J. Construct. Steel Res. Vol. 43, Nos. 1-3, pp.

119-140, 1997
© 1997 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0143-974X(97)00016-3 0143-974X/97 $17.00 + 0.00
ELSEVIER

Design of Bolted Endplate Connections

P. C. Olsen

Colberg Consult, Fasanvaenget 124, 2980 Kokkedal, DK-Denmark

(Received 25 January 1996; revised version received 3 January 1997;

accepted 23 January 1997)

ABSTRACT

Design formulae for bolted flush and extended endplates are presented based
on the theory of plasticity. The flush endplate is designed such that prying
forces are not present. The bolts of the extended endplate are designed such
that yielding at the boltline and at the flange can occur, thus reducing the
endplate thickness but increasing the required bolt force capacity. An explicit
expression for the prying ratio is presented. The effect of using several bolts
alongside the web is accounted for. The design formulae for flush and
extended endplates are channelled into one general set of design formulae.
To measure the adequacy of this simplified design method, comprehensive
numerical analyses of the bolt/endplate connection have been performed. The
numerical method is presented briefly. The agreement between the two
methods is excellent, with only minor differences in the ultimate load bearing
capacities. © 1997 Elsevier Science Ltd.

1 INTRODUCTION

Great attention has been drawn to the design of bolted endplate connections.
Primarily, the forces in the bolts were objects of great concern, whereas only
little attention was given to the design of the endplate. Although very compli-
cated moment distributions are present in the endplate, the endplate is mostly
designed based on simple T-stub models with yielding at the root of the T
[1] and possibly at the bolt line [2,3]. In the annex to EC3 [4], the design of
bolted endplates is based entirely on the model of equivalent T-stubs. How-
ever, the methods presented in Eurocode No. 3 [4] are applicable only for
connections with one line of bolts on each side of the web or, alternatively,
one line of bolts on each side of the tension flange. In addition the regulations

119
120 P. C. Olsen

are directly applicable only for small axial forces. For such simple cases, the
T-stub model seems to be adequate, although the moment distribution at the
junction between web and flange is very complex.
It is very important to consider the connection bolts/endplate in its entirety.
The dimension of the endplate influences the bolt forces, i.e. a very thick
endplate results in smaller bolt forces, whereas prying forces are introduced
for a thinner endplate, resulting in larger bolt forces. The existence of prying
forces is a matter of to what extent the endplate makes contact with the
resisting media, i.e. an opposite endplate for a beam splice or a flange for a
beam-column connection.
From this introductory discussion, it is obvious that more general methods
of analysis and design of bolted endplates, in particular methods physically
more comprehensive and suitable for computer implementation, are desirable.
In Gebbeken et al. [5] results of numerical computations using the finite
element method and the theory of elasticity are presented. However, the prob-
lem of contact between surfaces is very complex in the theory of elasticity
and requires great computational effort.
In this paper a general approach of analysis and design of bolted endplate
connections, based on the structural laws of plasticity, is presented. Both end-
plate and bolts must thus possess sufficient capacity of plasticity. The upper
bound theory is applied, i.e. the yield line theory is applied for the endplate.
Contrary to classical yield line theory, where a single parameter determines
the entire mechanism, the endplate is subdivided into triangular elements, the
displacements of each vertex of the mesh representing a partial mechanism.
The proper mechanism is determined by means of solving a linearly con-
strained optimization problem. The contact problem is simply solved by
adding additional constraints regarding the displacements at the vertices; this
does not increase the complexity. The optimization problem is solved by
means of a very efficient method developed by the author. This newly
developed method has a response time so minimal as to enable interactive
design.
One may utilize this methodology in one of two ways, either in terms of
a small variety of possible yieldline patterns, selected by engineering judge-
ment, with few parameters determining the mechanism, or in terms of applying
a very large number of elements, in which case the necessity of engineering
judgement becomes superfluous and whereby a very accurate estimation of
the bearing capacity is ensured. In the first case, the response time of the
methodology is comparable with that of applying a simple analytical solution
expressed in terms of a formula, whereas the response time in the second case,
especially in a PC-environment, unfortunately is unacceptable. The duration of
the analyses of the examples presented in this paper was typically 3-10 min,
which must be compared to an acceptable response time of maximum 30 s.
 ' Design of bolted endplate connections 121

The conclusion drawn was to develop analytical design aids by means of

simple yieldline patterns and to test the results against the elaborate and accur-
ate numerical method of using a very large number of elements. This must
be seen in context with the requirement in EC3 of having to demonstrate the
adequacy of the model by means of physical tests when the yieldline method
is used. The adequacy of the analytical solutions presented in this paper has
been proved by means of an accurate numerical method.

