You are on page 1of 16

~

Pergamon

Mech. Mach. Theory Vol.31, No. 2, pp. 229-244, 1996 Copyright 1996ElsevierScienceLtd 0094-114X(95)00041-0 Printed in Great Britain.All rights reserved

0094-114X/96 $15.00 +0.00

D E T E R M I N A T I O N OF LOAD D I S T R I B U T I O N IN A T H R E A D E D C O N N E C T O R WITH Y I E L D I N G T H R E A D S W. WANG


Department of Mechanical Engineering, National Yun Lin Engineering and Technical School, No. 64, Wen-Hwa Road, Hoo-Wei Town, Yun Lin, Taiwan, Republic of China

K. M. MARSHEK
Department of Mechanical Engineering, The University of Texas at Austin, Austin, TX 78712, U.S.A.
(Received 4 April 199l; in revised form 2 February 1995; received for publication 22 May 1995)

Abstract--Amodified spring model was developed to predict the load distribution in the threaded portion
of a connector having yielded threads. Distinction was made between a compression case (nut and bolt) and a tension case (turnbuckle). The load distribution was described using simple second order difference equations and linear first order constraints. The load distributions for elastic and yielded threaded connectors are compared, and a discussion of how the yielded threads affect the load distribution is given.

NOMENCLATURE A, B, C, D---constant coefficients of the finite difference equations b--length of thread per revolution at the clamped end L--effective thread height F---external applied stud load h--thread width at the clamped end K~--axial stiffness of body section i between threads K~c--axial stiffness of stud section i between threads K~---composite spring constant of thread pair i L,--load on the body section i n--total number of active threads P,--load on the thread i, axial Q--first moment of the cross section at the clamped end of the thread r---diameter at base of thread Si--load on the stud section i u~--absolute deflection of stud thread ring i I/--shear force w--uniformly distributed load on thread ~, 8, 7--constant coefficients 6~--relative axial deflection of the springs with constant K~ 6~--relative axial deflection of the springs with constant K~ 6~---relative axial deflection of the springs with constant K~ a~----equivalentstress av--yield stress, tension z,.y--shear stress in the x - y plane a,.--tensile or compressive stress in the x direction

INTRODUCTION This p a p e r develops a m e t h o d for d e t e r m i n i n g the load d i s t r i b u t i o n in a yielding threaded c o n n e c t i o n . F o r static a n d fatigue loading, it is i m p o r t a n t to be able to accurately predict the local load increase (when some threads are yielded) in order to determine c o n n e c t o r load capacity. The analysis procedure developed follows from the finite difference m e t h o d s presented by Miller [1] for the e v a l u a t i o n of load d i s t r i b u t i o n in a n o n - y i e l d i n g threaded connection. Previous researchers have studied the d i s t r i b u t i o n o f load a l o n g the threads o f threaded c o n n e c t o r s [2-17]. T h e first theoretical m e t h o d that included radial expansion of the n u t was presented by Sopwith [2]. He deduced that the d i s t r i b u t i o n o f load a l o n g the length o f a n u t was n o t u n i f o r m due to the strains set up in the bolt a n d n u t while u n d e r load. By analyzing the strains, 229

