You are on page 1of 9

CHINESEJOURNALOFMECHANICALENGINEERING

Vol.26,aNo. 3,a2013
DOI:10.3901/CJME.2013.03.***,availableonlineatwww.springerlink.comwww.cjmenet.comwww.cjmenet.com.cn

ThreeDimensionalFiniteElementAnalysis of the MechanicalProperties ofHelicalThread Connection


1,2 1,* 1 1 2 1 YANGGuoqing ,HONGJun ,ZHULinbo ,LIBaotong ,XIONGMeihua ,and WANGFei

1StateKeyLaboratoryforManufacturingSystems Engineering,XianJiaotongUniversity,Xian710049,China 2CollegeofElectromechanicalEngineering,HunanUniversity ofScienceandTechnology,Xiangtan411201,China


Received June7,2012revised January28,2013 accepted February27,2013

Abstract: Conventional analytical and numerical methods for the mechanical properties of helical threads are relied on many assumptionsandapproximationsandthushardlyyieldsatisfiedresults.Aparameterized3Dfiniteelementmodelofboltedjointswith realhelicalthreadgeometry isestablishedandmeshed with refined hexahedralelements.ThevonMisesplasticitycriterion,kinematic hardeningruleofmaterialsandinterfacialcontactsareemployedtomakeitpossibleforthesuggestedmodelbeabletoapproachreal assembly conditions.Then,themechanicalpropertiesofboltedjointswithdifferentthreadpitches,threadnumbersandmodularratios areinvestigated,includingthecontactpressuredistributionatjointinterfaces,theaxialloaddistributionandstressconcentrationin screw threads during the loading and unloading process. Simulation results indicate that the load distribution in screw threads producedbythesuggestedmodelagreeswelltheresultsfromCHENsphotoelastictests.Inaddition,aninterestingphenomenonis found that tightening the bolt with a large preload first and then adjusting the clamping force by unloading can make the load distributionmoreuniformandreducethemaximum residualequivalentstressinthreadrootsbyupto40%.Thisresearchprovidesa simpleandpracticalapproachtoconstructingthe3Dfiniteelementmodelandpredictingthemechanicalpropertiesofhelicalthread connection. Keywords:finiteelement analysis, threadconnection, loaddistribution, stressconcentration

1 Introduction*
Thread connection is a basic assembly method in mechanical systems due to easy disassembly, reassembly, adjustment and maintenance. There is no doubt that the geometrical sizes and material properties of thread connection influence its static, dynamic and thermal performances even the overall function of the mechanical system. So evaluating the load bearing capacity of thread connection hasattractedconsiderableattentionofscholars. [1] Electrometric method and three dimensional stress [24] freezing photoelastic method have been used to evaluatetheaxialloaddistributionandstressconcentration of screw threads. The electrometric method uses strain gages to measure the strain of screw threads, but the placement of the gages and wires may change the insitu environment. The photoelastic method can show us the distribution rules of stress and strain in screw threads. However, it is a pity that photoelastic model need to be designedspecially andslicedforpolarization.
* Correspondingauthor.Email:jhong@mail.xjtu.edu.cn This project is supported by National Natural Science Foundation of China (Grant No. 50935006), the Major Project of Highend CNC Machine Tool and Basic Manufacturing Equipment of China (2011ZX04016031) and theHiTechResearchandDevelopmentProgram ofChina(2012AA040701)
ChineseMechanicalEngineeringSociety andSpringerVerlagBerlinHeidelberg 2012

YAMATOTO[5] developedananalyticalmodelbasedon thirteen assumptions and approximationstoinvestigatethe mechanical properties of screw threads. In his model, the thread is simplified as a cantilever to formulize the load distributionasshowinEq.(1). Itexpressesthe axialloadat anysectionperpendicularto theaxisof theboltornut:
Fl = F sinh(l L - l l) , sinh( l L)

(1)

where F is the bolt preload, L is the thread engagement length,listhedistanceawayfromthenutbearingsurface, Fl istheaxialloadatthesectionwithadistanceoflaway from the nut bearing surface, is a constant and can be calculatedasfollows:
-1 ( Ab Eb ) -1 + ( An E n) , -1 ( kb Eb-1 +k n En ) tanb

l=

(2)

whereAb andAn aretheequivalentcrosssectionalareasof boltandnut,Eb,En aretheElasticmoduliof boltandnut, kb andkn are theelasticdeformationcoefficientsofboltand nutthreads, is the helix angleofthreads. [6] WANG and MARSHEK developed a modified spring model to predict the load distribution in threaded connectors with partial yield deformation. By contrast, YAMATOTOsmethodismorepopularbecauseofitsfast

