/I?r2.

90

WELDED
INTERIOR BEAM-TO-COLUMN
CONNECTIONS

AMERICAN INSTITUTE OF STEEL CONSTRVCI'ION, I.".

TA492
.W4G7

6163
 A9:l. 90

Welded Interior Beam-to-Column Connections

A report on inlJtJligalio11J carried out by

J. D. GRAHAM A. N . SHERBOURNE R. N . KHABBAZ

undtr the dir~dioll oj

C. D. JENSEN

American Instil ute of Slu/ Com/ruclion ReJtarrh al uhigh Univ,rJity

AMERICAN INSTITUTE OF STEEL CONSTRUCTION , INC.

One East Wacker Drive , Suite 3100 , Chicago IL 60601-2001
Copyrigl1l 1959. ew York. New York

Reprinted 1990 by American Institute of Steel Construction. All

rights rC3Crved. No pan of thi s publication may be reproduced
without wrllIcn pennission.

Publi ~ hcd by the American Institute of Stccl Construction. Inc.

alOne Ea" Wacker Drive. Suile 3100. Chicago. IL 60601-2001.
 Table of Contents

p,."
SYMBOLS 4

SYNOPSIS

OUTLINE OF INVESTIGATION 6

PART A- TEST PROGRAM 7

1. Two-way Connection Tests

2. Four.way Connection Tests

3. Simulated Connection Tests

I'ART B DISCUSSION OF TEST RESULTS. . . . . . . . . 20

1. Connection Requirements
2. Two-way Connection Tests
3. Four-way Connection Tests
4. Effect of Axial Load
5. Correlation of Tests

PART C ANALYSIS AND DESIGN OF CONNECTIONS . ... . 22

1. Analysis of Connections
2. Comparison of Test Results with Analysis
3. Limitations of This Investigation
4. Advocated Design Methods

CONCLUSIONS _ . . . . 28

APPENDIX . . . . . 29
1. Theoretical Analysis
2. Two-way Tests
3. Four-way Tests

BIBLIOGRAPHY . . ................ 39

ACKNOWLEDGMENTS . . . . . . . . 39

3
 Symbols

lkam Cross Sectional Ar.. Q. Resistance supplied by Stiffeners

b &am Flange Width Q, Resistance supplied by part of Connect,on adjacent

to !kam Tension Flange
b. Stiffener Width
d Beam Depth
s Elastic Section Modulus
Beam Flange Thickness
Column Depth
Column Flange Thickness
Mean Distance ~tween Beam Flanges
Stiffener Thickness
DeRmion of lkam Loading Point at V.
Young's Modulus
v. Load (also the shear) on the Test lkam to
produce M,
1 Resultant Str... on Fillet Weld in force per unit Load (also the shear) on the Test &am to
length produce M~
I~ Component of Resultant Stress on Fillet Weld due
v. Load (also the shear) on the Test Beam to
to Moment produce M.
I. Component of Resultant Str... on Fillet Weld due w Column Web Thickness
to Shear Force
H Hinge Angle
,u, Beam Web Thickness
z Plastic Section Modulus
Beam Moment of Inertia
Strain at Strain Hardening
k Column 'k' Distance
y Poisson's Ratio
L Beam Span
Average Plate Stress at Test Ultimate in Tension
m Distance between Filld Extremities of One Flange Tests
of Column
Stress at Proportional Limit
M, Theoretical Maximum or . Plastic' Moment of the
lkam 0'1111 . Ultim~te Stress

Beam Working Moment Lower Yield Stress

Moment to cause First Yitlding of the Beam Static Yield Stress
Column Working Load Upper Yield Stress

Q. Force on Conntion due to Bam Rotation per unit length

Q, Resistance supplied by Column Web &am Rotation per unit length at M.