230

w. Wang and K. M. Marshek

Sopwith obtained the load distribution along the thread helix. The various strains, excluding the axial tension in the bolt and compression in the nut, can be divided into three types: (1) bending deflection in the threads of the bolt and nut, (2) axial recession due to compression in the threads of the bolt and nut, and (3) expansion of the nut caused by radial pressure between the bolt and nut. His method can be utilized for both the bolt and nut case and the turnbuckle case with symmetrical triangular thread forms. Sopwith's technique is also capable of analyzing systems with components of dissimilar materials. Yazawa and Hongo [3] derived an expression for the distribution of load in the screw thread of a bolt-nut connection that was subjected not only to axial forces but also simultaneously to tangential forces and bending moments. In their analysis, they first assumed that the distribution of load in a screw thread is a polynomial function of turning angle as the variable with undetermined coefficients. Next, they used the polynomial function to calculate the resultant axial force, resultant tangential force, bending moment at the bearing surface of the bolt head, deflection and deflection angle of the bolt at a distance from the bolt head and then decide the undetermined coefficients in the postulated polynomial by a linear programming method. They showed a good match between calculated and experimental axial load distributions. Using an extensometer on the outside of a nut, Goodier [4] investigated the distribution of thread loads and deformations. He showed that the load distribution was governed by a number of mechanisms: stretching of the bolt, compression in the nut, bending of the threads, and circumferential stretching. Hetenyi [5] investigated a variety of nut designs using a threedimensional "stress freezing' photoelastic method. He found that by applying either a taper-thread nut or a nut with a tapered body, the fastener strength could be increased by 30%. While using a similar three-dimensional photoelastic method, Allison [6] modeled buttress threads in a turnbuckle coupling. He determined that the inclusion of tapered sections does not offer any advantages in terms of load sharing or reduction in critical fillet stress for the configurations tested. The distribution of load obtained from the tests showed peaks at either end of the engaged threads. Stoeckly and Macke [7] developed a testing apparatus that actually measured the axial displacement of the threads. The axial thread displacements were converted to axial thread loads by coefficients determined from single-and triple-thread tests. The tests were performed on (1) a variety of tapered, linearly varying pitch diameters, and (2) straight, constant pitch diameter, American National eight-pitch threads, cut on 2 inch steel bolts. In addition to these tests, Stoeckly and Macke modified the Sopwith equations to accommodate tapered screw-thread connections in order to compare their experimental data with Sopwith's. They concluded that the coefficient of friction between the lubricated bolt and nut has little effect on the thread load distribution and, the maximum stress in a bolt for a given load is reduced by choosing the proper thread taper. Utilizing the data and techniques for load distribution, Heywood [8, 9] developed methods for calculating thread fillet stress, joint efficiency, and load carrying capacity. He also made significant advances in the fatigue evaluation of threaded connections. He showed that determination and manipulation of load distribution played an important role in computing cyclic and static load capacity. Kenny and Patterson [10] used a frozen-stress photoelastic analysis employing a fringemultiplying polariscope in conjunction with a recording microdensitometer to find the load distribution and the normalized stress distribution in the threads of an ISO metric nut and bolt. The results of the method correlate closely with Sopwith's theoretical load distribution and with the finite element approach used by Bretl and Cook [12]. More recent investigations into the distribution of load in threaded connections have involved the use of finite element methods. O'Hara [11] for example, used two-dimensional axisymmetric models of hypothetical thread profiles to compare his analysis with Heywood's. The thread projections were based on characteristic lengths and not on the dimensions of the real components. There were indications of general agreement with Heywood's equations, except for the root load. The larger error at the root was attributed to the use of a point load rather than a distributed load. Bretl and Cook[12] developed a unique finite element modeling technique for threaded connections. Instead of modeling individual threads, the thread zone is replaced by a layer of elements with orthotropic properties. The principal directions of orthotropy depend upon the thread geometry and upon the direction of load. Both tapered and conventional threads were