Yang Guoqing,etal: ThreeDimensionalFiniteElementAnalysis of the MechanicalPropertiesofHelicalThread Connection

solution to the load distribution of engaged threads. However,itmerelyfocusesontheaxialload,andnomore information about the stress/strain behavior is available. Therefore, it is difficult for designers to get a deep understandingontherelationshipbetweenloaddistribution patternandstructural behaviorof threadconnection. On the other hand, with the development of numerical techniques,FEmethodhasbeenconsideredtobeoneofthe most powerful tools for engineers. Nonthreaded FE models[710] and2Daxisymmetricthreadmodels[1113] have been employed to simulate the mechanical behavior of threadedfasteners(boltandnut).Thenonthreadedmodels ignoretheinfluenceofscrewthreadsontheloadtransferin thread connections. The 2D axisymmetric thread models can consider the load transfer and stress concentration in screwthreads,buttheyignorethehelicaleffectofthreads. So it is necessary to build a more effective and accurate [14] model in the case of detailed design. IZUMI, et al investigated the tightening and loosening mechanism of threaded fastener using a 3D FE model with tetrahedral elements,buttheirmodelistooroughtoaccurately obtain [15, 16] the load distribution in threads. FUKUOKA, et al constructed a 3D FE model with hexahedral elements, which provides an approach for refining the modeling of the helical thread effectof threadconnection. Therefore, the objective of this paper is to propose a generally accepted procedure for building a 3D FEmodel ofboltedjointwithhelicalthreads,andthentoinvestigate the corresponding mechanical properties. In section 2, a parameterized 3D FE model is constructed, and it is meshed with highquality hexahedral elements. Then in section 3, the mechanical properties of the helical thread connection are analyzed in detail, including contact pressure distribution at the joint interfaces, nonuniform loaddistributioninengagedthreads,stressdistributionand stress concentration. Finally conclusions and recommendationsaregiveninsection4.

perpendiculartotheboltaxiscanbeobtainedbyprojecting the three parts into the plane as shown in Fig. 1b. The threadprofilecanbeexpressedby
d 2 - 7 H 8 + 2 r - r 2 - p2q 2 -2 4 0 q q1 , (3) r' = Hq + d / 2 - 7 H 8 q1 < q q 2 d 2 q 2 < q
0.5 where d isthenominaldiameter,H(H=3 P/2)isthe thread 0.5 height, p is the thread pitch, (3 P/12P/8) is the 0.5 root radius of external threads, 1=3 /P and 2=7/8. Similarly, the profile ofinternal threadscan be formulized asfollows:

d 1 2 H q + d 2 - 7 H 8 r' = d 2 + H 8 - 2r n + r 2 - p 2 -2 (p - q ) 2 4 n

0 q q1

q1 < q q 2 , q 2 < q

(4)

0.5 whered1 istheminordiameter,n(n3 p/24)istheroot 0.5 radiusof internalthreads, 1=/4and 2=(13 n/p).

(a) Axialsection ofexternalthread

C D D' C' 2 d1 1 d

B A Threadroot

2 Numerical Model
The complexity of thread structure and the nonlinear material properties and contact behavior of thread connectionmakeitimpossibletobuildaperfectanalytical model. In order to accurately study the mechanical propertiesof helical threads, anewprocedureforbuilding a 3DFE modelofboltedjoint isproposed. 2.1 Geometrical modeling Anew3Dhelicalthreadmodelisbuilt,andthemodeled thread profile coincides with ISO 68, 261, 262 and 724 standards. The thread pitch, root radius and nominal diameter of bolt are all parameterized to investigate their effectonthemechanicalpropertiesofengagedthreads.As shown in Fig. 1, the profile of the external thread can be divided into three parts such as AB(root radius), BC (thread flank), and CD(crest). The crosssection

B'

(b)Crosssectionof externalthread perpendicularto boltaxis

Fig.1. Thread profile of the externalthread

2.2 Finiteelementmodel In order to mesh the helical thread model with high quality hexahedral elements, the bolt and nut are divided intoalargenumberofvolumesallthecrosssectionshave thesameshape(seeFig.1b),buttheadjacentsectionshave an angle difference of 22.5(360/16) along the bolt axis. As shown in Fig. 2a, each pitch of the external thread is divided into 16 layers of solids, and each layer is divided into 5 solids. Similarly, each pitch oftheinternal threadis

