppressionRule 1: Based on feature type only weconcludedwhether model is suppressed or not.Feature likeholes, fillets, chamfers aresuppressed.
Rule 2: Based on feature dimension only
Figure 4: Decision treefor feature suppression
338
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
thatshould be executed if the condition is true.Theright node is the node that should be executedif thecondition is false.
Example algorithm for the above flowchart forsuppression of holes using decision treeinductionas shown below.
Algorithm for feature simplificationAlgorithm: Rule 1
if feature type ==hole
get diameter value (min)
then (DIAMETER (D) <5)
if boundary condition. Distance (d) <5
then
The feature leads to be suppressed
Else The feature could not be suppressed
End if
End if
Algorithm: Rule 2
if feature type = = hole
get diameter value (min)
then (DIAMETER (D) <5)
if Hole type = = blind
then The feature is to be suppressed
else
if Hole type = =through
then The feature could not be suppressed.
End if
End if
End if
Implementation and validation on exampleImplementation of approach:
The data processing implementation of
simplification approach was carried out onCATIAV5 software. In that the API development’splatform: “Knowledgeware”. Knowledgewareaspecial tool developed for 3D applications ofCADCAMmultiplatform. This platform is installedbyusing API toolkit installer and it is based onvisualbasic language.
Example of validationIn this section, one example of validation will allowvalidating the principal functionalities of thesimplification algorithm. (Fig 5) presents BogieSF200. The part Bogie selected because theyhave a broad variety of mechanical parts in termsof its forms, the boundary conditions and alsothe details which they contain.
In this first the STEP file format is encodedand details of the feature are obtained.
After that the 3D model is extracted fromneutral STEP file in Knowledgeware API platform.Below Fig presents the illustration of the principalstages to pass from a CAD model of the BogieSF200 (Fig 5), imported CAD model from STEPfile (Fig 6), shows that operation performed in ruleeditor (Fig 7), shows that check editorperformance (Fig 8), shows knowledgeinspector checks the rules (Fig 9), shows
Figure 5: Imported CAD modelfrom STEP file
339
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
Remove features language browser and finally(Fig 10), shows simplified model.Figure 6: Imported CAD model
from STEP file
Figure 7: Imported CAD modelfrom STEP file
Figure 8: knowledge inspectorchecks the rules
Figure 9: Remove features
Figure 10: Simplified model
Overview of hybrid approachIn order to be independent of the CADCAMsystems, the proposed algorithm relies on aneutral file (STEP) to recuperate the data of thepart to be simplified. To use various tools ofsimulation, the simplified CAD geometry will bealso stored under the same neutral format(STEP).
The overview diagram for hybrid approachisshown below. It consists of three main stagesinsimplifying of CAD model. This original
340
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
visionenables designer to visualize the leastinfluentialzones (high order of criticality) on thecomputationresults, giving him the possibilityeither to interactivelyeliminate the entities whichhave highorder of criticality, or to appeal toautomaticalgorithms of elimination.
Stage2: Identification of detailsThe stage (2) consists in identifying theIso-zonefor elimination. That implies theimplementationof algorithms of identificationbased on featurerecognition. The result of thisphase is arepresentation of Iso-zones targets forelimination.These Iso-zones are entities (edges,faces,chamfers, fillets, and rounds) coloredaccordingto a gradient of criticality.
Stage3: Remove and Reconstruction ofmodelThe stage (3) consists in removing theidentifieddetails, then in rebuilding the geometricalmodel
Figure 11: Overview of hybrid approach
Stage1: Initial CAD modelThe stage (1) of the algorithm consists of aphaseof identification of details. The structuredinformation relates to the faces, wires, theedgesand the vertex which includes the geometrymodelof the part.
Figure 12: Initial CAD model
Figure 13: Identificationof details(edges, faces, Izo-zones)
after suppression. The result of this phase isasimplified CAD model whose elementarytopologyis valid. At the exit of the algorithm, thedesignerhas at his disposal a simplified modelrecorded informat STEP for a simulation by finiteelements.
Implementation and validation on exampleImplementation of approach
The data processing implementation of
341
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
simplification approach was carried out onopenCASCADE platform. Open CASCADE aspecialtool developed for 3D applications of CAD-CAMmultiplatform. This platform available freelyinopen source on internet ant it is based onC++language encoding.
Example of validationIn this section, one example of validationwill allowvalidating the principal functionalities ofthesimplification algorithm. Fig presents BogieSF200. The part Bogie selected because they haveabroad variety of mechanical parts in terms of
itsforms, the boundary conditions and also thedetailswhich they contain.
RESULTS AND DISCUSSIONThe below figures shows that the analysedmodelsusing ANSYS 14.0.Fig (18) shows analysisoforiginal model it has an equivalent Von-misesstress of 1.8745e5 Max and analysis timetaken as38 s. Fig (19) shows analysis of simplifiedmodelusing knowledge based approach it hasanequivalent Von-mises stress of 1.4891e5 Maxandanalysis time taken as 16 s. Fig (20) showsanalysisof simplified model using knowledgebasedapproach it has an equivalent Von-misesstress of1.4236e5 Max and analysis time takenas 12 s.