Load distribution in a threaded connector

231

considered on a nut and bolt configuration where the nut is in compression and the bolt in tension. Bretl and Cook's numerical results for conventional and tapered threads agreed well with theoretical and experimental results in the literature. Tanaka [13, 14] introduced a modified two-dimensional finite element method to analyze such cases as threaded connections subjected to a transverse displacement and flange coupling under an arbitrary type of loading for the typical examples of the threaded connections subjected to an external load. He concluded the following: (1) In the case with fastened plates of equal thickness, he found that a hexagon socket head cap bolt is the hardest to separate, followed by stud, through bolt and tap bolt in this order. (2) The axial tension is not linearly related to the external load. The ratio of the axial tension increase to the external load increase is not constant, and its value changes with the amount of external load. In some cases, the value of the force ratio becomes negative. (3) In the case of a T-flange, the effect of flange thickness upon the force ratio is greater than that of the bolt hole location. Fukuoka et al. [15] used an axisymmetric finite element method with four types of models that include the effects of friction on two contact surfaces between threads of the bolt and nut, and between fastened plate and nut. They concluded that the friction coefficient on the bearing surface of the nut has a more dominant effect than on the pressure flank because of the stress concentration at the root of the bolt. Newport and Glinka [16] investigated the effect of changing a dimension of the thread geometry on the maximum local stress occurring in a threaded tether connection. They used a previously validated numerical technique and presented the results for changes in thread fillet radius, pitch, wall thickness and the number of engaged threads. They derived a parametric equation for the stress concentration factor in tubular threaded connections based on their results. They also compared the theoretical predictions of local stress based on their equation with Fessler's photoelastic results. The paper concluded that in order to reduce the critical local stress in a conventional tubular threaded connection: (a) the thread fillet radius should be as large as possible, (b) the pitch should be as large as possible, and (c) the wall thickness should be above a critical minimum. The parametric equation allowed for the use of a stress concentration factor in threaded connections. This equation has been tested against published photoelastic results and has been shown to be slightly conservative. Crum [17], in the most recent work on the distribution of load in threaded connections, made an unproved assumption that the thread displacements are in direct ratio to their distance from the external loading point. The effects of vessel geometry changes, temperature effects, seal movement, and thread imperfections are studied. In the present paper we give a review of spring models and the development of the finite difference equations used to predict the load distribution along the threaded portions of a bolt and nut connection and a pipe union connection. Detailed description of a spring model for a yielding fastener is also included. This paper compares finite difference load distribution results for non-yielding and yielding fasteners. The results of these comparisons are graphically presented. The effect of yielding on the performance of threaded fasteners for both a compression and a tension case will be discussed. MODEL DEVELOPMENT The threaded connector is comprised of two major components: the internally threaded member (body) and the externally threaded member (stud). Physical models in Figs 1 and 2 represent two opposite loading patterns normally encountered by threaded components. Two models of threaded connectors, classified as tension or compression, were developed to analyze the behavior of threaded connectors by Miller [1]. His models consist of an assembly of springs, simulating elements of the threaded connectors. The interactions of these elements lead to the development of a second order difference equation. In order to solve more complicated problems, we have to derive a generalized expression for the second order difference equation, Locally imposed linear constraint together with a conservative yielding criterion and Crout's elimination method will be applied to solve for the load distribution in a yielding connector.

232

W. Wang and K. M. Marshek

Fig. 1. Compression case: nut and bolt,


Model L Compression case

Fig. 2. Tension case: body and stud.

Model I will represent the compressison case of the threaded connector. A diagram of the generalized spring model for the compression case is shown in Fig. 3. The external load F, as shown, is distributed into the threads inducing compression in the body sections between threads, thus the name "compression case." In the spring model, there are springs for the axial stiffness of the stud's root area with constant K~ and for the axial stiffness of the body with constant K~:. The spring with constant K~r is a composite spring which combines in series the axial stiffness of the threads in bending, shear and AXISOFSYMMETRY

FIRSTACTIVETHREAD OF STUD

LASTACTIVETHREAD OF STUD

/
SO S1
Si-I

STUD
Si Si Sn-2 Sn-1
Sn-I Sn

:_.__LAAA -L~I ~LAA~ -L~i tlA ~-~A_.~.I_~AA ~L.~_L_AAAA 9-,

;
;

,,Vt-IV,,,-

,VVI- VVVt-VVVIBODY

[AAA bll

V,,l-/

/~_._ FIRSTACTIVETHREAD
OF BODY

LASTACTIVE THREAD
OF BODY

Fig. 3. Model 1--compression case

Load distribution in a threaded connector

233

the radial stiffness of the body and stud sections. See [18] for the development of the equations for the spring constants. The load Pi represents the load on the i th thread. The load 5:,.is the load on the section between the i and i + 1 thread. The load L~ is the load on body section i. The threads are numbered from the first to the last thread. From force equilibrium
P,=S,_,-&

l~<l~<n
i __ i

(1)