2 THE NUMERICAL MODEL OF ENDPLATE AND BOLTS

It is well known [6] that the ultimate load factor of the upper bound theory
may be determined by means of solving linear optimization problems. Formu-
lation of the yield line theory as a problem of linear optimization is
accomplished by subdividing the plate under consideration into a number of
triangular elements as shown in Fig. 1, and applying the principle of virtual
work and the upper bound theorem. Subdivision of the domain into triangles
is advantageous in the sense that all partial mechanisms automatically are
geometrically admissible.
Displacements and relative rotations are assigned to all vertices and all
triangular edges, respectively, in the subdivided domain. All displacements
are collected in a vector u and all rotations in a vector 0. Compatibility
requires the following equation to be satisfied:

0 = Bu (1)

where B is a rectangular matrix with the number of rows corresponding to

the number of edges and the number of columns corresponding to the number
of vertices. Writing the rotations as the difference between two positive num-
bers, eqn (1) can be rewritten accordingly as:

0 + -- 0 - =Bu0 + ---00- -->0. (2)

Fig. 1. Subdivision of part of an endplate around the boltholes into triangular elements.
122 P. C. Olsen

The intemal work is determined as the product of the yield moments and
the rotations. Letting m + and in - represent the yield moments, the internal
work is determined from eqn (3):

Wintern = m + 0 + +m-0-. (3)

The extemal work is principally determined as the product of the extemal

forces and the displacements. This can be expressed as a linear equation, as
shown in eqn (4):

Wextern • C u (4)

where the components of C represent the external work for unit displacements.
By means of the upper bound theorem, the ultimate load factor is determined
as the quotient:

-
Wintern
- - (5)
Wextern "

For each mechanism an upper value of the ultimate load factor is determined
and the problem is hereafter to determine the particular mechanism yielding
the smallest load factor. By requiring the external work to be constant, for
example

Cu = 1 (6)

it is easily seen that this: mechanism and thereby the ultimate load factor are
determined by solving the following optimization problem:

rain[m+ 0 + + m - 0-],0 + - 0 - - B . u = 0 C-u = 1 0 + --- 0 0- --- 0. (7)

The incorporation of the bolts requires only a small extension to eqn (7).
In addition to the internal work from the endplate, the internal work done by
the bolts, which is simply the product of bolt force and displacement, must
also be considered. Thus eqn (3) is amended as follows:

Wintern = m + 0 + + m - 0 - + Pbolt u (8)

where Pboxt is a vector expressing the yield condition of the individual bolts.
Finally, the problem of contact between the endplate and the resisting media
is solved by requiring the displacements to be non-negative. As can be seen,
 Design of bolted endplate connections 123

this does not increase the complexity of the problem. Thus, the bearing
capacity of the assemblage of endplate/bolts with automatic consideration of
potential prying forces is determined by solution of the following optimiz-
ation problem:

min[m+0 + + m - 0 - +Pbo~tU],0+ - 0- - B . u = 0 C . u = l 0 ÷ >-00 ->0. (9)

Requiring the displacements to be non-negative is also a requirement to the

resisting media in terms of sufficient strength. A beam splice is reckoned to
have sufficient strength when the thickness of the endplates at opposite beam
ends is equal. For a beam-column connection the method is also applicable
if one models both endplate and column flange with the appropriate strengths
and requires the difference in displacements between endplate and flange to
be non-negative. With this latter technique practically any type of bolted con-
nection can be analysed.
Endplate connections for symmetrical I sections will now be discussed. In
this case the tensile stresses in the flange and in part of the web are transferred
via the endplate to the bolts, whereas the compression stresses are transferred
by means of contact between the endplates. However, the flanges and the web
also support the endplate, in that the section practically deforms rigidly. This
effect is accommodated by introducing infinitely strong beams. Because of
symmetry, only half of the endplate is considered, and in order to reduce the
computational resources only the part of the endplate around the boltholes in
the tension zone is descretized. This is justified by the fact that the I section
rotates rigidly around the compression flange. This requires, however, that
flanges are positioned opposite to one another. One thus arrives at the model
shown in Fig. 2.
In Fig. 3 the yieldline patterns for a thin and a thick endplate are shown.
The bolts are in both cases designed assuming an infinitely strong endplate.
As can be seen, completely different yieldline patterns are decisive in the
two cases.

3 DESIGN OF BOLTS ASSUMING AN INFINITELY STRONG

ENDPLATE

In the case of an infinitely strong endplate, the ultimate load factor is easy to
determine, as it corresponds to a mechanism, where the assemblage of beams
and endplate rotates rigidly about the compression flange as shown in Fig. 3.
In this case the internal work is solely due to the yielding of the bolts:
124 P. C. Olsen

Fig. 2. Computational model of bolted endplate.

