You are on page 1of 51

CaliforniaStateUniversity,Northridge

DepartmentofMechanicalEngineering

ComputerAidedAnalysisandDesign
ME386/L
ProfessorKhachatourians
Spring2015

RobertTimm
RachelForeman
KevinMatsuno
PiotrOrzechowski
MarioSolorzano
JeremyWard

May6,2015

TableofContents

ExecutiveSummary.....2

DesignMethodology....3

LoadPathandBoundaryConditions...5

CriticalComponentAnalysis.......9

SS1001:LeftLeverArm....9

SS1002&SS1004:InputandOutputTieRod...22

SS1003:SteeringPlate.30

SS1005:SteeringKnuckle....38

Conclusion.46

References..47

APPENDIX....48

ExecutiveSummary

The Human Powered Vehicle Challenge (HPVC) is a yearly competition hosted by the
American Society of Mechanical Engineers (ASME). Students work in teams to develop
sustainable and practical humanpowered transportation alternatives. The specific objective of
this project was to improve the steering components of the existing California State University,
Northridge (CSUN) HPVteamvehicle.Goalsweresettoreducetheweightsofeachcomponent
60% reduction for the lever arm, 40% reduction for each tierod, 50% reductionforthesteering
plate,and40%reductionforthesteeringknuckle.Additionally,basedonSolidWorksSimulation
finite element analysis (FEA), no part was to fall below a factor of safety (F.O.S.) of 1.5. By
changing the geometry of the lever arm, first tie rod, and second tie rod, their weights were
reduced by 60, 81, and 79 percent, respectively. By changing the geometry and material of the
steering plate and knuckle, their weights were reduced by 78 and 26 percent, respectively.Each
part met the minimum factor of safety except for the steering plate, which fell just short with a
factorofsafetyof1.46.

DesignMethodology

The method used to optimize the steering system began with understanding the current
CSUN HPV steering system. This definition came from the HPV team. Theyprovidedmodels
and identified each of the components in the system. The interface between each of these
components, other HPV components, and the driver were clarified. Critical inputs into the
analysis were the maximum forces that could be applied by the driver, the limit of travel of the
steeringplate,andtheoriginalmaterialsselectedfortheeachofthecomponents.
With this geometric definition of the assembly, the kinematics of the system and its
components could be determined. Thepositionalkinematicsofthesystemweresolvedbasedon
the tie rod configuration. The forces acting on each of thecomponentswas quantifiedwithfree
body diagrams (FBDs) and hand calculations. With this information the loading and boundary
conditionsforeachofthesystemcomponentswasdetermined.
With solid models and boundary conditions for each component, the FEA effort could
begin. The team was separated into groups with each managing a component inthesystem. In
all cases, the components requiredsomeidealizationtobemeshed. Examplesofidealizationare
removing threading, adding fillets to shark, reentrant corners, or simply fixing incorrect
geometries from the original HPV models. Withmeshingcompletethestaticsimulationscould
be run to find the maximum stresses. To achieve better than 5% convergence, manual and
automated mesh iteration was used. Generally, all models required generous use of fillets to
achieve convergence. Achieving convergence on the original parts established baseline for
proper analysis of the optimized parts. The focus could now be turned to the optimization of
eachoftheparts.
The FEA process of creating a Solidworks CAD model, idealizing the model, meshing
the model, running the simulations, demonstrating convergence, and calculating the F.O.S. for
various material candidateswascompleted. Thisprocesswasrepeatedasneededuntilthemodel
meshed properly, converged, and performed well relative to the set goals (F.O.S. and weight).
Meeting the F.O.S. requirements for the system was the primary requirement. Any design that
did not achieve the minimum FOS of 1.5 was rejected. The weight reduction goals for the
3

project were important, but were best effort. The final designs (combination of geometry and
material changes) all showed improvements, butthepercomponentweightreductiongoalswere
not all met. The time constraints for the project eventually required all of the optimization
efforts for the each of the components to end. In the case of the more complex parts (the lever
arm and the steering knuckle), more improvement (reducing weight while maintaining the
minimum FOS) was certainly possible. This additional optimization represents an area of
opportunity for future efforts. Additionally,theinterfacingfeaturesofeachofthepartswerenot
changed so that the system kinematic behavior and the integration of the system components in
Solidworks wouldbeunaffected. Futureworkcouldinvolveimprovingtheseconnectionsaswell
astheindividualcomponents.

Figure1:
FEAdrivendesignprocess.
4

LoadPathandBoundaryConditions

Referring to Figure2,theusergripseachoftheleverarmhandles.Toturn,oneleverarm
is pushed forward while the other is pulled towards the user (green arrows). This causes both
levers to rotate opposite to one another (yellow arrows). The input force from both lever arms
translate through each tie rod and into the the steering plate. Since both input tierodforcesare
off centered, the steering plate rotates andtranslatesforceintotheoutput tierods. Notethatpart
of the initial input forces will diffuse intotheHPVframefromthehingedportionofthesteering
plate. The remaining forces translate through both output tie rods to the steering knuckles, and
ultimately rotate the front wheels. The remaining forces dissipate into theportionsoftheframe
supportingthefronttiresaswellasintotheground(bluearrows).