CHINESEJOURNALOFMECHANICALENGINEERING
divided into 16 layers of solids and each layer is divided into 12 solids (see Fig. 2b).The modeled bolt andthenut are shown in Fig. 2c and Fig. 2d, respectively. The proposedgeometrymodelingoftheboltandthenutenable the model to be meshed mostly with 8node solid hexahedral elements so as to improve the solution efficiency and accuracy. Fig. 3 gives out the generated elements in the bolted joint. The modeling method of threadscannotonlyguaranteethediscreteaccuracyofFE elements,but also wellcontrolthe elementnumber.

(a) One externalthread

(b) One internalthread

(c)Bolt

(d) Nut

Fig.2. Solid volumemodelof the bolt and nut

(a) Bolt

(b) Nut

* * threadpitchesp,modularratiosE (E =En/Eb)andengaged threadnumbersNaslistedinTable1.Thetensilestrength, yield strength, elastic modulus and Poisson's ratios of the bolt material are b=1 GPa, s=900 MPa, Eb=200 GPa, v=0.3 respectively. The von Mises plasticity criterion and kinematic hardening rule of the materials are considered. Thetangentmodulioftheboltand nutareobtainedfroms, b andtheircorrespondingstrain(0.2%andtheelongation), aswellasthematerialpropertiesoftheboltedmembersare definedasthesameasthoseinRef.[10].Thethicknessof the members is 15mm, and the radial dimensions of the upper and lower members are greater than the possible contact radius R at the member interface. The assembly clearance between the hole and the bolt is defined as 0.1 mm. The friction coefficient of joint interfaces (the interface between thread pair and those between bolt, nut andmembers)aredefinedas0.1sinceitseldomaffectsthe loaddistributioninengagedthreads[4]. Inthisconstructed3DFEmodel,thecontactrelationship between the bolt,nut andmembersand that of thread pair are established to investigate their contact behavior and corresponding contact characteristics. The end of the bolt head is constrained, and the preload is applied to the unthreaded portion of the bolt with PRETS 179 elements. Thetotalfiniteelementsarelessthan100thousand,which iswithinthescopeofcommoncomputersandFEsoftware. The 3D FE model considers the fine structure of screw threads,andisfullyfittomodelthemechanicalproperties of threadconnection. Meanwhile, a new 2D axisymmetric FE model of the correspondingboltedjoint isconstructedasshowninFig.4. The 2D FE model can consider the nonlinear contact behavior of the joint interfaces. Themember, bolthead or nut interfaces are not constrained in the axial direction and can model the nonliear contact behavior of thread connection, which is great different from those in Refs. [1113]. A2D FEAis conducted for the comparison with the3DFEAand YAMATOTOsanalyticalmethods.

(c) Boltedjoint(the bolted members arehalfshown)

Fig.3. Hexahedralmeshmodel of the boltedjoint Fig.4. 2DaxialaxisymmetricFE modeloftheboltedjoint

TheboltandnutjointswiththenominalthreadsizeM12, property class 10.9/10 are studied by inputting different

Yang Guoqing,etal: ThreeDimensionalFiniteElementAnalysis of the MechanicalPropertiesofHelicalThread Connection