(a) (b)
t =20ram
t = lOmm
k : 1.13
k = 0.72

).
Fig. 3. Yieldline patterns for (A) a thin endplate and (B) a thick endplate
 Design of bolted endplate connections 125

Wint~m= OZPB,ihi (10)

1

where PB.i is the yield strength of the bolts, 0 is the angle of rotation and h i
is the distance between the bolts and the compression flange. The external
work is determined from eqn (11):

Wextern = Oh(o'Tbntfa + 1/6(20"T + o'c)twh) (11)

where O'T is the stress at the flange in tension, Oc is the stress at the flange
in compression, t~ and bn are the thickness and the width, respectively, of the
flange in tension, tw is the thickness of the web, and finally h is the height of
the section [eqn (11) implies a linear stress distribution through the section].
The bolts can now be designed by means of eqns (5), (10) and (11) in that
the load factor must be larger than one, or, equivalently:

~PB.ihi ~ h(o'Tbtatn + 1/6(2OFT + o'c)twh). (12)

1

For example, for given bolt dimensions, bolt strength and bolt spacing, the
number of required bolts may be determined.

4 DESIGN OF FLUSH ENDPLATE

The bolt design method of eqn (12) is completely traditional. Choosing an

endplate thickness large enough to prevent any yielding in the endplate, the
bolt dimensions thus determined are also the final dimensions. A question of
great importance is, thus, what the lower limit of thickness of the endplate is
in order to prevent yielding in the endplate.
Before addressing this question, however, another limiting value of the end-
plate thickness, assuming full strength utilization of the bolts, is determined
from the punching shear resistance. According to EC3/4/, the punching shear
resistance is determined on the basis of a critical diameter corresponding to
the mean dimension dm of the cross flats and the cross points of the bolt head
or the nut. Requiring full strength utilization of the bolts, this leads to the
following minimal endplate thickness tp:

tp --> (13)
4dmfp.y
126 P. C. Olsen

where dB is the diameter of the net stress area of the bolt, fB,y is the yield
strength of the bolts and fp,y is the yield strength of the endplate. Using the
specifications for the M-bolts, dm/dB "~ 1.9 and dN/dB ~ 1.15, where dN is
the nominal diameter of the bolt. This leads to the following requirements:

> /0.54dN ......... (8.8 bolts)

tr, -- L0.76dN ......... (10.9 bolts) (14)

assuming fp,y = 235 MPa and utilizing the yield strength of the bolt only.
A simple analysis including one bolt only and assuming the simple yield
mechanism, which is shown in Fig. 4, is now performed. The internal work
is solely due to the yielding of the endplate and it is a simple matter to deter-
mine this to be:

Wintern =
((hch)hc)1
m 2
X
+
C
- 1 + 2
e
(c .-}-x ) ~ ~pfp,yO (15)

where the symbols used are seen in Fig. 4. The internal work is minimal for:

x = ,,/m-7. (16)

N,-x

Fig. 4. Simple yield mechanism for an endplate with one bolt only.
 Design of bolted endplate connections 127

The external work is determined by means of eqn (11) and thus the minimal
thickness of the endplate is determined from:

2 h(orTbntn + (1/6)(20"T + O'c)twh)

tp --> . (17)
m 2~- +--c 1 +2 e (c+ ) ,y

In Fig. 5 the endplate thickness is plotted for the IPE-series of sections,

assuming the sections carry only a bending moment causing yielding in both
flanges. Also, the design yield strengths of section and endplate are equal.
When placing several bolts alongside the web, the yield mechanism shown
in Fig. 6 is decisive. As a result of the rotation about the compression flange,
the area denoted ABCD in Fig. 6 can not deform rigidly and must therefore
be divided as shown. The internal work is determined to be:

Wintem =
((hc+h)
m 2
x
+
c
1 +2
h-c-d
e
(c + d +
1
x) ~ tp2fp,yO
(18)

where the symbols used are seen in Fig. 6. Similarly to the yield mechanism
of Fig. 4, the internal work is minimal for the x-value determined in eqn (16)
and the external work is determined by means of eqn (11). The minimal end-
plate thickness is thus determined from:
.35 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

i i
i t
i
30 . . . . . - . . . . . . . . . . . . . . F - - - i-/ ~ - ....

~25- . . . . . . . . . I- . . . . r--- -r" . . . . I

.r. . .
I I I I
~Z~ I t I I
x~" I I I

._u 20-
(-
. . . . . . . . . . . .
i ~- . . . . I- . . . . I-
. . . .