Figure2:
(Left)Theloadpathofthesteeringsubassembly.(Right)Coloredlegendofeacharrow.

In addition to the load path, boundary conditions were established to better represent
realisticscenarios.First, incalculatingtheloadforeachpart,themaximum amount ofpushing or
pulling force generated from an average adult male arm needed tobedetermined. TheNational
Aeronautics and Space Administration (NASA) published a paper that measured the average
adult male strength to the 95th percentile. The acquired results were used for the HPV steering
subassemblyandcanbeseeninFigure3.

Figure3:
ChartsfromNASAHumanStrengthStudyshowingamaximumforceof56lb.

Another boundary condition established was the operating condition. All analysis was
conducted in ambient,noncorrosive,environmentalconditions. Asaresult,anyfailurefromthe
component is due to the stress from the applied loads. In addition, each component of the
subassembly was examined under a locked, static scenario in order to calculatethemaximum
loadingconditions.
The final boundary conditions were determined by describing the system kinematics
mathematically. The steering system kinematics can be described by a four bar linkage. The
fourlinkagesontheHPVvehicleare(Figure4):
Thesteeringplateedgea.Thisisthedrivinglinklength5.00inches.
Thetierodb.Thisisthecouplinglinkof14.25inches.
Thesteeringknucklec.Theisrockerlinkof3.79inches.
Thefixedlinkd.Thislinkisfixedatbothendstotheframe.Itis15.78inches.
The four bar linkage position equations are derived by understanding each linkage is a
vector and the four vectors always add to zero(evenastheangleofthedrivinglinkischanged).
Theinputsintotheequationarethelengthsofeachlinkandtheangleofthedrivinglink.
6

TherangeofmotionpossiblefromforthesteeringplatewasgivenbytheHPVteam.It
is+/18fromstraight.Withthisinputandtrigonometry,theanglesforeachoftheother
linkagescanbesolved.Theangleforeachlinkageisdefinedrelativetothegroundlink,d.

anglebetweenaandd.78.84whensteeringisstraight.Canvaryfrom
2
60.84to96.84.Thisisaninputintothefourbarlinkageequations.Thisangle
drivesthepositionoftheotherlinkages.

anglebetweenbandd.
3

anglebetweencandd.
4

Figure4:
Thesteeringsystemgeometrywiththefourlinkagesdefined.


Figure5:
Thefourbarlinkagesolutionfor
and
(asafunctionofthesteeringangle).The
3
4
steeringangleis0whenthesteeringisstraight.

Figure6:
Thegraphicalrepresentationofthetravelofthesteeringsystemwhenthesteering
plateisrotated+/18.Thetoplinkage(d)isthegroundlinkitdoesnotmove.
8

CriticalComponentAnalysis

LeftLeverArm(SS1001)
The HPV teams left lever arm (Figure 2) is used to initiatingmotionwithinthesteering
subassembly. The lever arm has already been builtbytheHPVteamtothedimensionsspecified
in their original SolidWorks model. The arm spans approximately 22 in. by 18 in. and its total
width is approximately 3 in. (Figure 7). The lever arm is made of Aluminum 6061T6, and it
weighs approximately 1.91 lb. (Table 1). It is a 1 by 2 rectangular beam weldment with a
thicknessof0.125(Figure8).

Figure7:
Originalleverarmleftandfrontviewwithdimensions.

Figure8:
Leverarmbodycrosssectionwithdimensions(handlehasdifferentdimensions).
9

The lever arm is made up of nineseparatepieces.Theywereweldedtogetherusingfiller


weld rod 4043 (Figure 9). Yield strength of aluminum60601T6is39,885psi.Yieldstrengthof
postweld, heattreated, and aged alloy 4043 with aluminum 6061 is approximately 40,000 psi
(see Appendix, Figure 44). Therefore, for FEA purposes the lever arm can be considered one,
solidcomponentwithauniformstiffnessmatrixandayieldstrengthof39,885psi.

Figure9:
Originalleverarmweldmentshowingeachseparatesection.

The HPV team provided the fully dimensioned model of their original lever armdesign.
Since our goalwastooptimizethemodelbyreducingweightwhileachievingaminimumF.O.S.
of 1.5, we needed to: (1) identify the worst case loadingscenario,(2)drawafreebodydiagram,
(3) apply the HPV team loading conditions and fixtures, (4) verify the HPV team FEA model
and results, and (5) optimize the lever arm model and compare its FEA driven results to the
originalHPVteamdata.

Based on the previously determined force of 56 lb,aswellthefixturesassumedfromthe

lockedscenario,aFBDwascreatedusingtheoriginalSolidWorksmodel(Figure10).

10


Figure10:
LeverarmsFBDwithmaximuminputforceof56lb.

Defining all distances and taking a moment about the lever arms pivotpoint,NewtonsSecond
law produces a general equation of the output force for theleverarm(Eq.1).TheForceoutin
Figure10,representsthereactionforceexertedbythetierodbackontotheLeverArm.

(Eq.1)

Applying 56 lb. on the handle results in the tierod experiencing 425 pounds of force. Table 1
summarizesallFBDfindings.

11

Table1:
GivenandcalculatedFBDleverarmvalues.