* Dimensionlessengagedthreadlength l

Fig. 6(a) shows the contact pressure distribution at the 3D thread interface which is expanded in the axial and Thread Thread Modular Deformation Deformation Bolted Constant circumferential directions. The horizontal and vertical pitch number ratio coef.ofbolt coef.ofnut joint * coordinates stand for the circumferential radian and the p /mm N E kb kn * * dimensionlessengagedthreadlengthl (l =l/p)respectively. Joint 1 1.75 6 1 3.44 4.49 0.172 Joint 2 1.75 6 1/3 3.44 4.49 0.121 Fortheconsiderationofthe filletofthenutnearthebearing Joint 3 1.50 6 1 3.57 4.86 0.178 surface, the thread in the distance range 00.5p from the Joint 4 1.75 10 1 3.44 4.49 0.172 nutbearingsurfacetransfersnocontactloads. Therearesix threads transferring load in the distance range 0.5p6.5p. Thegray stripesstandforthecontactareaofthreads,which 3 SimulationResults is obviously smaller than the noncontact area (the blank areain therectangle).The contact pressure decreases with The above constructed 3D FE model is applied to the distance from the nut bearing surface. The maximum analyze the contact pressure distribution at the joint contactpressureinthe3Dthreadinterfaceapproaches350 interfaces, the load distribution andstress concentration in MPa which islarger thanthat of2D FEA. Fig. 6(b) gives engagedthreads.The resultsof3DFEA arecomparedwith out the contact pressure in 3D member interfaces whose thoseof 2DFEA and analyticalmethods. contactradiusRisabout3.1timesofthenominalradiusr ofthebolt.Themaximumcontactpressure of 3DFEAis33 3.1 Contactpressuredistributionatjointinterfaces MPawhichislargerthanthat of 2DFEA. The interfacial contact pressure distribution under the preload ofF=15 kNisobtained (see Fig.5andFig.6)by using FE methods. As shown in Fig. 5, the maximum contactpressureinboltheadornutbearinginterfacesisas highas 452 MPa which is tentimes morethan that of the member interface. The contact pressure in the bolt or nut bearingsurfacesshowsconcavenonlinearity becauseofthe stress concentration at the interface edges. The maximum contactpressurenearthefirstengagedthreadis274.6MPa, and decreases with the distance from the nut bearing surface.

Table1. Parameters forloaddistributionin threads

(a) Contactpressurein the thread inteface

35 30 Contactpressure P /MPa 25 20 15 10 5 0 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 Dimensionlesscontactradius R*
(b)Contactpressure distributionin thememberinterface

(a)Contactpressure in thread,nutorbolthead bearingsurfaces

Fig.6. Interfacialcontactpressureof joint1 (3DFEA)

(b)Contactpressure in thememberinterface

Fig.5. 2DFEAcontactpressure at joint interfacesof joint1

3.2 Axial loaddistributioninengaged threads Theloaddistributioninthe boltedjointswithdifferentp, * E andN(seeTable1)areanalyzedwith2DFEA,3DFEA and YAMATOTOs analytical methods. The deformation coefficientskb andkn (inTable1)areobtainedaccordingto

CHINESEJOURNALOFMECHANICALENGINEERING
the geometrical sizes, material properties of joints 14. Thentheconstant(inTable1) andtheloadofeachthread arecalculatedbasedonEq.(1). TheloadofeachengagedthreadFi (istandsforthreadi#, i=1,2,,10,1#,2#,,10#standforthe1sttothe10th engaged thread away from the nut bearing surface) is obtainedandusedtocalculatethecorrespondingloadratio * F *(F =Fi /F). As shown in Table 2,the loadratio on each engaged thread indicates great nonuniformity. In joint 1, the total load ratios on thread 13# in three methods are larger than 65%. In joint 2, the load ratios on thread 1# obtainedfromtheanalyticalmethodand3DFEAdecrease from27.7%and28.7%(injoint1) to 22.8%and 26.0%.By contrast, the load ratio on thread 1# in the 2D FEA increases from 27.8% to 29.2%, which contradicts the result in Ref. [4]. This phenomenon may be explained by the different constrained nut bearing surface mentioned in 2D FEA (see section 2). The axisymmetric assumption make the result of 2D FEA different from those of the analyticalmethodand3DFEA.Afurtheranalysisindicates thatthetotalloadratiosofthreads13#obtainedfromthe analytical method, 2D FEA and 3D FEA decrease from 65.2%,67.5%and66.4%(injoint1)to58.6%,66.0%and 62.6%respectively.SoalowermodularratioE* canmake theloaddistributioninthreadsmoreuniform.Injoint3,the total load ratios of threads 13# obtained from the analytical method, 2D FEA and 3D FEA decrease from 65.2%,66.4%and67.5%(injoint1)to62.6%,63.1%and 61.8%.Theresultsshowthatafinerthreadpitchcanmake the load distribtution of screw threads more uniform. In joint 4, theresults of the analyticaland FE methods show that the load distribution in engaged threads is more nonuniform.Thetotalloadratiosofthreads13#obtained from the analytical method, 2D FEA and 3D FEA are 59.9%, 51.2% and 64.4% respectively, and threads 910# carry load of less than10%. Theresults of2D FEA show more uniform load distribution in engaged threads than those of3D FEAandanalyticalmethods dueto the useof the axisymmetric structure. Compared with the analytical solutions, the 3D FEA results show that the load ratio difference on each engaged thread ranging in 18# is less than2%,and threads910#carrynearlynoloads. The load ratio on each engaged thread in 2D FEA decreaseswhenthethreadisfarawayfromthenutbearing surface, which agrees well with the variation rules of the analytical solutions. While the 3D FEA results show that theloadratioonthread6#isgreaterthanthatofthread5# * * when the jointhas alower modularratioE (E 1/3) ora finer thread pitch p, which coincides well with theresults [4] obtained from CHENs photoelastic test . Compared with the analytical results of the load ratio on each engaged threadinjoints14,the3DFEAand2DFEAresultsshow the largest differences of 3.2% and 6.4% respectively. Generallyspeaking,theresultsof3DFEAareclosertothe analyticalsolutionsthanthoseof2DFEA. The loading and unloading process is considered to