The thread deflections are included in the composite spring constant K T - PilOT, where 3!r is the combined axial component of deflection due to load P~. It is assumed that the load P is applied at the mid-height of the thread. The effective body section i between the thread i - 1 and i has a spring constant KL = Li/6i~, where 6L is the axial deflection due to load Li. The first active body section which is between the fixed end and the first body thread sees the sum of the load Pj transferred into the threads. Stiffness of the stud sections between threads are simulated by the spring constant K~c = &/gL, where 6L is the elongation of the section i due to load & (Fig. 4). The deflection of the k th thread ring is given by
k

u~=a~= y, a~:
j=l
i Us-U~+ I i = a~

1 .<k .<,,

(2)

and the elongation of the stud section between any two consecutive rings is
l <. i <~ n - 1
(3)

Substituting the expression for u~ from equation (2) into equation (3) yields

6~--6~+'--6~'=6~
Pi
K~-

l<~i~n--I

(4)

Replacing the deflections in equation (4) by their equivalent spring constant ratios produces

Pi+I
K~r+'

Li+ I - Si
K~' K~

(5)

The load L k can be written as

Lk= ~ S',
j=k

u~

Ki

.__q

uis+i

Si-I

Si+l

STUD

,----,
/ "1
KI~

I /,---,
\ ~"
Kib~l

I
\ ~'
BODY

/"~1

VVV Fig. 4. Thread sections i and i + I of compression case.


MMT 31/2--H

234

W. Wang and K. M. Marshek

When Pj from equation (1) is substituted into the above equation, the load Lk becomes
L k = S k - | -- Sn

(6)

where S. is the load in the stud section n beyond the last active stud thread. For compatibility with the boundary conditions of this system, the S. term must equal zero. Equation (6) becomes L~ = Sk_l (7)

Substituting the expressions for P and L from equations (1) and (7) into equation (5) yields
Si-i--Si

I 1 + v---7~ -~Tr"i+ K ~ ] + S i + l ~K~ = 0 .K'T + v---7~ -~ i+ K Ti/ K T

(8)

Replacing the constant coefficient terms in the bracket by ]/i and order difference equation of the form
S i - i -- ~iSi -~- ~iSi+ 1 ~ 0

by g~, yields a second (9)

From equation (9) we have n - 1 simultaneous equations with n - 1 linear independent unknowns. A modified Gauss elimination method, known as Crout's elimination method is applied here to solve the (n - 1) by (n - 1) system of equations.
M o d e l H. Tension case

The spring model for the tension case is shown in Fig. 5. The difference between this model and the compression model is the location of fixity. In the tension model, the fixity location is near the last active body thread. With the direction of the applied load F as shown, the loads transferred into the body threads would induce tension in the sections between threads, thus the name "tension case." Due to the similarity between the two models, the compression model description will adequately describe both configurations. However, the first active body section in the tension model follows the first active body thread. In the compression case the first active body section comes before the
AXIS OF SYMMETRY

FIRST ACTIVETHREAD OF STUD

LAST ACTIVETHREAD OF STUD /

STUD

So

/ / /

'
f 1 1 /

FIRST ACTIVETHREAD OF BODY Fig. 5. Model II--tension case.