Although the HPV Team had previously performed FEA analysis, they were not able to
locate their numerical results to pass on to our team. However, they provided their lever arm
model and the external loads and fixtures that they applied totheirmodel(Figure11).TheHPV
Team applied a standard fixed geometry to the inner circumference of the pivot pin surface.
Unfortunately, the HPV Team did not consider the reactionary force from the tierod, which is
necessaryintheanalysisofthispart.

Figure11:
HPVteamFEAmodel.
12


The HPV team treated the entire boxcross section of the model as a solid. The cylindrical
handles (Part 7 and 8,Figure9)weretreatedasshellswiththicknessof0.125.Theendsurfaces
of the cylindrical shells were manually bonded to each other and tothesolidLeverArmsection
(Part6,Figure9).Theresultsofcoarse,default,andfinemeshdensitiesareshowninTable2.

Table2:
FEAresultsdrivenbyHPVTeaminterpretationofloadsandfixtures.

Although there appears to be a convergence in the maximum von Mises stress across
course, default, and fine mesh densities, using mesh control applied toarandomreentrantedge
th
revealed a slightly higher von Mises stress (Table 1,4
column,FineMeshwithMeshControl).

An hadaptive study confirmed that the maximum von Mises stresses did not converge (Figure
12). Therefore the results undercourse,default,andfinemeshdensitycolumns usedbytheHPV
teamtoidentifyquietareasandminimumfactorofsafetyintheLeverArmweresuspect.

13


Figure12:
HadaptivestudyonHPVteamleverarmmodel.

While each mesh refinement shown in Table 1 brings about an increase in the maximum
displacement, the difference between consecutive results decreases. Therefore, theonlyresultin
the HPV Teams FEA study that provides accurate values is the displacement (0.33). Because
factor of safety is dependent on stress, italongwiththevonMisesstressespreviouslycalculated
could not be used in our study. Due to the significant amount of stress singularities, the HPV
TeamsFEALeverArmmodelhadtobecleanedup.

In order to eliminate all stress singularities, 0.1 inch radius fillets were applied to all

sharp reentrant corners (Figure 13). Since maximum displacement would occur at the location
of theinputload(56lb),thecylindricalhandles(Part7and8,Figure9)wereremovedandPart 6
of Figure 9 was extended in order to imitate the contact force (Figure 14). A circular split line
was created on the surface of the Lever Arm where the tie rod is attached. The split line
represented a washer that would be placed under the tie rod bolt, holding it in place. Standard
fixed geometry was applied to the split line surface and to inner wall (Figure 13). A shaft was
created and inserted into the Lever Arms pivot point hole. This shaft was necessary to It was
bonded to the Lever Arm and a fixed hinge was applied totheinnerwallsof theshaft(Figure
13).

14


Figure13:
FixturesandfilletsaddedtoleverarmFEAmodel.

Figure14:
AppliedfixturesandexternalLoadsoncleanedupFEAleverarmmodel.

15

After applying the necessary fixtures and loads(Figure14),anhadaptivestudyrevealed


(after 3 loops and convergence of 0.66%) a maximum von Mises stress of 27,367 psi with a
minimum factor of safety of 1.46 (Figure 15). Results from the hadaptive study are shown in
Table3.

Table3:
FEAdrivenresultsformodified(cleanedup)leverarmmodel.

Maximum von Mises stress and minimum factor of safety occurred at the intersection of Part 1
and Part 2 (see Figure 9 for reference) (Figures 15 & 16). Displacement from Table 3 was
disregarded since the handle was not necessary for the hadaptive stress study. Maximum
displacement wastakenfromTable1.TheModifiedFEALeverArmmodelwascreatedinorder
to determine the maximum von Mises stress in the original HPV Teams model. Displacement
from Table 1 and maximum von Mises Stress from Table 3 will be used to compare the
optimizedmodelwithHPVTeamsmodel.

16


Figures15&16:
LocationsofmaximumvonMisesstress(left)andminimumFOS(right)on
modifiedleverarmmodel.

The decision was made to continue theuseof6061T6aluminumfortheoptimizedlever


arm. The initial optimized design included large radii at thebends(Figure17,left).Thepurpose
of this was to immediately reduce the chance for stress singularities in the analysis.Ratherthan
using weldmentsinSolidWorks,thepartwasmodeledasonesolidpiece.Thehandleofthelever
wasshortenedanditsthicknesswasreducedtolowertheoverallweight(Figure18,bottom).

Figure17:
Initialleverarmoptimizationwithlargeradii(left)comparedtooriginal(Right).

17


Figure18:
Initialleverarmhandle(top)versusreducedhandle(bottom).

The original dimensions of the rectangular cross section were 2.00 x 1.00 x 0.125. To

continue to reduce the lever arm weight, the thickness of the rectangular bodywasfirstreduced
to 0.056 the smallest standard thickness of 2 by 1 tubing based on SolidWorks Weldment
Profiles. This thickness was found to produce stresses well above the yield strength of the
material. The thickness was gradually increased based on standard available rectangular pipe
sizes until a suitably low stress was achieved. The thickness eventually chosen was 0.09375, or
3/32inches(Figure19).