obtainitseffectontheloaddistributioninengagedthreads. The local plastic deformation taking place near the thread root enables the load and load ratio curves to vary nonlinearlyasshowninFigs.79.
Table2. Loadratio F* oneach thread at F=15 kN /%
Method Analytical 2DFEA 3DFEA Method Analytical 2DFEA 3DFEA Method Analytical 2DFEA 3DFEA Method
* Joint1 (p=1.75, E =1, N=6)

1# 27.7 27.8 28.7 1# 22.8 29.2 26.0 1# 25.7 25.5 25.3 1# 2#

2# 21.1 21.3 21.7 2# 19.2 20.6 20.3 2# 20.4 20.3 19.9 3# 4#

3# 16.4 17.3 17.1 3# 16.6 16.2 16.3 3# 16.5 17.3 16.6 5#

4# 13.2 14.2 13.4


*

5# 11.3 11.4 10.3 5# 13.6 11.2 11.7 5# 12.2 12.6 12.0 8# 3.9 6.0 2.4

6# 10.3 8.0 8.8 6# 13.0 9.4 12.0 6# 11.4 9.3 12.1 9# 10# 3.3 5.8 1.0 3.0 3.8 0

Joint 2(p=1.75, E =1/3, N=6) 4# 14.8 13.4 13.7 4# 13.8 15.0 14.1 6# 6.2 7.5 6.5 7# 4.8 6.6 4.2

* Joint 3(p=1.5, E =1, N=6)

* Joint 4(p=1.75, E =1, N=10)

Analytical 26.1 19.4 14.4 10.8 8.1 2DFEA 3DFEA 22.3 16.2 12.7 10.4 8.7 27.6 20.7 16.1 12.4 9.3

18 16 Loadofeachthread F /kN i 14 12 10 8 6 4 2 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN


(a) Loadofeach thread

Loading Unloading

1# 2# 3#

4# 5# 6#

30 Loadratiooneachthread F /% i 25 20 15 10

Loading 1# 2# 3# 4# 5# 6#

Unloading

5 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN
(b)Loadratioon each thread

Fig.7. Axialloaddistributionin threads of joint1 (2DFEA)

Yang Guoqing,etal: ThreeDimensionalFiniteElementAnalysis of the MechanicalPropertiesofHelicalThread Connection

Loadratiooneachthread F /% i

Fig. 7 gives out the load and load ratio on each 2D engagedthreadinjoint1duringtheloadingandunloading process.In2DFEA,eachengagedthreadcarriesloadwith anearlylinearincreasewiththepreloadandtheloadratio on each engaged thread keeps nearly constant during the loadingprocess.However,Theaxialloadofeach2Dthread varies nonlinearly during the unloading process. It is interestingtodiscoverthatthe loadingcurveofthread1#or 2#isabovethecorrespondingunloadingcurve.Whileeach loadingcurveofthreads36#isbelowtheunloadingcurve. So the unloading process enables the load distribution in engaged threads to be more uniform, and may guide the assembly processof threadconnection. 3DFEAisconductedtoobtaintheloadandloadratioof each engaged thread during the loading and unloading process. The corresponding load distribution of engaged threadsinjoints 14 isshowninFigs.810.
20 18 Loadofeachthread F /kN i 16 14 12 10 8 6 4 2 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN
(a) Axialloadof each thread

process,eachloadratioonthreads13#isinitiallyhighat small preload and then decreases, while those of other threads increase with the bolt preload. This may be explainedbytheFEmodelingmethodmentionedinsection 2. The load ratio on each thread varies slightly when the preload rangesin15kN 60kNduringtheloadingprocess.
30 25 20 15 10 Loading Unloading 5 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN
(a) Loadratiosonthreadsofjoint 2