4-,

--~ / I £P,y: £yd

- o 1o . . . . . - /----" m " = O.'Sb,F't
c'- / i
III I C : U,Jm

I e = O,Sm
5 . . . . . . ;- . . . . i-
i i
i i
i i
o I I I I I
0 100 200 300 4-00 500 600
IPE

Fig. 5. Endplate thickness for the IPE section applying eqn (17).
128 P. C. Olsen

-Jr-

Nr~

Fig. 6. Yield mechanism for a row of bolts alongside the web.

h(crTbntn + (1/6)(2trT + o'c)twh)

~2
(( )
tp
h-c-d h h-c-d
m 2 7- + - 1 +2 (c+d+ ) ,y
c e
(19)

eqn (19) is identical to eqn (17) when d = 0 is inserted. For extraordinary

bolt positions the decisive yieldline pattern may be one, where the position
of point B in Fig. 6 coincides with the intersection of web and compression
flange. In this case the endplate thickness is determined as:

O'Tbntn + (1/6)(20"T + o'~)twh

tp --> 2 (20)

(c + 2
e
,y

In Fig. 7 the effect of placing several bolts alongside the web is plotted for
the IPE-series of sections, d = 0 corresponds to the yield mechanism of Fig.
4. d = 50 and d = 100 correspond approximately to two and three bolts,
 Design of bolted endplate connections 129

35- . . . . . . . . . . . . . . . 7 .... ? .... 7 ....

d=O mm I J
.?.;.;.;.L d=50 mm I I . ..,~f~¢~_ - !
30-
,
I I I
o3 I I I
~25- . . . . P . . . . r . . . . I- . . . . . . l- . . . . I
I 1 I I I I

I I ~ I I I
.o 20- .... .~ .... ~'~-- .... ~ ..........
c- l i I
.+~ I I I
, off = ~yol
~15- .... ,t_. . . . ,~ -~ ~C =-£yd
4.#
I I I --
c~ ~ ~ £P,y- -Fyd
o_ . .
_ _ ] _ _ _
. .
I
P .... ~- m = 0,Sbft
-o10-
c-- ~ ~ C = 0,5rn
I i I
IM
, , e = 0.5m
. . . . . p . . . . ~ p
5- . . . .

i I
I I I

o I I I I I
o 100 200 300 400 500 600
IPE

Fig. 7. The effect of placing several bolts alongside the web.

respectively. It is seen that the result of placing several bolts alongside the
web may be partly a strengthening, partly a weakening of the endplate,
depending on the distance between the bolts and the compression flange. The
endplate is strengthened if this distance is large.
In what follows, it is demonstrated that prying forces are not present at the
ultimate limit state when designing the endplate thickness according to eqns
(17), (19) and (20). The yield mechanism of Fig. 8 is considered, in which
the contact surface of potential prying forces is lifted as shown, corresponding
to a rotation about the compression flange. The internal work is now:

Wintern -~- W ( e q n l 8 ) -.I-

h-c-d h \ h-c-d
U (~PB ihi - m(2 - - + - - 1)+2 (21)
h-c
),
X C e

(c + a + x) tgf.,,

and it is seen that x is once again determined by eqn (16). The endplate
thickness is deduced from the requirement that the derivative of the internal
work with respect to the lifting u must be non-negative, and it is seen that,
when applying the bolt forces of eqn (12), the thickness is once again determ-
ined by eqns (17) and (19), thus concluding that prying forces are not present.
130 P. C. Olsen

UI)= u
UA = u(1-~-dc)

Fig. 8. Lifting of the surface of potential prying forces.

5 TESTING THE DESIGN PROCEDURE FOR FLUSH ENDPLATES

The combined procedure of designing the bolts, assuming an infinitely strong

endplate and designing the actual endplate thickness according to eqns (17),
(19) and (20), is now tested by means of the numerical method. An IPE300
is selected and the ultimate load factor is firstly determined for the parameters
used in the chart of Figs 5 and 7 for one and three bolts, respectively. Sec-
ondly, for extraordinary bolt positions, an endplate with a width of 200 mm
with the first bolt positioned 100 mm below the tension flange, in line with
the tip of the flanges is examined. A yield strength fyd = 183.5 MPa is applied.
The results are shown in Fig. 9, in which the required total bolt force capacity
PB, the endplate thickness tp and the ultimate load factor A calculated are
tabulated. The yieldline patterns shown in this figure indicate that the design
results in an ultimate load factor corresponding to an infinitely strong endplate.
This is, however, not true for the endplate P3, the reason being that the end-
plate is weakened by the missing support of the tension flange at the extended
part of the endplate, resulting in a lower ultimate load factor.
In Fig. 10 the effect of increasing the total bolt force capacity is shown. It
is seen that the forces in the endplate to some extent may be redistributed
due to prying action, resulting in larger ultimate load factors. It is also notice-
 Design of bolted endplate connections 131

P1 P2 P3 P4

410 KN 507 KN 539 KN 710 KN

21,51 mm 21.39 Mm 26,60 mm 30,93 mm

1.00 1.00 0,94 1,00

Fig. 9. Design values, ultimate load factors and yielding patterns for varying flush endplate.