To furtherreducethe weight,theouterdimensionsoftherectangulartubewereshortened

as much as possible. This was also an iterative process the longer side of the tube wasreduced
by 1/8 increments until the factor of safety no longer met the goal of 1.5. After the smallest
acceptable dimension was determined to be 1.75 for the long side, the same procedure was
repeated for the short side. After numerous iterations, the final dimensions of the rectangular
tubewere1.75x0.75x0.09375(Figure19).
18

Figure19:
Originalleverarmcrosssection(left)andoptimizedleverarmcrosssection(right).

Withthenewdimensions,thepartwasanalyzedusingsecondordersolidelements.A
default,curvaturebasedmeshwasusedtomoreaccuratelysolveforthestresses.Inaddition,an
hadaptivestudywasimplementedwithatargetaccuracyof99%andamaximumofthreeloops.
Byfullyfixingtheholewherethetierodconnects,puttingafixedhingeatthesecondhole,and
applyinga56poundforcenormaltothehandle,thestressresultsseeninFigure20were
obtained.
The stress results between the second and third loop had a convergence error of 2.00%
within the 5% convergence criteria.Thisresultedinafactorofsafetyof1.58,abovetherequired
minimum of 1.5. This also reduced the weight by 47% in comparison to the original model
(Table 4). While this did not quite meet the reduction goal of 60%, the results were acceptable,
andthuswerepresentedinthegroupscriticaldesignreview(CDR)onMonday,April20,2015.

19


Figure20:
FEAvonMisesstressresultsonoptimizedleverarm.

Table4:
FEAdrivenresultsfromoriginalandoptimizedleverarms.

Although the results of the optimized design were already an improvement on the

original lever arm, further optimization was performed after the CDR presentation. To more
accurately represent the model, the design was altered back to weldments per the original HPV
model. The handle was still shortened and reduced in diameter and thickness to begintoreduce
the weight. The final handle was 0.75 diameter by 0.035 thickness this is the smallest
thicknessavailableatthisdiameter.6061T6Aluminumwasstillused.

20

The rectangular tubing was reduced to 1.75 x 0.75 based on the results of the first
optimization. However, the thickness was reduced to 0.064 to see if the weldment geometry
could handle the forces with lessmaterial.Topreventstresssingularities,0.2filletswereadded
to all of the interior edges of bends along the lever arm. The same fixtures and force that were
applied to the firstoptimizedleverarmwereappliedtothissecondoptimization.Thesame mesh
propertieswereused,withtheexceptionofthemaximumnumberofloopsbeingraisedtofive.

The results of the secondoptimizationsurpassedthoseofthefirstoptimization.Thenew,

lower stress results can be seen in Figure 21. The lower maximum stress yielded an improved
factor of safety of 1.78. Most importantly, the new optimization reduced the partweightto0.80
pounds. Based on the original weight, this would be a 58%weightreduction.However,anerror
was found in the original weight calculation (the weight was initially measured without both
sections of the handle included)incomparisontothecorrectedoriginalweight,thispartmetthe
weight reduction goal with a total reduction of 60.2%. The comparison between the original,
first,andsecondoptimizationswiththeadjustedoriginalweightcanbeviewedinTable5.

Figure21:
FEAvonMisesstressresultsonthesecondoptimizedleverarm.

21

Table5:
FEAdrivenresultsfortheoriginal,firstoptimized,andsecondoptimizedleverarmmodels
withcorrectedweightvalues.

InputTieRod(SS1002)andOutputTieRod(SS1004)

Figure22:
Tierodlocationwithinsteeringsubassembly.

Shown highlighted in red in Figure 22 are both tie rods, their general dimensions, and

placement in the steering subsystem. In optimizing the tie rods, one of the main constraints on
optimization was to keep the original lengths of the members,andtonotchangethetierodends
that connect thesystem.Bothtierodswere runusing similaranalyses,thereforeanalysisofthese
parts will have results shown together to compare and contrast between the two. The input and
output tierodsaresimple,twoforcemembers.Theinputtierodreceivesload fromtheleverarm
via its tierod end connection to themember.Thisforceistranslatedintotherod,wheretheload
22

then distributes to the ternary plate via a similar connection. The ternary then translates the
forces into the output tie rods, and finally that load is transferred to the knuckle through the
output tie rod. With the locked steering scenario, the opposite sense of the forces on each
respective side of the system mean that the tie rods canandwillbesubjectedtobothtensileand
compressive loading in right and left turn scenarios, respectively. These two loading scenarios
areshowninFigure23below:

Figure23:
Loadingconditionsappliedtoeachtierod.

Being long, slender members that are subjected to compressive loads, it is obvious that
another mode of failure for these parts is buckling. Therefore, compressive and tensile static
linear structural analyses were carried out, along with buckling analysis for both members in
hopes of reducing the weight of these components by a minimum of 40%throughoptimization.
Interestingly, the output tie rod sees much less force than the input receives, at just 152 pounds
of force versus 425 pounds of force for the input tie rod. This was taken into consideration,but
because of the differing lengths of the tworods,staticstressanalysisandbucklinganalysiswere
performedonbothtoconfirmthattheinputtierodwasindeedthecriticalmember.