1# 2#

3# 4#

5# 6#

Loading Unloading

1# 2# 3#

4# 5# 6#
35 Loadratiooneachthread F /% i 30 25 20 15 10

Loading Unloading

1# 2#

3# 4#

5# 6#

40 Loadratiooneachthread F /% i 35 30 25 Loading Unloading 1# 2# 3# 4# 5# 6#

5 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN
(b) Load ratiosonthreadsofjoint 3

30 Loadratiooneachthread F /% i 25 20 Loading 15 10 5 0 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN


(c)Loadratiosonthreadsofjoint 4

15 10 5 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN
(b)Loadratioon each thread

20

1# 2#

3# 4#

5# 6#

7# 8#

9# 10#

Unloading

Fig.8. Axialloaddistributionin threads of joint1 (3DFEA)

Compared withthe2D FEAresults, the 3D FEAresults (seeFig.8andFig.9)showthattheloadingcurveofeach thread is closer to the corresponding unloading curve. However,Fig.8andFig.9showthevariationrulesofthe 3D FEAresults whichare similar to those of the 2D FEA results (in Fig. 7).Theloading curve of thread1# or 2# is abovethecorrespondingunloadingone,eachloadingcurve ofthreads310#isbelowthecorrespondingunloadingone. when the preload is less than 15 kN during the loading

Fig.9. Axialloadratiooneach thread in joints 24 (3DFEA)

Comparedwiththeloaddistributioninjoint1(inFig.8), Fig.9aand9bshowlowerloadratiosofthreads1#and2#, but higher load ratios of threads 5# and 6#. The results * showthatthesmallerE andfinerthreadpitchcanimprove the uniformity of the load distribution in threads. Fig .9c shows the nonuniform load ratio on each engaged thread

CHINESEJOURNALOFMECHANICALENGINEERING
inJoint4,thetotalloadratioonthreads810#islessthan 10%.Moreover,theresultsofFig.9aand9bshowthatthe loadratio doesnot always decrease whenthe threadis far away from the nut bearing surface. For example, the load ratio on thread 5# is close to and even larger than that of thread6#. The3D FEArevealsnearly the samerules of axialload distribution in threads as those from YAMATOTOs analytical method as shown in Fig. 10. The load ratio curvesin Fig.10bshowthattheboltpreloadhardlyaffect the load ratio on the enaged thread length l when the bolt preloadranges in30 kN60 kN.In each studied joint, the load ratio on the three engaged threads near the nut interface (l=0.5p3.5p) always keeps higher than 60%. * Lower value of E or finer pitch can make the load distribution in engaged threads more uniform. For joints with thelarge number of engaged threads, the threads far awayfromthenutbearingsurfacecarrynearlynoloads.
100 Loadratioonengagedthreads F /%
*

stressexistsinthethreadrootnearthenutbearingsurface or thefilletunderthebolthead.Thecompressivestressof the members indicates a hollow conical shape. The maximum equivalent (von Mises) stress in threads always liesinthefirst threadrootnearthenutbearingsurface. When the preload is increased to 61 kN, the maximum equivalentandaxialtensilestressgetfarbeyondthetensile strength. As shown in Fig. 13the maximum equivalent reaches 1.11 GPa, and obvious plastic deformation occurs near the first thread root and bolt head fillet. The further increase ofthepreload may resultintensionboltfractureor thread failure. The first thread root is weaker and more vulnerabletotension bolt fracturethanthe bolt head fillet because of the smaller crosssection perpendicular to the boltaxis.
Axialstress z /MPa 383.96 159.84 50.72 2.54 50.81 Equivalentstress e /MPa 0.02 10.64 63.69 114.75 167.87 214.84 268.98 451.11 702.07 (b)Equivalentstress

80 60 40 20 0 0 Joint1 Joint2 Joint3 Joint4

187.69 431.46 579.34 751.33 (a) Axialstress

Fig. 11. Stress distributionof joint1 atF=15kN (2DFEA)


1 2 3 4 5 6 7 8 9 10 11 * Dimensionlessengagedthreadlength l
(a) Analyticalmethod