Figure 14: Remove and Reconstructionof model
Figure 15: Initial CAD model(STEP FORMAT)
Figure 16: Identification of Izo-zones
Figure 17: Simplified model
342
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
CONCLUSIONThis paper presents an estimating theeffect ofsimplification algorithm on CAD modelsfor asimulation by finite element method. Thealgorithmproposed consists on reading the B- repmodelof the CAD geometry in order to identify,then toremove the details considered to besuperfluousfor mechanical analysis. The example ofvalidation show that asuitable elimination of thedetails in a CAD modelallows saving a veryimportant time (up to 69%) inthe procedure ofsimulation while keeping a highquality of thecomputation results. Theseobservations werefound by doing simulations byfinite elementsbefore and after simplification. Thiswork helps tothe industrialists who are interestedmore and
Figure 18: Analysis of original model
Figure 19: Analysis of simplified model(Knowledge based approach)
Figure 20: Analysis of simplified model(Hybrid approach)
The below Chart 1 shows that thecomparisonof CAD model simplificationtechniques based onEquivalent Von-mises stressand chart 2 showscomparison of computation time reduction.
Chart 1: Equivalent Von-misesStress comparison
Chart 2: Comparison of Computational Time
343
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
more in preparing CAD models forsimulationpurposes. Thus the comparison showshybridapproach gives acceptable results withminimumerror.
SCOPE AND FUTURE WORKThis work helps to the industrialists whoareinterested more and more in preparingCADmodels for simulation purposes. It is fullybased onsimplification of features like holes,chamfers andfillets. At short notice, In Future it isimportant toconsider other criteria of simplificationsuch asvolumes, size and surface entity basedwith thissimplification technique.
REFERENCES1. Danaan Metge, “Modeling and Simulating
Mechanical Systems on a TransformingDicycle”, BPG Motors, 2013.
2. LEE sang Hun and LEE Kunwoo, “Feature –based multiresolution techniques for productdesign”, Journal ofZhejiang UniversitySCIENCE A 7(9):1535-1543, 2006.
3. Yogesh H Kulkarni, Anil Sahasrabudhe,Mukund Kale, “Strategies for using featureinformation in model simplification”.International Conference on Computer AidedEngineering, 2011.
4. J.Edlund, E.Kermati and M.Servin, “A Long-tracked Bogie for forestry machines on softand rough terrain”, Journal ofTerramechanics, 2012.
5. Metge, “Modeling and Simulation ofMechanical Systems on a TransformingDicycle”, BPG Motors, 2013.
6. Erick R. Denlinger, Seffirwin and PanMichaleris, “Thermomechanical modeling
and simulations of additive manufacturingof Aerospace parts”,J.Manuf.sci and engg136(6), 2014.
7. Letherwood M. gunter, “Virtual modeling andsimulation of military ground vehicles”,SAEtechnical paper,2000-01-1580, 2000.
8. Wang.H.Maz and Kikuchi.N.Pierre, “Multi-domain topology optimization for vehiclestructure crash worthiness design”, SAEtechnical paper, 2004-01-11734, 2004.
9. Atul Thakur, Ashis Gopal Banerjee, andSatyandra K. Gupta, “A survey of cadmodel simplification techniques forphysics-based simulation applications”,Computer-Aided Design, 41(2):65- 80,2009.
10. Hamdi Mounir, Aifaoui Nizar and BanamaraAbdelmajid,“CAD model simplificationusing a removing details and merging facestechnique for FEM simulations”. Journal ofMechanical Science and Technology26(11):3539-3548, 2012.
11. H Zhu and C.H Menq, “B-rep modelsimplification by automatic fillet/roundsuppressing for efficient automatic featurerecognition”.Computer-Aided Design,34(2):109 - 123, 2002.
12. R. Ferandes, “A posteriori evaluation ofsimplification details for Finite ElementModel Preparation”, computers andstructureVol-87, 73-80, 2009.
13. Shuming Gao, “Feature Suppressionbased mesh model simplification”, CADVol-42 1178-1188, 2010.
14. R.J.Donaghy & C.G.Amstrong,“Dimensional Reduction of surface for
344
Int. J. Engg. Res. & Sci. & Tech. 2015 Subash karan R and Ravichandran R, 2015
This article can be downloaded from http://www.ijerst.com/RASET-2015.php#1
analysis”, Engineers with Computers, 2000.
15. M. Handi, B.Louchichi &N.Aifaooi, “CAD /CAE interoperation, an automatic generationof analysis model based on thesimplification of CAD geometry”, CFM,France, 2009.
16. Ousama Jaider, “Manufacturing computeraided process planning for rotational parts”.International Journal of Innovation andApplied Studies, 307-319, 2014.
17. Bob Evans, Doug Fisher, “OvercomingProcess Delays with Decision treeInduction. Machine Learning”, IEEEEXPERT, 1996.
18. Jian-Fengyu, Hong Xiao, Jiezhong & HuiCheng, “CAD model simplification forassembly field”, Int J Adv Manf Technol,2013.
19. A.Sheffer, “model simplification for meshingusing face clustering”, CAD-33(925-934),2001.