23

The first step in the analysis process of these components was defeaturing and
idealization of the model. This process included removal of the threads on the two ends of the
rods, and the addition of some generous fillets on the interior change in diameters of the rod.
Additionally, it should be noted that the two ends of the rods have inserts to place the smaller
quarterinchnominalthreads intothemuchlargerinnerdiameterofthe rodmaterialcomingfrom
McMaster Carr. For this analysis these inserts areassumedtobesolidandapartoftheroditself
and this was part of the idealization of the tie rods.Inpracticetheseinserts would eitherneedto
be welded, press fit, or glued into the ends of therodsthemselves.Figure24belowshowssome
ofthedefeaturingandidealizationofthemodel.

Figure24:
Defeaturing,idealization,andadditionoffilletstothetierods.

In order validate the idealization made by removing the threads from the model, aquick
thread tearout hand calculation was performed to make sure that the threads would not tear out
under the applied loading. This calculation was absolutely necessary prior to performing any
FEA analysis on this rod. If the threads wouldhavefailedonthispart,theloadingappliedtothe
rest of the rod would never reach the value being put into the FEA model. Because only three
threads typically yield and hold load, three times the pitch was used for the thread engagement

24

length for a conservative estimate. This calculation confirmed that the threads would not be a
problem, so long as 6061 T6 was the agedstateofthethreads.Thefactorofsafetyfoundforthe
threadtearoutwasabout7.2.TheresultsofthesehandcalculationscanbeseeninFigure25.

Figure25:
Threadtearoutcalculationstovalidateidealizationassumptions.

After defeaturing and idealizing the tie rods, the restraints and loads were applied. It was

determined that it would be beneficial to split the model in half and apply symmetry boundary
conditions to the model. This had twofold benefits it reduced the amount of time to mesh and
run each of the many studies needed, and made the visualization of maximum stress locations
much easier. This also allowed for quickupdatestononconvergentareas like theinteriorfillets.
A fixed restraint was applied to the cylindrical face on one side of the rod, where the threads
would be taking the load. Similarly, the load was applied to the opposite end of the rods onthe
same cylindrical face where the threads wouldtaketheload. Becauseofthesymmetryboundary
conditions, half the expected load was applied to this cylindrical surface, normaltotheendface
of the rod in both tensile and compressive loading scenarios, respectively. Figure 26 shows the
symmetryboundarycondition,restraintsandloadsappliedtothemodel.

25


Figure26:
Restraintsandloadsappliedtotierodmodels.

Once all of the idealization and boundary conditions were applied, a mesh was created.
Solid, standard, high quality elementswereusedforthesepartssincetheaddedinsertsattheend
of the rod didnt allow for shell or beam elements to be used in this case. A static linear study
was then carried out for the initial tubing size to find outhowhighafactorofsafetythetierods
had before moving forward with optimization. The mesh was manually edited, using coarse,
default, and then fine element size in order to confirm convergence of the model. This
convergence was calculated for each step by finding the convergence error between the default
and fine mesh steps. Other than the first static study performed, and one other study on the
output tie rod with extremely low stresses, all models achieved convergence onthefirsttry. For
the output tie rod study that failed the first convergence test, a quick change of the mesh to a
curvature based fine mesh compared with the last fine mesh produced convergent results. The
initial factor of safety for the input tie rod with an outer diameter of was found to be 29.
Since this factor ofsafety wassohigh,thetubing sizeselectedfortheseconditerationwascutin
half, with a outer diameter. This resulted in a factor of safety for stress of 11.68 for both
tensile and compressive loading scenarios. FEA von Mises plots for thetensileloadingscenario

26

are shown in Figure 27. The results showing von Mises Stress, max deflection and minimum
factorofsafetyfromthesestaticstudiesforbothloadingscenariosareshowninTable6.

Figure27:
FEAstaticvonMisesstressfororiginal(left)andoptimized(right)tierodmodels.

Table6:
FEAdrivenstaticresultsoforiginalandoptimizedtierods.

AsisevidentinlookingatTable6,stressvalueswerestillnowhereneartheyieldstress
ofaluminum6061T6,howeveritwasnecessarytoconfirmthatthesestudiesforcompression
andtensionwereactuallythecriticalmodeoffailureforthepartmovingforward.Sincetheyare
longandslendermembers,bucklinganalysiswascarriedoutforeachrod.Bucklingplotsare
showninFigure28,whereasTable7liststhebucklingloadfactorsfoundforeachtubesize.
Mode1wasthefirstpositiveBLFvaluefound,sothismodewastakentogivetheBLFforall
bucklingtestsperformed.

27


Figure28:
Bucklingplotsfororiginal(left)andoptimized(right)tierodsizes.

Table7:
BLFfactorsfrommodeoneofbucklinganalysisforeachrodandsize.