100 Loadratioonengagedthreads F /%
*

80 60 40 20 70 3.5 0 0 1 4.0 4.5 5.0 85

Joint1 Joint2 80 Joint3 Joint4 75


(a) Axialstress

2 3 4 5 6 7 8 9 10 11 * Dimensionlessengagedthreadlength l
(b) 3DFEA method

Fig.10. Axialloaddistributioninthreadsof joints14

3.3 Stressdistributionpattern The vonMisesequivalentstresse, the axialstressz and thefirst principal stress 1 ofjoints14are analyzed. The stressdistributionsofjoint1areshowninFigs.1113.The results show that the thread root, thread undercut and the fillet radius under the bolt head are prone to local plastic deformationandstressconcentration.Themaximumtensile

(b)Equivalentstress

Fig.12. Stress distributionof joint1 at F=15kN (3DFEA)

Yang Guoqing,etal: ThreeDimensionalFiniteElementAnalysis of the MechanicalPropertiesofHelicalThread Connection

Equivalentstress e /MPa 0.05 54.68 125.34 249.31 498.57 623.19 793.82 897.34 1 122.11

F30kN,andthenincreaseveryslowlyandevendecrease duringthesubsequentloadingprocesswhenF>30kN.The maximumprincipalstressdecreasescontinuously,whilethe maximum equivalent stress decreases at first and then increases during the unloading process because of the generated large compressive stress. Moreover, reasonable loadingandunloadingprocessmaydecreasethemaximum equivalentstressbyup40%atthesameclampingforce.

4 Conclusions
(1) Contact pressure distribution at joint interfaces, axial load distribution and stress concentration in screw threadsareanalyzedquantitativelybyusingrefined3DFE models. The constructed 3D FE model shows great advantages for the full consideration of the geometrical sizes and material properties of helical threads. It can be used to optimize the design and improve the overall performance of threadconnection,especiallythe reliability. (2)Theclearancebetweenthreads,thefilletradiiofbolt headandthreadrootarealsofoundtoaffectthemechanical properties of helical threads, which need to be further studiedinthefuture. (3)Tighteningtheboltwithalargepreloadfirstandthen adjustingtheclampingforcebyunloadingcanimprovethe uniformity of the load distribution in engaged threads, which may guide the assembly process of thread connection. Moreover, the constructed 3D FE model may contribute to the study on thetightening and loosening of boltedjoints. References
[1] ZUODawei,CHENXiumin,SHENGuangxian.Experimentalstudy on theload distribution ofplate mill pressure screwpairs[J].China MechanicalEngineering,2006, 17(3): 307311.(in Chinese) [2] MARINORL,RILEYWF.Optimizingthreadrootcontoursusing photoelastic methods[J].ExperimentalMechanics, 1964, 4(1): 110. [3] DRAGONI E. Effect of thread shape on screw stress concentration by photoelastic measurements[J]. Journal of Offshore Mechanics andArcticEngineering,1994,116(4):228232. [4] CHENHaiping,ZENGPan,FANGGang,etal.Loaddistributionof bolted joint[J]. Journal of Mechanical Engineering, 2010, 46(9): 171178.(in Chinese) [5] YAMATOTO A. The theory and computation of threads connection[M]. Tokoy: Yokendo, 1980. (inJapanese) [6] WANGW,MARSHEKK M.Determinationofloaddistributionina threaded connector with yielding threads[J]. Mechanism and MachineTheory,1996, 31(2): 229244. [7] KIM J, YOON J C, KANG B S. Finite element analysis and modeling of structure with bolted joints[J]. Applied Mathematical Modelling,2007,31(5):895911. [8] YANG Guoqing, HONG Jun, WANG Ning, et al. Member stiffnesses and interface contact characteristics of bolted joints[C]//Proceedings 2011 IEEE International Symposium on AssemblyandManufacturing(ISAM2011),Tampere, Finland, May 2527,2011: 16. [9] NASSAR S A, WU Zhijun, YANG Xianjie. Achieving uniform clamploadingasketedboltedjointsusinganonlinearfiniteelement model[J]. Journal of Pressure Vessel TechnologyTransactions of theASME,2010,132(3):110.