15o ......... ? . . . . ?-. . . . ? . . . .

&-" " "-~, P2 I i

1.40 5~k'~ P3 L . . . . . . . . . . . . . .
F,-'='=',,'- P 4 I I I I I
- I I t I

. . . . L . . . . . . . . L . . . . L . . . . I
1.30 I I I I I
0
q,.-, I I I I

0 I I I I I I
I I I I I
0
. . . . . . L . . . . L . . . . L . . . . L . . . . I
~r- 1 . 2 0 I I I I I
I I I I I I
"0 I I I I I I
I I I I I I
0
_.j 1.10 I I I I I I
II I I I I I

I I I I I
1.00 - -L . . . . L . . . . L . . . . L . . . . L . . . . I
I I I I I I
I I I I I I
I I I I I I

0.90
1.00 1.50 2.00 2.50 3.00 3.50 4.00
Bolt force factor
Fig. 10. The bearing capacity as a function of increase in bolt force capacity.

able that the increase in the ultimate load factor is non-proportional with the
increase in bolt force capacity, indicating that the connection bolts/endplate
in its entirety becomes decisive for the bearing capacity. This is also clearly
seen in Fig. 11, where the yieldline patterns for an increase in bolt force
capacity of 1.25 and 4.0, respectively, are shown.

6 DESIGN OF E X T E N D E D E N D P L A T E

Extended endplates are considered next. This type of connection could also
be designed such that no prying forces occur. In this case, however, only
yielding at the flange would be possible, resulting in rather thick endplates.
132 P. C. Olsen

PI P2 P3 P4

1.25PB

4,00P B

Fig. 11. Yieldline pattems for varying endplates and bolt force capacities.

x.
\

Fig. 12. Yield mechanism for an extended endplate.

 Design of bolted endplate connections 133

Using a slightly more complicated design procedure, the connection may be

designed with yieldlines at both the boltline of the extended part of the end-
plate and at the flange. To this end, the yield mechanism of Fig. 12 is con-
sidered. The internal work is determined from eqn (22), whereas the external
work is still given by eqn (11). Also, it is easily seen that the minimizing
value of x is still given by eqn (16). Thus, from eqns (11) and (22), the
endplate thickness and the bolt forces are derived as follows in eqns (23-25):

Wintern = (22)

m 2 +2 +2 - 1 +2 (c+d+x) 0
x c a e

-v~/hi-m 2 - +2 - 1
Intern X C

+2
bed
e
(c+d+x)
))
+

UExt "
2 t2fpy m l -- a + t~,fp,y.

tp (23)

h(crTbntn + (1/6)(20"T + crc)twh)

(m( h2- c - d +2-+2
h h - 1) + 2 h - c - d (C + d "1- ~ ' ~ ) ~ p , y
~me c a e

,((h-c-d
£ PB,ihi >-- ~ t~fpy m 2 +2--1
C
h) (24)
Intern ~e

+2 (c+d+ )
e

a
2-
1 m l
PB,e >-- ~ t~fey a (25)
a
1-
l
134 P. C. Olsen

By means of eqn (23) the endplate thickness is first determined, whereafter

the internal and the external bolts are designed by applying eqn (24) and eqn
(25), respectively. Only one row of external bolts is considered. The design
procedure is iterative in that the distance d depends on the number of internal
bolt rows.
It is seen that the required bolt force capacities of the internal and the
external bolts are probably not the same. However, applying identical bolts
and designing the length of the extended part of the endplate according to
eqn (26):
m
k-2
a a
(26)
l- m
k-
a

where:

m 2
c-. +2
)
- 1 +2 (c+d+ )
c
k= (27)
hi
Intern

yields identical required bolt force capacities for both internal and external
bolts.
Finally, the prying ratio is determined as the ratio between the total bolt
force capacity of eqns (24) and (25) and the bolt force capacity of eqn (12)
for an infinitely strong endplate where no prying forces are present. The prying
ratio is thus determined from eqn (28):

m 2 - (a/l)l
~,hi k + a l - (~l) J
All "
3 =

m 2 +2-+2 - 1 +2 (c+d+ )
~/me c a e
(28)

and it is seen that the prying ratio depends only on the geometry of the connec-
tion.
In Fig. 13 the required endplate thickness is plotted for the IPE-series of
sections, assuming the sections to carry only a bending moment causing yield-
ing in both flanges. Also the design yield strengths of section and endplate
 Design of bolted endplate connections 135