FromTable7andTable6,itisclearlyevidentthattheinputrodisthecriticalmemberof
thetwotierodsinthesystem.Evenwiththelongerlengthoftheoutputrod,theBLFstillcomes
outhigherduetothelowerforceexertedintothismemberbytheternary.WithaBLFofjust
1.845,theinputtierodwasdeemedtobefullyoptimized,asthenextsizedownfromwould
haveproducedamemberthatbuckledunderthesameloadingconditions.Additionally,thenext
sizedownfromwouldalsorequirereducingthesizeofthetierods,asthemajorthread
diameterwouldthenbelargerthantheouterdiameterofthenextsizedownintubingfrom
McMasterCarr.Thiswouldhavedefeatedtheoptimizationintentofkeepingtheoriginaltie
rods,thereforeouterdiametertubingwasthefinaloptimizedsizingselectedforthetierods
basedonbucklingfailure,havingacriticalfactorofsafetyof1.845.MultiplyingtheBLFfactor
bytheappliedloadof425poundsgivesabucklingloadof784poundsallowedfortheinputtie
rod.
Changingthematerialofthetierodswasreallyoutofthequestionforthisoptimization
project.Theonlymaterialthatcouldhavereducedtheweightsignificantlywouldhavebeen
magnesium,butbecauseoftheabilitytogetthestockforthispartofftheshelfinaluminum
28

6061,itwaspointlesstotryandfabricatethispartoutofmagnesiumtotryandshaveoffafew
extraounces.

Table8:
FEAdrivenresultsfororiginalandoutputtierodsinstaticandbucklinganalysis.

Optimizationofthetierodsresultedinaboutan80%reductioninweightfromthe
originaldesign(Table8),whichwasdoublethegoalof40%weightreductionthatwasoriginally
settoachieve.Theoriginaldesignwasveryoverdesigned,andthisoptimizedsizingwillresultin
bettersteeringresponseandalightervehicleoverall.Notmuchelsecanbedonewiththetierods
movingforward,asensitivitystudyonthispartwouldlikelyshownegligiblegainsbyremoving
morematerial.Lookingatreducingthetierodendsizewouldbethenextstepinreducingweight
ofthispartofthesteeringsubsystem.

29

SteeringPlate(SS1003)

Figure29:
Envelopedimensionsofthesteeringplate.

The steering plate receives the input force from the two lever arms and thentranslatesit
to the steering knuckles this causes thefront tirestorotate. Likeothercomponents,thesteering
plate was studied as a linear, static structural analysis. To begin analysis, the maximum forces
experienced by the part needed to bedetermined.Asaresult,twoFBDswerecreatedthatvaried
the steering plates andoutputtierodspositions:onein theidleposition(norotation)andoneat
the maximum rotation (18 from the yaxis). The rest of the steering subassembly components
have mirror parts as a result, loads from only one of each of these parts is needed for the
analysis of those parts. Alternatively, the steering plates loading scenario must include the
forces from all duplicates parts. Ultimately, the fully rotated scenario experienced the greatest
forces. Both output tierodsgeneratedaresultantforceof152lb.onthesteeringplate. Belowis
the FBD for this loading scenario in (Figure 30) (Scenario 1 can be found in the Appendix,
Figure45).

30


Figure30:
(Right)FBDoftheoriginalsteeringplateatfullrotation.(Left)Summaryofforces
andassumptions.

Next, SolidWorks Simulation was used to determine the original plate design weight,
maximum von Mises stress, maximum deflection, and minimum F.O.S. Due to its simplistic
design and clean CAD model, no defeaturing or cleanup was needed to begin the analysis.
Model preparation was completed by adding split lines to divide the two smaller holes
cylindrical faces in half. This was necessary to simulate the approximate area on which the tie
rod input and output forces would act. When applying loads, the resultant output tie rod force
was broken into cartesian (x and y) components. These components accounted for both the
steeringplateandtierods'anglesofrotationwithrespecttooneanother. Finally,thedirectionof
each force was applied normal to eitherthestandardfront,right,ortopplaneandplacedononly
one side of the split line face. In order to fix the plate, an advanced fixture was used on the
larger (bottom) cylindrical face that restrained the part from moving in the radial,
circumferential,andaxialdirections.Figure31showstheFEAmodelusedforanalysis.

31


Figure31:
FEAmodelwithboundaryconditionsfororiginalstaticanalysis.

Due to the steering plate volume, second order solid elements were used. To obtain
accurate results, a finer meshdensitywasusedineverysuccessiveiteration.Coarse,default,and
fine meshes were tried for the original and each optimized geometry. If more iterations were
needed, mesh control and curvature based meshes were applied in local areas.

For each

iteration, a standard spreadsheet established by the group was filled out (see Appendix, Figure
46). Finally, the official data was collected once the von Mises stress between two consecutive
runsmetthe5%convergencecriteria.TheresultscanbeseeninFigure32.

32


Figure32:
(Top)FEAvonMisesstressresultsofthefifthiterationofanalysis.(Bottom)
Summaryofmanualconvergenceiterationresults.

The maximum von Mises stress, maximum deflection, and minimum F.O.S. occuratthe
upper right corner of the steering plate. This is the case when the steering plate was rotated
completely to the right. Due to symmetry, the upper left corner would have identical results if
theplatewasrotatedfullytotheleft.
Based on the results, theplategeometry, material,andthicknesscouldbealteredtoreach
the 50% weight reduction goal while maintaining a F.O.S. of 1.5 or greater. To find the best
optimized design, a manual sensitivity study was conducted. First, keeping the geometry and
thickness fixed, each material from the table of material candidates was applied (see Appendix,
Table 16). The same data and convergence criteria from analysis of the original design was

33

used. Next, the geometry of the design was changed while the material and thickness were
maintained.Inoptimizingthegeometry,twodesignswerechosen(Figure33).