Fig.13. Equivalent stressofjoint1at F=61kN (3DFEA)

Fig. 14 shows the variation rules of the maximum equivalent stress and principal stress during loading and unloading process. The thread pitch and themodular ratio are found to have little effect on the stress concentration occurring at the thread roots. The maximum equivalent stressandprincipalstressof3DFEAincreasemoreslowly with the preload than those of 2D FEAduringthe loading process.The2Dand3DFEAresultsshowalargeresidual equivalent stress after unloading, similar to those in Ref. [12], but is less than the yield strength. The residual equivalentstressof3DFEAislessthanthatof2DFEA.
1,200 MaxvonMisesstress s /MPa e,max 1,000 800 600 400 200 3DFEA_Joint1 3DFEA_Joint2 3DFEA_Joint3 3DFEA_Joint4 2DFEA_Joint1 Loading Unloading

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN

(a) Maximumvon Misesstress

1,800 Maxprincipalstress s /MPa 1,max 1,600 1,400 1,200 1,000 800 600 400 200 3DFEA_Joint1 3DFEA_Joint2 3DFEA_Joint3 3DFEA_Joint4 2DFEA_Joint1 Loading Unloading

0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Boltpreload F /kN
(b)Maximumprincipalstress

Fig.14. Maximumstressin threadsatdifferentpreload

The3D FEAresults show thatthemaximum equivalent stress and principal stress increase nearly linearly when

CHINESEJOURNALOFMECHANICALENGINEERING
[10] OSKOUEI R H, KEIKHOSRAVY M, SOUTIS C. Estimating clamping pressure distribution and stiffness in aircraft boltedjoints by finiteelement analysis[J]. Proceedings of the Institution of Mechanical Engineers Part G: Journal of Aerospace Engineering, 2009,223(7):863871. [11] CHEN Shoujun, LI Qiang, ZHANG Yi, et al. Research on deformation and stress distribution on thread teeth at conic thread connections[J]. China Mechanical Engineering, 2010, 21(17): 2 0442 049.(in Chinese) [12] LIAORidong,SUNYujuan,ZHANGWeizheng.Nonlinearanalysis of axialload and stress distribution for threaded connection[J]. ChineseJournalofMechanicalEngineering,2009,22(6):869875. [13] FUKUOKA T. Finite element analysis of the thermal and mechanicalbehaviorofaboltedjoint[J].JournalofPressureVessel TechnologyTransactionsoftheASME,2005, 127(4):402407. [14] IZUMIS,YOKOYAMAT,IWASAKI A, etal.Threedimensional finite element analysis of tightening and loosening mechanism of threaded fastener[J]. Engineering Failure Analysis, 2005, 12(4): 604615. [15] FUKUOKAT,NOMURAM.Propositionofhelicalthreadmodeling with accurate geometry and finite element analysis[J]. Journal of Pressure Vessel TechnologyTransactions of the ASME, 2008, 130(1):16. [16] FUKUOKATand NOMURAM.Truecrosssectionalareaofscrew threads with helix and root radius geometries taken into consideration[J]. Journal of Pressure Vessel Technology, 2009, 131(2): 024501.1024501.5.

XianJiaotong University, China.He receivedhis master degree onmechatronicsinCentralSouthUniversity,China,in2004.His research interests include mechatronics engineering and digital design. Tel:+8618991843105 Email:guoqing.yang@stu.xjtu.edu.cn HONG Jun, born in 1968, is currently a professor at Xian Jiaotong University, China. He received his PhD degree from Xian Jiaotong Universtiy, China,in 2001. Hisresearch interests includemechatronicsengineeringanddigitaldesign. Tel:+8613709296696Email:jhong@mail.xjtu.edu.cn ZHU Linbo, born in 1985,is currently a PhD candidate at State Key Laboratory for Manufacturing Systems Engineering, Xian JiaotongUniversity,China. Email:chdyuhan@foxmail.com LI Baotong, born in 1982, is currentlya PhD candidateat State Key Laboratory for Manufacturing Systems Engineering, Xian JiaotongUniversity,China. Email: baotong.csu@stu.xjtu.edu.cn XIONG Meihua, bornin1976,is currentlya Lecturerin Hunan UniversityofScienceandTechnology,China. Email:mhxiong@hnust.cn WANGFei,bornin1987,iscurrentlyamastercandidateatState Key Laboratory for Manufacturing Systems Engineering, Xian JiaotongUniversity,China. Email:wangf061126@163.com

Biographicalnotes
YANG Guoqing, born in 1976, is currently a PhD candidate at State Key Laboratory for Manufacturing Systems Engineering,

You might also like