4
 Synopsis
Previous research on beam-ta-column connections has not I. Column stiffeners are not required
been carried to the point where definite co nclusio ns, suit- (A) adjacent to the beam compression Banges if
able for the designer, (an be reached. In particular. in-
bl
w::. ---
formation is lacking on the criteria for the need of column
stiffening and on the criteria for designing it when it is P,+ ~k
needed. Information is also lacking concerning the (B) adjacent to the beam t.nsion flanges if
moment-rotation capacity of a connection and concerning
I, ;;. 0.4 Ybi
the effect on connections of beams (earning into the col-
umn web as OCCurs in four-way connections. II. Column stiffeners are required if the formulas in
A satisfactory connection is defined as one which is (A) and (B) are not satisfi.d, and l"eir minimum thick-
capable of nesses are given by
a. developing the theoretical maximum or "plastic" mo- (C) in th. case of horizontal plate sh!feners
ment of the beam when working axial load is on the
column and I, =: b1 [b I - w (I + ~ k) "J
h. permitting sufficient rotation at this moment to allow
and, as a further limitation,
the second plastic moment to form at the mid -span
of th. beam. b,
ThiS report is a summary of experimental and analyti- " >-
16
cal investigations into the behavior of connections both (D) in the case of horizontal plate stiffeners eccentric
with and without stiffeners. The first stage of this work by 2" or less,
comprised an investigation into two-way beam-to-column
1.7
connections, first by detailed tests copying practical con- I, =: -b-[bl - W (I + ~k) 1
ditions and later by simpler tests simulating these condi-
tions. The second stage comprised an investigation into where, again,
four-way beam-to-column connections, again by detailed
b,
tests copying practical conditions. The design rules stem-
ming fcom these investigations apply to those connections
" >-
16
in which - (E) in the case of vertical plate stiffen.rs parallel to
1. Beams and columns are members of the wide flar.ge the column web and located at the toes of the col
series listed in the AISC Manual. umn flanges
2. Beams are connected to both column flanges and may bl
----w
or may not be connected to both sides of the column "= I + ~k
web such that approximately equal moments are ap-
and , as a further limitation,
plied on opposit. sides of the column.
3. Connecting welds are so designed and executed that d,
they are as strong as, or stronger than, the parts con- '. >-
30
nect.d. The limitations of this investigation, the analysis lead-
The design rules finally arrived at, for the connections ing to the above formulas and design examples are given
of fully-load.d beams to column flang.s, are : in Part C.

.. )
 Outline of Investigation
In this investigation , studies arc made of two-way and
four-way interior beam-la-column connections. Attempts
are first made to copy the most severe conditions found
in practice, while in later tests those items having :1 neg-
ligible effect on the connection performance arc: elimi-
nated . Beam and column sizes used are typical of those
in a building frame.
The primary purpose is the: study of the connection
under the follo wing items:
3. Stiffening reqlli,.ements. \'V'hcn are stiffeners needed?
\"'hat are the factors involved in the behavior of the
connection with and without stiffeners? These as -
sume significance in the application of "pbstic an-
alysis" to the design of tier buildings. To :lssure the
formation of plastic hinges in the beams. the connec-
tion and the column should be capable of sustaining a
pla'itic moment in excess of, or at least equ:d to, the
plastic moment v;'Ilue of tile be;'lms.
b. Rota/ion capacity. This is another import:tnt fe;'lture
in the "plastic" ;'Ina lysis of structures since it expresses
the ability of the connection to sustain a full plastiC
moment through the required hinge angle.
The beams were welded directly to the columns for
three reasons :
1. The direct welded co nnection has certain advantlges
and may eventually come in to more general use.
2. The emph;'lsis in this in\'estigation being upon the
study of the stresses and strains in the column at the
intersection, the e1imin:J,tion of top plates and seat
angles removed a few unnecessary variables.
3. The direct-welded connection, without seat angles.
represents the seve rest loading on the column at the
connection.
Howeve r, the formulas developed by this investigation
may be used for determining the need for, and the design
of stiffeners when the beam Ranges are connected to the
columns by butt-welded plates. In this case the width and
thickness of the connection plate is used in design.