LAST ACTIVETHREAD OF BODY

Load distribution in a threaded connector

235

first active body thread. These differences will be accounted for in the mathematical model and are highlighted below. The first difference is the determination of load L. As in the compression model, the load L is the combination of loads transmitted into the body sections in the axial direction, However, in the tension model, the first thread load P is carried through each successive body section. The last body section (see Fig. 5) carries the sum of the loads transferred into the threads The load Lk is given by
Lk = Sk_, -- S. Substitution of equation (1) into equation (10) yields (10)

Lk = So - Sk

(ll)

The absolute deflection us o f a ring section for the tension case is a combination of local thread deflection and the sum of the section deflections following this location (Fig. 6). Or, in mathematical terms
k Us

6 ~ + ~ 6{.
j=k

(12)

Combining equation (12) with equation (3) produces


6~--6~+'+6~ ~-rs~ (13)

Replacing the 6 variables with their equivalent load to spring constant ratios, equation (13) becomes Pi Pi + l

Li

Si

(14)

Elimination of the variables P and L using equation (1) and equation (11) leads to an equation of the form

K~-. +-~-+-~-?-/-e "'T , T| -- S S,_,--S, 1 +~v+l ,+, K,-----~ = T _S0~_~


*xbc l~-sc I T Kbc

r,~i

Ki7

Ki

K~

(15)

Ki-I SC Si-1

~.q
si

U~

Ksc

.--q

ui+l
Si+l

STUD

-+'

/ "1

\
K~c
Li

/''1

\
tri+ 1

BODY

Fig. 6. Thread sections i and i + 1 of tension case.

236

w. Wang and K. M. Marshek

Indexing this equation by one and letting t, ,t and y equal the coefficients yields
S,_, - fl, S, + ~,S, +, = - ?,S0

(16)

The boundary conditions for S are given as


So= F

(17)
(18)

sn = o

where So and Sn are defined in Fig. 5. Again, as in the compression case we can solve the (n - 1) unknown Si by Crout's elimination method.
Modified spring model for yielding threads

In this section, we will develop a method to determine the load distribution in a connector where the threads deform in an ideally elastic, perfectly plastic, manner. Elasto-plastic behavior is characterized by an initial elastic material response to an external force followed by plastic deformation after a certain higher value of force is reached. Figure 7 shows an idealized stress-strain curve for a ductile engineering material. The material initially deforms according to the elastic modulus, E, until the stress level reaches a value av, the uniaxial yield stress. By increasing the load further, the material exhibits a linear strain-hardening characterized by the tangent modulus, Er. For an ideally elastic, perfectly plastic material, the tangent modulus Er is equal to zero, and for this material, the strain hardening parameter h which is defined as h = dtr/dcp = ET/(1 -- E~/E) is also zero. In the following analysis, we will assume that the i th thread is yielded. Since the yielded thread deflection depends on the displacements of the adjacent threads, to avoid including the yielded thread deflection in our derivation, we will use the deflection at stud thread i - 1 and i + 1. In what follows, we first develop the equation to determine the load distribution for the compression case and then we will develop the equations for the tension case. (a) Compression case:
i-I

(20)
i+1

+'

+ ' + jE =l
., ,5~-

(21)

Subtracting equation (21) from equation (20) gives


= -~,~ ,,~ + -

(22)

Slress, Ge

60

Elastoplastic behavior slope, ET Perfectly plastic behavior ET=0 ~ lastic behavior Slope, E

Strain, E Fig. 7. Elastic, linear strain-hardening stress-strain behaviour.

L o a d distribution in a threaded connector but with

237

u~- 1 -

/,/~+ I __ - f i xi - 1 +

i [derived from equation (3)], equation (22) becomes 6~

~ - J ~ ' + fi~r-' - ~r +' = &~-l + 6L

(23)

Replacing the deflections in equation (23) with their equivalent spring constant ratios produces
Li Li + i Pi - ! Pi + I Si - 1 Si

K~

K~+K~

'-

K~r+'= ~---7-si-I +K-7~K~

(24)

The load Lk can be written as


L, = Sk_ l

Substituting the expressions for P and L from equation (1) and (7) into equation (24) yields
S,_2 + A,S,_, + B,S, + C,S,+, = D,

(25)

where
Ai=-K~-

,,I-1_ 1 K'~] [~-~ + K ~ +


i i[| l

(26)

K~r-' Ci = K~+ I Di = 0 With the constraint equation Si_ ~- Si = yield limit of thread = YLMT,

(28) (29)

(30)

we can solve for the load distribution in a connector with yielding threads using equations (25)-(30). If n adjacent threads are yielded, then the generalized form of equation (25) is

-S and the n linear constraint equations are

S~_ t - S; = yield limit of thread i, S~ - Si+ ~= yield limit of thread i,

S~+,_2 - S;+n_ ~= yield limit of thread i + n - 1.