35-
.......... F.... F.... F . . . . - ? .- . -. .-,,~l,
I I I .,~--I -- _ _ iI
30- oh- = {'vd ' ' '
- r .... r . . . . r--- i .....
"c =-%a,
I
,
I
, .Z',
I
/1'
25- -
CP,,
J]
= £,,c1'
] r . . . .
'
r - -
/ - r . . .
JJ'~' . .
0
E I t I I
,,(
:o,5~ , / .,x.J~y" :
.0_ 2 0 - , ~ - - -I- . . . . . . . I
c- a =o , -~-r-/',~_h~,'- .... :
I I I I
. . . . . . . . . . . . . . L . . . . L . . . . I
q) 1 5 - I I I I
I I I I
I I I I
.... I I I I
-0~10 -
c-
iI I ~ Flushendplate
5-
. . . . ~;. . . . . ; .... ; ~ o=O.2_5~
i i s "..'..~,~.~ a=O.50m
i i t .LLA-- a=l.00m
I i I
0 I I I i I
0 1 O0 200 300 400 500 600
IPE
Fig. 13. Endplate thickness for the IPE sections using extended endplates.

are equal. It is seen, in c o m p a r i s o n w i t h flush endplates, that the r e q u i r e d

t h i c k n e s s o f e x t e n d e d e n d p l a t e s is r e d u c e d c o n s i d e r a b l y .
I n Fig. 14 the r e q u i r e d internal b o l t f o r c e c a p a c i t i e s are p l o t t e d t o g e t h e r
w i t h the r e q u i r e d e x t e r n a l b o l t f o r c e c a p a c i t y f o r v a r i o u s l e n g t h s o f the
e x t e n d e d part o f the endplate. A y i e l d s t r e n g t h f o r s e c t i o n a n d e n d p l a t e Offyd
= 183.5 M P a is a s s u m e d .

4o0 ........... ? .... ;- . . . . ? .... ;-----/~

I I I I I
-i = e yeli
-, ,
I
,
,
,
l ./
/: I
O'P =--~" -- I i I , P~ I

z3°°-'r"Y ^-Ya r .... r .... r .... ~ .... ,

"~/ PI =U,~IODI I I I / I I
4t.
C"
: o ' Sr~ ,
I
,I I /
P I /
F i
"- d =0 ' ' ' ~/ ' ~
@ t i /J I/
o 2oo-- ~ =o,5~ ~. . . . ~--~-,~---~F, ---~
~,,~=I84MP~ , ~/ , J t ~K ,
'~' y '*.,I I If I/ I / I

I I I I I
I~ I I I I I
0 i i i i
r~ 100 . . . . . . r . . . . . . . .

I I
I I
Internal
External, a~1=0.67 ;
ij=J~(dl~ i ~,-,~3~.~F..xtemal, a'11=O,SO i
0 -i I i I I I
0 ,00 200 300 ,00 s~o 600
IPE

Fig. 14. Bolt force capacities for the IPE sections using extended endplates.
136 P. C. Olsen

PLate P1 P2 P3 P4
PB,In± 296 KN 296 KN 372 KN 372 KN
PB,Ex± 149 KN 198 KN 147 KN 196 KN
tp 16.42 ~ 16,42 m~ 16.33 Mm 16.33

X 1,02 1.02 1,03 1,03

Fig. 15. Design values, ultimate load factors and yieldline patterns for varying extended end-
plates.

7 TESTING THE DESIGN PROCEDURE FOR EXTENDED

ENDPLATES

The design procedure for extended endplates is now tested by means of the
numerical method. An IPE 300 is selected and the ultimate load factor is
determined for the parameters used in Fig. 14. Also, for the same parameters
but using three internal bolts instead with d = 100 mm, the ultimate load factor
is determined. The results are shown in Fig. 15, in which the required total
bolt force capacity PB,],t and PB,Ext for the internal and external bolts, respect-
ively, the endplate thickness tp and the ultimate load factor )t calculated are
tabulated. It is seen that the design procedure is satisfactory and leads to the
expected behaviour of the connection at the ultimate limit state.
In Fig. 16 the ultimate load factor for connection P3 is plotted as a function