6
 Part A- Test Program
T. Two.way Conn ection Tests in a tier building. Three basic co lu mn sizes were chose n.
This program consisted of the design, preparation and The first used was an 8 WF } 1 column which was loaded
testing of specimens as shown in Table I and Figures to simulate conditio ns existing at the top of a building
I, 2, 3 and 4 for the purpose of determining the be frame where axial loads are small compared to beam
havior and stress distributions in the co nnection and its loads. The second group utilized 8 WF 67, 12 WF 40,
component members. Attention was limited primarily to and 12 WF 65 columm o n the basis of beam and column
the study of what was considered to be the most impor. loads being of the same o rder of magnitude. The third
tant practical problem, viz.! whether co lumn stiffeners are
requi red and if so how to design them, although other SYMMUI!CAL AIQUT CENTEIlINE

aspects of the problem merited considerltion. Beam and .\

column sizes were chosen to duplicate conditions existing

..
'
.,;
JACo( ENLARGEO OtT AIL
Of 'UTI WELO

I'
LATUI;Al SVPPOIT TO
ROt AlION IAIS
COM PIESSION RANGf

COWt.4NS
.Wf) 1
'WF.7
,.GUllt Oll't NTtO TO SHOW C.ONN["ION
11WFoIO
n:IIJJlo IN C.ONSTIIUCTION '2WhS

FIGURE I fiGURE 2
General Vj ew of Two- \X'a)' Test In Progress Test Arrangement-Two-Way Connections

T "BLI: I
PROGRAM OF TWOWAY DlRECfWELDED BEAMTOCOLUMN TESTS
Test Column Beam Stiffener
No Shape Web' Flange Shape Web' Fla nge Type Dimension
A l 8 \XII' 31 0.2S8 0.4>3 16 \XIF}6 0299 0428 Non e None
A2 8 WF67 0575 0.9}}
A4 12 WF 65 o WO 0.60(,
A5 12 WF99 0.580 0921
B6 8WF}1 0288 0.4}} t 3.9"xYl6"
8 8 12 WF 40 0.291 0516 t }.9"x \4"
C-9 8 WF}1 0.288 0.4}3 tt ~~ 6" X 22"
C-ll 12 \XII' 40 0.294 0.5 16 :t::!: ~r."x22"
D 12 12 \XIF 40 0.291 0.51 (, T ee sti ffener ST 6 WF }2 .5 x 22"
H I 8WF}1 o 2SS 0.4}} Doubler pbte o/i G" x 20"
Indi cates AISC MAnual v3lue.
t Horizontal plate <;tlfleners, at level of tension and compression Ranges.
tt Vertical pllte stiffene rs at t~s of column Ran~es .
7
size was a 12 WF 99 column used und~r conditions r~pr~ The point of load application on the beams was at a dis
senting th~ low~r ti~rs of a f ram~ wh~r~ axial loads ar~ tance of 4'0" from the face of the column Bange. Axial
high in comparison with beam loads. One size of beam load was applied to the specimen by an 800 kip screw
was selected throughout this program to eliminate beam type universal testing machine. The specimen was in
size as a variable, and because it is likely that floor load verted in the machine to permit the beam loads to be
ings will be constant through successive stories of a build. applied by mechanical compression jacks which were
ing. The size selected (16 WF 36) has dimensions that mounted on dynamometers. The dynamometers, in turn,
ensure the development of the plastic bending strength, were set on bearing blocks seated on the table of the
M" without local buckling of either the Range or the machine (See Figures 1 and 2).
web. All welding was done by qualified welders using 'X 0"
The test program was divided into five groups of tests diameter E6020 electrodes, except that an E6012 elec
depending upon the type of stiffening employed. (See trade was used for the first weld pass. There was much
Figures 3 and 4) . The specimens consisted of two instrumentation on the specimens, measurements being
16 WF 36 beam stubs, 4'6" long, welded directly to the taken during the test of strain distribution, deflections,
Banges of the WF column sections as shown in Figure 2. rotations and tendencies towards both local and lateral

t===3r=1F===/ ~ aBBa
I If
11
II
II
II
SERIES A SERIES B SERIES C SERIES 0
HORIZONTAL PLATE SPLIT WF
NO COLUMN VERTICAL PLATE
STIFFENERS STIFFENERS STIFFENERS
STIFFENERS
FIGURE 3