Figure33:
Original,firstoptimized,andsecondoptimizedsteeringplatedesignsaslabeled.

In addition, all combinations of geometric designs and material candidates wentthrough


a linear, static structural study with the same data objectives and convergence criteria. Finally,
the thickness was altered in combination with the geometric design and material candidates.
Originally, the thickness oftheplatewas0.25inches. Theoptimizedthickness,0.16inches,was
chosen from McMaster Carrs standard sheet/bar/strip 6061T6 thickness list. The individual
material and geometric designvariablestudiesunderwentthesamemanualconvergenceprocess.
However, due to the limitation of time, the studies combining material,geometry,andthickness
were solved using hadaptive studies with a 99% target accuracy. Tables 9 and 10 provide a
summaryofallthestudiesperformed.

34

Table9:
Summaryofanalysisresultsforeachsteeringplategeometryandeachmaterialtried.
DarkorangecellsindicatebothweightandF.O.S.goalsweremet.

Table10:
FinalanalysiswithpreviousproposedgeometriesandmaterialANDreducedthickness.

35

Figure 34 showsthechosenoptimizeddesign. Basedonthemanualsensitivitystudy,the


best design is optimized design 1, and should be made from AZ61AF magnesium alloy with a
thickness of 0.16 in. Unlike the original design, the optimized design has its maximum von
Mises stress, maximum deflection, and minimum F.O.S. in the upper left corner. Although the
minimum F.O.S. is below the required 1.5, it is highly localized. The majority oftheparthasa
calculated F.O.S. well above 1.5. As a general observation, cutting the geometry introduces
more stress concentrations. As a result, optimized design 2 was rejected due to the increased
amounts ofcriticalstressesproducedinthepart. TheresultsusingaAZ61AF,optimizeddesign
2,anda0.16in.thicknesscanbeseenintheAppendix(Figure47).

Figure34:
FEAvonMisesstressplotandsummaryofresultsforoptimization1(chosendesign).

Another study was conducted to calculate the minimum F.O.S.atthetierodholesdueto


tearout shear and direct bearing stresses. The tabulated summary, formulas, and minimum
F.O.S. for each material can be seen in Figure 35. In the optimized design, tearout andbearing
stresseswillnotcausetheparttofail.

36

Figure35:
(Left)Summaryoftearoutshearstresscalculations.(Right)Summaryofbearing
stressanalysis.Verificationthatthepartwithnotfailduetotearout.

In summary, the optimized design can successfully reduced the weight by 78% and still
maintain a F.O.S. of 1.5 and above. The changes in material, geometry, and thickness allowed
the component to perform in the worst case scenario. A comparison between the original and
optimized steeringplatecanbeseeninFigure36. Intermsofmanufacturing,optimizeddesign1
would need tobemilledtoremovetheinteriormaterial. Inaddition,platecanbepurchasedwith
the 0.16 thickness or machined to achieve this. Future work for this part includes performing
bucklinganalysisonthechosenoptimizeddesign.

Figure36:
(Left)Chosensteeringplategeometry.(Right)FEAdrivenresultsfororiginaland
chosenoptimizeddesigns.

37

SteeringKnuckle(SS1005)
The boundary conditions for the steering knuckle were established based on input from
the HPV team, assuming the locked scenario, and guided by the previously determined
kinematic positional results (Figure 37). The knuckle attaches to the frame, the wheel hub, the
output tie rod, and the brake calipers. The forces applied during breaking were given by the
HPV team. From the kinematic analysis, the tie rodforcevectorisknowntobenominallyatan
angle of +14.0 (Figure 5). When the turning fully totheleft,thetierodvectorisatan angleof
+44.2. When the steering is hard right, the tie rod vector is at10.8. Therearemanydifferent
possible combinations for the steering position, brake application, and turning force direction
between these two angles. The worst case condition used for modeling and optimization was
steering fully left (or right), maximum applied force from the rider, and brakes applied. The
sensitivityofthedesigntochangesintheboundaryconditionswastestedandwillbediscussed.

Figure37:
Fixturesandloadsappliedtotheoriginalsteeringknucklemodel.

38


Figure38:
Angleoftheforceinputtotheleftsteeringknuckle.

At the tie rodmounting,theangleofforceinputtotheleftknucklevaries. Forahardleft


turn, the angle is 44.2 on the other hand, a hard right turn yieldsanangle is10.8. Forinitial
analysisthenominal,14anglewasused(Figure38).

Figure39:
Explanationofinandoutforces.
39

Theforcefromthetierodcanbeeitherintheindirectionortheoutdirection,asseenin
figure39.Bothboundaryconditionswereanalyzedandgaveverysimilarresults.
The FEA effort began with the analysis of the existing HPV team part (the left side
steering knuckle). Due to the presence of sharp corners, the model required idealization in the
form of adding fillets to obtain convergence (Figure 40). The model wasmanuallymeshedand
the convergencecriteriaof5%wasmet. ThemaximumvonMisesstresswasfoundtobe 17,627
psi and the minimum F.O.S. was2.26. Themaximumstresswaslocatedonthetransitiontothe
flat section of the part, where the frame mounts to the knuckle. The results for both the tie rod
inandoutforceswereverysimilar(Table11).