(32)

Equations (31) and (32) give the load distribution for the compression case having yielded threads. Next we develop the equations for the tension case. (b) Tension case: Following the same procedure used for the compression case gives
Si_ 2 + A~Si_ t + B~Si + C~S~+~= Di

(25)

where
= i i1- 1 +K---~+ 1 1

(33) (34)

238

W. Wang and K. M. Marshek K~r- t


Ci = Kit+ l

(35)
(36)

D i = --S O

q-

and the linear constraint equations


Si-- 1 -- Si =

yield limit

of thread i

(37)

If n adjacent threads have yielded, the generalized form of equation (25) becomes S 1 S x I ~ + 1 1 1 1 -] SF1 1-] =
. . . . .

1 +...+

b=2xr

Y
V =wb V

w = distributed line load

y = 0.0042 in., where the

yielding begins Y Fig. 8. A thread model under a transverse load V and an axial load V tan ~.

Load distribution in a threaded connector a n d the n l i n e a r c o n s t r a i n t e q u a t i o n s are Si_ j - Si = yield limit o f t h r e a d i, Si - S~+ t = yield limit o f t h r e a d i + 1,

239

S~+,_2 - Sin_ z = yield limit o f t h r e a d i + n - 1. Equations threads. (38) and (39) give the load distribution for the t e n s i o n case with

(39) yielded

1. 2. 3. 4.

INPUT Compression case or tension case Fastener material properties Boundary conditions Yield limit (YLMT) I Calculate if,i, 13i,~ --~ Si-1 - ~iSi + ff.iZi+l= "~

Construct the finite difference equations then solve for S's

--1,

Yes Si > YLMT No

Set Si-1 - Si = YLMT Calculate Ai, Bi, Ci, and Di...

Next i

Build a new stiffness matrix

Fig. 9. Flow chart for the calculation of load distribution in the threaded fastener with yielding threads.

240

w. Wang and K. M. Marshek

Conservative yield criterion In the spring model analysis, the thread is modeled as a short beam. The deflection of the short beam is due to the bending m o m e n t and the shear force. Unlike the long beam, bending stresses are not dominant and shear stresses must also be included in the analysis. We applied a conservative Tresca yield criterion to determine a yield limit for the threads, and a thread is considered to have failed when the combined stress exceeds the yield limit. Figure 8 shows a cantilever thread of rectangular cross section having width b, depth h, length I, and subjected to a force at the free end. The force can be divided into two components; a transverse shear V and a normal force V tan q~, where ~b is the slope angle of the thread face. The Tresca yield criterion is applied to determine if the thread has yielded or not. The Tresca yield criterion is taken to be
2 2 2 fix + 4Zxy = lYe

(40)

where G is the equivalent stress. When the thread begins to yield. In equation and zxy is the shear stress in the x - y Since the m a x i m u m effective stress can be written as

G equals lyy (the yield stress in pure tension or compression), (40), G is the tensile or compressive stress in the x direction, plane. is found in the clamped end of the thread, the shear stress

*xY= Ib =-~3 where V= w= b= Q = I= shear force = wd, uniformly distributed load, length of thread per revolution = diameter at base of a thread times rr, first m o m e n t of the thread's cross section with respect to the neutral Z axis, m o m e n t of inertia of the thread's cross section. 40

--

Photoelastic Theoretical FEM Spring Model

30

--

.o
.to
t~

e-

20

o,

10

Thread, I

Fig. 10. Compression case: comparison of photoelastic, theoretical, finite clement, and spring model of a 1 inch diameter steel bolt and nut with 8 engaged Whitworth threads (8 threads/inch).