The Ttst $tries of TwoWay lkam-toColumn Connections

,-

-
.. -
"""SOl
oo.&.P I'la'[
111"10111:11$

FIGUIlE 4 FIGURE ,

The Test Series (Continued) Instrumentation Plan--Se:ries B

8
buckling of the boeam. Figure S shows the instrumenta- column failed by column web buckling at a load slightly
tion in Series B, there being few differences in the other above the beam working load, namely 1. 12 V",. Con nec-
series. tion A-4, with a thicker web showed much straining, both
Before proceeding with a test, the column was checked tension and compressio n, in the co lumn webs opposite
for axial alignment by observing the strains in four elec- the beam flanges and fai lure occurred by column web
trical strain gages located at the same level in the column buckling at a beam load of 44 kips, which is 1.82 V W.
and mounted at the outer edges of each column flange. In both cases the decrease in moment carrying capacity
The maximum variation permitted in the gage reading was quite rapid but no local buckling of the beam flanges
was about 10% at full co lumn working load. was experienced. The co lumn flanges in Test A4 de
The sequence of loading in the tests was arranged in formed co nsiderably on the second column overload.
five stages as follows: Specimens A-2 and A-5 behaved extremely well with-
I. The column load was increased in five equal incre out stiffening. Local buckling of the beam flanges oc-
ments to working load, P10, with no load on the beams. curred at 2.08 V wand 2.26 V w respectively. The loss of
(This axial load was the same for the full height of beam strength was quite gradual and the specimens sus-
the column). tained large rotations before the tests were concluded.
Upon application of the second column overload addi-
2. The beam load was increased in four equal increments
tional deformation of the column flanges was noted , but
to ,",orking load , VI(" whil e maintaining working load,
no other effect on the column was observed that would
Pw , at all times in the portion of the co lumn "below"*
indicate that column failure was imminent.
the beams. At the conclusion of this stage the "upper"
portion of the column sustained a load equal to Series B
Pw - 2Vw where
Horizontal stiffeners were placed across the column
P-Y' = the column working load (refer to Section 2.2
flanges at the level of the beam flanges in this series as
of Appendix) and
shown in Figure 3. These stiffeners were welded to both
V ~ = the applied beam working load. column flanges and to the column web. In test B6 the
3. With this working load , V w' maintained on the stiffeners were of a thickness equal to the beam flanges
beams, the column was then subjected to a first over- but in B-8 the stiffeners were thinner. This is a very
load which increased the load in the "lower" portion strong type of connection as borne out by the test results,
to 1.65 times the working load and which increased as both exhibited excellent load and rotation capacities.
the load in the " upper" portion correspondingly. This Both specimens suffered local buckling of the beam com
was done in three equal increments. The column load pression flanges at the onset of the strain hardening range
was subsequently reduced to working load in the and the increase in beam load above this level was slight.
"lower" portion. This left the specimen under the The decline of strength from the maximum value was
same loading that existed at the end of stage 2. gradual as jacking continued and no harmful effects were
4. With working load, Pte, maintained in ' the " lower" observed in the column stiffeners beyond the presence of
section of the column the beams were loaded in incre- a few strain lines. The princi pal deformations occurred
ments until failure occurred. in the beams.
S. As a last step in the testing, with the connections dam-
aged and with the last beam load still in the jacks, Series C
the column was subjected to a second overload equal The stiffening provided in this series of tests consisted
to twice the working axial load . of plates positioned vertically near the toes of the column
The test program was divided into five groups of tests flange as sllown in Figure 3. The stiffeners were arbi -
(namely A, B, C, 0 and H) depending upon the type of trarily made the same thickness as the column web. Both
stiffening employed (See Figures 3 and 4). Specimen connections C-9 and C-Il carried the required loads. In
dimensions are given in Table 1. both tests there was evidence of some slight loca l buckling
on the beam compression flanges at loads of 2.17 V w'
Series A In both tests, the column web between the beam compres
In thi s group no stiffening was provided and the tests sion flanges buckled. For specimen Cll the crilical load
ranged from the very light, thinweb 8 WF 31 column to at which this effect was first noticed was 1.97 V W' In C-II
the heavier 12 WF 99. Connection A-I with the 8 WF 31 weld failure occurred just after this in the tension flange
butt welds. In test C-9 the connection continued to carry
"Below' or "lower" and "upJ>f'r" refer to the portions of a
column below and abo,re the beam as used in actual construc-
load until at 2.16 V ~ the south stiffener plate buckled.
tion, not as in the laboratory. From this point the load fell off rapidly.