Table11:
FEAdrivenresultsfortheoriginalsteeringknuckledesign.

40


Figure40:
FEAresultsfortheoriginaldesignwiththeinforceapplied.

For the optimized design, material was removed from quiet areas of the part based on
the original FEA. The stress concentration at the transition to the flat surface was addressed by
adding material at this corner. Additionally, ribs were added to resistthemomentcreatedbythe
tie rod force (Figure 41). The result was approximately a 60% reduction in maximum stress.
Although the part volume was increased approximately 10% which these changes, the weight
was still reduced by switching to a lighter, magnesium alloy material. The minimum F.O.S. for
this critical part was increased from 2.3 to 5.0 while the weightwascutfrom0.27lb.to0.20lb.
(Table12).

41

Table12:
FEAdrivenresultsfortheimprovedknuckledesignusingnonadaptivestudies.

Figure41:
FEAresultsoftheoptimizeddesignwiththeinforceapplied.
42

Analysis was conducted to understandthesensitivityofthe FEAresults tochangesinthe


boundary conditions. It was determined that the analysis was insensitive to a reversal in the
direction of the tie rod force from the nominal. However, changes in the tie rod force angle
(representing turning fully to the left or right) did have a significant effect on the maximum
stress FEA result. Although the quantitative result was different for these varying boundary
conditions, the overall trend was the same. The optimized knuckle design had lowermaximum
stressandgreaterstiffness(Figures42and43,Table13).

Figure42:
Changingthetierodforceanglefrom44.2(top)to10.8(bottom)ontheoriginal
knucklehadasignificanteffectonthemaxstressquantityandlocation.
43

Figure43:
Aswiththeoriginaldesign,changingthetierodforceanglehadasignificanteffect
onthemaxstressquantityandlocationfortheimproveddesign.

Table13:
Theeffectofchangingthetierodinputforcedirectionfororiginalandoptimizeddesigns.

44

There is ample opportunity for futureworktobeconductedonthesteeringknuckle. The


design could be further optimized to improve manufacturability while reducing weight and
maintaining the the use of inexpensive and widely available 6061T6 aluminum alloy. Ideally,
the effect of other variations to the boundary conditions would be studied taking into account
weight transfer due to breaking. This would also callforverificationofthe brakingcaliperinput
forcesobtainedfromtheHPVteam.

Table14:
FEAdrivenresultsforbothforcescenariosfortheoriginalandoptimizeddesigns.

45

Conclusion

In summary, the CSUN HPV steering subassembly was optimized. Based on the
established boundary conditions, each part had a targetweightreductiongoal tomakethesenior
design vehicle lighter. In addition, a minimum F.O.S. of 1.5 was established to constrain the
amountofoptimizationandensuretheriderssafety.
Using FEA, asensitivitystudywasconductedforeachparttoselectthebestcombination
of design variables, material, thickness, and geometry. Adjusting thethicknessineachpartwas
determined to be most effective in reducing the weight. Out of the five components, only the
steering knuckle was unable to fulfill its weight reduction goal (Table 15). Still, the overall
weightoftheoptimizedsubassemblyis1.20lb,whichisa62%reductionoftheinitialweight.
A minimum F.O.S. of 1.5 was met for eachpartof thesubassemblyundertheworstcase
conditions. The steering plate has the lowest F.O.S., and therefore would be the weakest
componentinthesubassembly.
For future improvements, redesigning the steering knuckle to further reduce weight is
desired. Additionally, designing a simpler lever arm with less complex angles would make the
component easier to manufacture. One area that wouldbebeneficialtoresearchisthecorrosive
resistanceofeachpartinregardstohumansweatandotherenvironmentalfactors.
In terms of lessons learned, being in constant communication isessentialinstreamlining
the design process. Attending group meetings and responding to group members messages
allowsallmemberstogaugethegroupsoverallprogress.
Table15:
AsummaryoftheoptimizedsteeringsubassemblyweightandminimumF.O.S.

46

References

[1] ASME.
RulesForThe2015HumanPoweredVehicleChallenge
.Chicago:ASME,25
Sept.2014.PDF.
[2] BeamsandTrussesOverview.2013SOLIDWORKSHelp.DassaultSystemes,n.d.
Web.19Apr.2015.
[3] Jeeverajan,Anthony.HUMANPERFORMANCECAPABILITIES.
HUMAN
PERFORMANCECAPABILITIES
.nasa,7May2008.Web.17Mar.2015.

47

Appendix

Figure44:
Datasheetfor4043weldwire.
48

Figure45:
FBDofthesteeringplateinidleposition.Forces,equations,andassumptionsused.

Figure46:
(Top)Astandardtableusedforeachgroupmembertotrackanalysisiterations.
(Bottom)Astandardtableforverifyingconvergencebetweeniterations.
49

Table16:
Listofmaterialcandidatesforallsubassemblycomponents.

Figure47:
FEAvonMisesstressplotandsummaryofresultsforoptimization2(notchosen).

50

You might also like