1.50 . . . . . . . . . . . . . . . . ~- . . . . / . . . . 7. . . . i
i Internal i i i i
Entomal i i i n Yietdllne p~cterns ~or v~rylng Internal bot±
1.40 . . . . . . L . . . . L . . . . L . . . . k . . . . k . . . . I +'trEe capacl±les,
i i J i i
I ~ g i i i
i i a i i i
L i i i :
1,25 PB,Int 1.50 PB,In± 2.0 PB,Int 4.0 PB,In±
~1.30 ~..... IlIL....
O " I'IL.... IlL.... Ill" I11.... II
I i i i i
01.20 ...... L- --L .... L .... t. . . . . L .... i

ri r
I i i i I i

~o 1.1o ..... '~n . . . . i

'L . . . . i
'L . . . . . . .i:. . i
'~ . . . . a
',

1.oo . . . . . . L .... L .... L .... L .... U ....

I I t U I
I I = I I
I I = I I I
I I I O
0.90
1.00 4.00
Bolt force foctor

Fig. 16. The bearing capacity as a function of increase in external and internal bolt force
capacities and corresponding yieldline patterns.
 Design of bolted endplate connections 137

of increasing internal and external bolt force capacities, respectively. In this

figure the yieldline patterns for varying internal bolt force capacities are also
shown. It is seen that the bearing capacity does not improve by increasing
the external bolt force capacity above that given by eqn (25). However, as
with flush endplates, the bearing capacity increases by increasing the internal
bolt force capacity, indicating that the forces in the endplate are, to some
extent, redistributed.

8 CONNECTIONS WITH SEVERAL BOLTS PER ROW

The wide-flanged H-series of sections offers the possibility of placing two

bolts per row on each side of the web and this type of connection is considered
next. In Fig. 17 the bearing capacity of a connection for an HEB300 with an
extended endplate designed according to eqns (23), (24) and (25) is plotted
as a function of increasing bolt force capacity. It is seen that the design pro-
cedure slightly overestimates the bearing capacity and that the reserves in
bearing capacity in terms of increasing bolt force capacity are less than for
the connection with only one bolt per row, as seen in Figs 10 and 16.
It is probably not possible by simple yieldline patterns to derive a simple
formula that reflects a better approximation for this type of connection. There-
fore, as a simple way of increasing the bearing capacity, the author suggests
application of the design formulae of eqns (23), (24) and (25), increasing the
endplate thickness of eqn (23) by 5%. This has been tested on several connec-
tions and leads to safe design. The effect on the bearing capacity of increasing
the endplate thickness is seen in Fig. 17.
It must be emphasized that no reduction in endplate strength due to the bolt
holes has been introduced. This is justified by the excellent agreement for the

1.50 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yletoiline pQt±erns For vorylng In±ern(~L h o t ±

Jnen~KI by ~xq i i i Force using "tp o f eq,(23),
c a p a c i ± i e s
i i i
. . . . ~- . . . . [- . . . . . . . . . . . . . . . . .

t i I i i 1,00 PB,In± 1,50 PB,]n't 2,00 PB,In± 4,00 PB,In±

. . . . L . . . . L . . . . . . . L . . . . L . . . . I
,.. ~.~.
o ',
i i ,, ,,
t ,,
t ,,
t

1.20 . . . . . . . L . . . . L . . . . L, . . . . k. . . . . I
t ~ i t

L : ~ t i
,,OIO 1 . 1 0 . . . . L . . . . U . . . . . . . . L . . . . I

i t i t

i J ~ t t
1.00. " - - ' - - - - ' - - - - ' - - - - ' - - - - . - - - -
i i = =
i i = ~ i
I i i = J
J i i i i

o.,o. ,.~o ~.~o ,.~o ~.~o ~.~o ,.~o

Bolt f o r c e f a c t o r

Fig. 17. Endplate with four bolts per row. The bearing capacity as a function of increase in
internal bolt force capacity and correspondingyieldline patterns.
138 P. C. Olsen

endplates with only one row of bolts each side of the web. However, for
connections with several bolts per row, a more rigorous approach could be
to reduce the strength of the endplate due to the bolt holes. However, being
satisfied with the simplified procedure of increasing the endplate thickness by
5%, this has not been pursued.