9
Series 0 Seri es H
Only one test, 0 12, WJ S performed in this group, the Only one test, II -I, was pe rformed in this g roup. SinC<.
connection being a modification of the C type using split test A I was st ronger in the tensio n region of the can
beam tee stiffeners instead of plates as show n in Figure 3. nection, this test investiga ted the effect of strengthening
The tee stiffener, while devised principa ll y for use in a the column web by th e addi ti on of a 0/,." doub ler plate
four-way beamtocolumn connection, actually served to welded flu sh with th e column web. Failure in 11 1 oc
eliminate buckli ng of both the stiffeners and the column curred by th e tension weld teaflng at mid- length of the
web. The connecti on was found to be extremely stiff, the butt weld bet\\ ee n the east he:tm :tnd th e column. The
primary cause of failure being the local buckling of the failure occurred at a beam load of 49.6 kipo; which IS
beam compressio n flanges which became large at loads 2.05 V K'" just below the load corresponding to beam pbs
in excess of 2.22 V .. Although large deformations oc tic moment The rotatiOn was adequate but the load fell
curred in the beams, the co nned ion appe.:tred to rem:tin off rapidly :Jfter the te:trin~ of the weld
elastic and little strain was obse rved in the Range of the
tee stiffener. A marked difference was noted in the be
havior of the two be:tms of the specimen and weld tears A comparison of teo;l be.lIn dc(let tions IS pre'ienled In
were observed in the beam tension Ranges at loads greater Figu re 6. Views of fOllr specimens at the completion of
than those requi red to cause bc:tm buckling. testi ng arc shown in Figu re 7.

A 2 A -5
THEORETICAL
CURVE / B-6
0 12

ULTIMA TE LOAD
v,/v,
B8
C9
1.00

->
-y
0 0 .75
0(
0~
:I
0(
w
=
WORKING LOAD
}{.f:jy~ ~-==:J -=B- J [ ---:m ]
0.50 P2V P2V P2V
A I

V V V v
I t
I
o.2511/-- - - --1--

A 8
P P P P

0L---------~10~--------2~0~------~3~0~--------4~0~------~5~0~--~
BEAM DEFLECTION. did,
FIGURE 6

Summary of Test R('su lu: B(,2m lo2d vs. B('am Ddl('("tion

10
 The results in Figures 8 :md 10 show that the co lumns applies to the C tests as shown in Figure 9 From In In
of th e A and H series, with no column fbnge stifTening. spection of the strain readings taken on the C specimens
MC not .lS stlfT ag;\lnst rout Ion .1S .HC the 16 \XlF 36 it IS noted that the column web earned .l major pHI of
beams whi ch framed to the columns. In the 13 tots (See the applied load. approximately 2Y2 to 3 times as much
Figure 9) the stiffeners pro\,jde the cqUlv:dcnt of beam as the plate stiffeners at beam working load.
fbnges to the columns, and the co lumns become as stifT The A serlCS of tests sho\\ cd high stress (QnCCntr.ltIOns
against rotation as are the framing -i n beams. The S;1me at the center of the belm tension Aangcs as indicated In

flGUur 7
I t,!!, A-I, B-S. <-9. ,lnJ D - [ "2 (Idt tl) rl~ht)

3000
'OVERALL ROTATION COLUMN WEB J
END10f USABLE 6ATA L O~RALL ROT~TlON
2500
1 "--., >" f-o s:::: THEORETICAL ULTI~ATE BEAM ~OMENT ,M.