Load distribution in a threaded connector The normal stress on the cross section of the thread at the clamped end can be written as

241

VLy
O'x = O'x due to the bending m . . . . t + O'x due to th . . . . pressive force - - T "[

V 12VLy V bh - ~ "t bh

(42)

Substituting equations (40) and (41) into equation (42) gives

a-L J2-F12VLy +
ay

V-]2 + 4 [6V

(43)

To locate the initial yield point, we take the partial derivative of equation (43) with respect to y 0 = Otre dy where (44)

Oae [-24VL'][-12VLy

V]
-1

6~6VTFh2

L JL -y

] (2y)=0 (45)

For a l i n c h diameter bolt with Whitworth threads, L = 0 . 0 4 2 0 2 i n c h (1.0673mm), b = 2.6136inch (66.385mm), and h = 0.10417 inch (2.6459 mm). Substituting these values into equation (45) and solving for y in inches gives y - 0.0454, 0.004215, and -0.00496. When y = 0.004215 inch (0.1071 mm), equation (43) gives a maximum value for fie, and the thread will begin yielding at 0.004215 inch from the neutral axis of the root cross section. Replacing tre by a0 ( = 30,000 psi or 207 Pa for steel) in equation (43), we find the maximum external load V that a bolt thread can resist is 2665 lb (11,883 N), and the maximum external load that a nut thread can resist is 3203 lb (14,282 N). Because the nut threads have a higher yield limit, yielding can occur in the bolt thread and not in the nut thread. Therefore, we set the yield limit (YLMT) equal to 2665 lb in the spring model analysis, the results of which follow.

25
Theoretical

20

FEM

15
0 .-,= ,.Q

"O

-o
0 ._J

10

0 0

Thread, i

Fig. 11. Tension case: comparison of theoretical, finite element, and spring model of a 1 inch steel stud and body with 8 engaged UNC threads (8 threads/inch).

242

w. Wang and K. M. Marshek

An algorithm useful in determining the load distribution for the case of yielding threaded connectors is shown in Fig. 9. R E S U L T S AND D I S C U S S I O N S In this section, we present and interpret the results for the yielding fastener. The elastic spring model results are shown in Figs 10 and 11: (a) for a 1 inch diameter steel bolt and a nut with 8 engaged Whitworth threads (8 threads/in.) for the compression case, and (b) for a 1 inch diameter steel bolt and nut with 8 engaged UNC threads (8 threads/in.) for the tension case. A two-dimensional axisymmetric contact finite element program [18] was developed to verify the spring model results. In the compression case, a three-dimensional "stress freezing" method used by Hetenyi [3] to investigate nut and bolt configurations is utilized for comparison with the spring model. The compression case is also evaluated using the theoretical method developed by Sopwith [2]. Results for the photoelastic model, Sopwith's model, the spring model, and the finite element model for the same threaded connection are shown in Fig. 10. The load distribution results for each analysis are plotted as "percentage of the total load" transmitted in each thread versus "thread number, i". The results from the spring model and finite element analysis compare well in this case, but the finite element method predicts a slightly higher load on the first thread. Figure 11 shows the results for the finite element analysis, Sopwith's analysis, and the spring model analysis for the tension case. For the yielding thread analysis, the Whitworth thread (8 threads/in.) is used for both the tension and compression case in the spring model analysis. By conservatively assuming that the thread will fail when the yield begins, we can compute the load to yield for each thread. When the load on a thread is over yielding limit, a set of finite difference equations and linear constraints are generated for determining the load distributions for this condition. To observe how each thread responds, we increase the external load until the first five threads are yielded. The results are shown in Figs 12 and 13, where the two horizontal lines represent respectively the loading limit for the nut thread and the bolt thread. It is apparent that the
4000

3000
8
tO

YLMT (Bolt) YLMT (Nut) F = 8700 Ib F =115001b

= 2000

..Q 13 "0 0

::3

o,

--

F =14000 Ib

r_ 1000 ,L

F =16200 Ib F =18500 Ib

Thread, i

Fig. 12. Compressioncase: load distribution after some threads are yielded.