9 RECAPITULATION

The design formulae of the previous sections for flush and extended endplates
can be channelled into one general set of design formulae as follows:
Endplate thickness:

h(trTbntn + (1/6)(2o'T + trc)twh)

tp ~--- ~2
((bed h h )
m2 +kl-+k2-- 1 +2 h-c-d
e
(c+d+ ~'~ )~p.y"

(29)

Required bolt force capacity of internal bolts:

~ P.,ihi >- ~t2pfpy

1 (30)
Intern

m 2 ~ +kl _hc_1)+2 h-ec-d (c+d+(G))

Required bolt force capacity of external bolts:

1 m 2 - (a/l)
Pae >- ~ t2fpy (31)
" a 1 - (a/l)

where:

12 flush endplate
(32)
kl = extended endplate

{~ flush endplate
(33)
k2 = extended endplate
 Design of bolted endplate connections 139

A prerequisite for applying these design formulae is that the opposing end-
plate has the same thickness as the one considered and that the compression
flange is rigidly supported in its entire length.
Unstiffened beam/column connections have not been considered in depth
in this paper. If the conditions above are satisfied for a beam/column connec-
tion, i.e. the column flange thickness is not less than the endplate thickness
and a column flange stiffener is provided in front of the compression flange
of the beam, the design formulae are applicable. Supplementary investigations
must naturally be done for the column web. However, applying the numerical
approach described in section 2, supplemented with a similar approach for
the in-plane forces, unstiffened beam/column connections could be thoroughly
investigated based on the ideal-plasticity theory. The author has unsuccessfully
applied for funds to carry out this work.
Endplate thickness and boltforce capacity are only two of several important
factors involved in a proper endplate design. Two other factors will be dis-
cussed here, requirements for the welding and requirements regarding through-
thickness effects.
The design of endplate thickness and boltforce capacity is based on a stress
distribution in the section corresponding to the Navier-Bernouilli distribution.
The endplate is considered to be simply supported at the flanges. Thus, the
welds must resist only the direct stresses arising from the external forces.
Using the linear-elastic stress distribution in the design of the welds, full com-
patibility with the endplate design is achieved.
In order to transfer the tensile stresses of the section to the bolts, large
perpendicular tensile stresses arise in portions of the endplate next to the
welds. Therefore the endplate must be checked for sufficient strength in direc-
tions perpendicular to the endplate.

10 ROTATIONAL STIFFNESS

The rotational stiffness of the connection at the ultimate limit state is easily
determined. Adhering to the design formulae of eqns (30), (31), (32) and (33),
yielding of the bolts and yielding of the endplate are equally likely to occur
for flush endplates. The strain energy evaluated from a deformed endplate and
stiff bolts is equal to that of a straight endplate and flexible bolts. For extended
endplates the strain energy can be evaluated partly from the deformations of
the internal bolts and partly from the deformation of the extended part of the
endplate. Thus, determining the strain energy of the extended part of the end-
plate from simple beam bending properties, the rotational stiffness S at the
ultimate limit state can be evaluated as follows:
140 P. C. Olsen

s=e + -#+
a
ai)
tp
(34)
Intern

in which the strains of the bolts are based on a length corresponding to the
endplate thickness (A i a r e a of bolts). In order to evaluate eqn (34) in terms
of the classification rules in EC3 [4], in which the stiffness of the connection
is related to the stiffness EI/l of the connected beam, eqn (34) is rewritten as:

l ((~ ~ ) m )~
S=~p + --PP+a Eaih2 (35)
Intern

where I and l are the moment of inertia and the length of the beam, respect-
ively. Examining this expression, for example for the IPE-series of sections,
indicates that the connections according to EC3 can be considered as rigid for
normally applied beam lengths as well as for braced as for unbraced frames.

ACKNOWLEDGEMENTS

This study was carried out in order to be implemented in a PC steel design

programme to expedite the daily routine work and optimize the design of
bolted endplates. This programme has been implemented by several Danish
steel manufacturing companies, one of which is A-L Sthl A/S of Skjern, one
of the largest steel manufacturers in Denmark, producing steel frames and
load bearing steel structures for the building trade, who supported this study
financially. The author gratefully acknowledges this support.

REFERENCES
1. Schineis, M., Vereinfachte Berechnung geschraubter Rahmenecken. Der Bauing-
enieur, 1969, 44, 12.
2. Grundy, P., Thomas, I. R. and Bennetts, I. D., Beam-to-column moment connec-
tions. Journal of the Structural Division, 1980, 106, 313-330.
3. Thomsen, K. and Agerskov, H., Versuche zur Ermitdung des Tragverhaltens von
Kopfplattensti~flen in biegebeanspruchten gewalzten IPE und HEB-Profil-Tri~gern.
Der Stahlbau, 1973, H. 8.
4. Eurocode No. 3, Design of Steel Structures, Part 1-1: General Rules and Rules
for Buildings, April 1992.
5. Gebbeken, N., Binder, B. and Rothert, H., Zur numerischen Analyse yon
Kopfplatten-Verbindungen. Der Stahlbau, 1992, H. 9.
6. Anderheggen, E. and Kn6ppel, H., Finite element limit analysis using linear pro-
gramming. International Journal of Solids and Structures, 1972, 8, 1413-1431.