Load distribution in a threaded connector


4000

243

YLMT (Bolt)

3000
4 YLMT (Nut) F = 12700 Ib F =16000 Ib

o~ =O ta

2000

"O "O t~ O _.d

--

F =18670 Ib F =18690 Ib

1000
& F =20500 Ib

0 0 3 4 5 6 7 8 9

Thread, i

Fig. 13. Tension case: load distribution after some threads are yielded. nut threads have a higher yield limit than the bolt threads, so failure will occur in the bolt threads only. For the compression case (see Fig. 12), the first five threads yield when the fastener is under 8700, 11,500, 14,000, 16,200, and 18,500 lb, respectively. Since the lower number threads always carry a higher load, once these threads are yielded, the thread that is next to them will carry a greater load as the external force increases. Therefore, the thread yielding order follows the thread number order exactly. But for the tension case (see Fig. 13), only thread 1 is yielded when the turnbuckle is under a 12,700 lb force, and the first five threads will not yield until the external force reaches 20,500 lb, which is 2000 lb higher than for the same condition as in the compression case. The thread yielding order is thread 1, 2, 3, 8, then 4, where thread 3 and thread 8 are yielded almost simultaneously. CONCLUSIONS Procedures have been presented for determining the load distribution in connectors having yielded threads for the compression and tension cases. The load distribution results from the spring model analysis compare well with theoretical, experimental and numerical results for a uniform threaded connector. In modeling the fasteners with yielding threads, we show that yielding starts from the first active thread then spreads throughout the contacting threads for the compression case. The yielding starts from the end threads then moves to the inside threads for the tension case. We also show that the turnbuckle connector (tension case) can resist a higher load without yielding than the nut and bolt connector (compression case).
REFERENCES

1. D. L. Miller, Mech. Mach. Theory 18, 421-430 (1983). 2. D. G. Sopwith, Institution o f Mechanical Engineers, Proc. 159, 373-383, 319-398 (1948). 3. S. Yazawa and K. Hongo, J S M E Int. J. Series I 31, 174--180(1988). 4. J. N. Goodier, Trans. A S M E 62, AI0-A16 (1940). 5. M. Hetenyi, Trans. A S M E 65, A93-A100 (1943). 6. I. M. Allison, Congress on Material Testing, Seventh Lecture, Budapest, Hung, Vol. 2, pp. 413-416 (1978).

244 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

W. Wang and K. M. Marshek E. E. Stoeckly and H. J. Macke, Trans. ASME 74, 103-112 (1952). R. B. Heywood, Designing By Photoelasticity. Chapman & Hall (1952). R. B. Heywood, Designing Against Fatigue of Metals. Reinhold (1962). B. Kenny and E. A. Patterson, Experimental Mechanics, pp. 208-213 (September 1985). P. O'Hara, Proc. of the Army Symposium on Solid Mechanics, Army Materials and Mechanics Research Center, AMMRC MS 74-8 pp. 99-119. (1974). J. L. Bretl and R. D. Cook, Int. J. Numer. Methods Engng 14, 1359-1377 (1979). M. Tanaka and A. Yamada, Bull. JSME 29, 617-624 (February 1986). M. Tanaka, K. Hongo and E. Asaba, Bull. JSME 25, 291-298 (February 1982). T. Fukuoka, N. Yamasaki, H. Kitagawa and M. Hamada, Bull. JSME 29, 3275-3279 (October 1986). A. Newport and G. Glinka, Proc. of the Int. Offshort Mechanics and Arctic Engineering Syrup. 7th, Vol. 1, pp. 141-147. ASME, New York, (1988). A. S. D. Crum, ASME, Pressure Vessels and Piping Division, Vol. 148, pp. 43-53. ASME, New York (1988). W. Wang, PhD Dissertation, The Univ. of Texas at Austin (1991).

You might also like