I
/COLU~N /
r'y</,
oZ d
!J
"?~ / J.ll
7 WEB
I; ~f T ---OVE~ALL t=--=:;::1..- -
-
( TH EORETICAL FIRST YIELDING OF THE BEAM M,
r-- - r-'~ -
2~

TlEORETICAL! i BEAM : ':OTATION i p;; ~ ------- COLUMN WEB

[SOD
ff I ~ ROT'N"REO'D. AT M.
I'
I OVER~LL C~~~N I i II OVERALL ROT A~'ON COLUMN

- ~ ----
J-,.-::::
~ J WORKING LOAD ON BEAM 7
- - - - - - - - - - - - -"-
\
--->.
/ WEB
ROTATION
I~~II~~~~~~--~----~----~---+----~
t<-~2
li( l /fl====?
I I
A-I A-2 A-S A' /1F===l
500

0.001

RADIAN S
SERIES A
OL-_ _- L_ _ _ ~ _ _- L_ _ _ PER INCH ~~~~ __ ~ _________ ~

ANGU L AR ROTAT ION - RAD IAN S PER INCH

MOMENT - ROTATION OF BEAM BA SED ON G'y' 36 k s,
FIGURI. 8

Com p.:Irison o f Column Web and Overall ConncctlOn Rotatiun With that for the Beam: Ser ies A

11
 3000

END 01 USABLE DA
\
~/
7'"
k- I
MA1MUM MolNT
I
/" r,TH EORETICAL ULTIMATE BEAM MOMENT,M
.A
/
~

2500

/f- G--t--
I -h/ -/- ~-
- VI
~

'"
w
I 2000
(-

!
I
.-
I

I
THEORETICAL FIRST
YIELDING OF THE II T/-
:,
1 - - - 1 - - - - ~ --
L ROT'N. REO ' D. AT M.
U
!!; I BEAMf' /' STIFFENERS
NOT SHOWN OVERALL
I
~
I I ON DWG. / ROTATION
it
:1 1500
I
I
THEORETICAL BEAM I ;
!p /
I
~
zw ROTATION rI
I
:1
~
THEORE~'CAL BEAM / ~ /
f- - - \ - _ . -
I
:1 1000
+----
MOMENT AT
--
WORKING LOAD)/I.
I
I~ 6-b :,. 6-8 Ir:-- C, ~C." :..-- 012

I SERIES B. C & 0
500

0.001
RADIANS
PER INCH
o
ANGULAR ROTATION - RADIANS PER INCH
FIGUU9
Comparison of Column Wt!b and Overa ll Conne<tion Rotation with that for th~ !kam : St-lie!l B, C, and D

3000
END OF U'S ABLE DA~A ] .1
---- /-
_____ --~'.~~..L - --- _B.!'~~ f~A.!.I~ ~Q~E_N.!L~ ------ /------
2500
7
-,'- - -"'b ~_ ~H5:q,RE:'!.!~A~ y.!.R.~ :!)E:'o.D.!.N~.9i_B~~y ~- ------

::l2000
I oA
ift
:I:
U OVERALL
Z
0..
COLUMN WEB \
':' 1500
I.. ROT 'N. REO'D AT M.
::E I. 'FTHEORETICAL BEAM ROTATION OVERA LL COLUMN
...,
Z
I
I
I I
BEAM WORKING MOMENT
ROTATI~ V WEB
W t ---. -----r----------------- 11--.. I

~ 1000 I i
::E
P 0 .001
\" h, J!
500
I
I RADIANS
fER INCH
r \
\ \
I

I J.I I
"\
SERIES H
o I I

ANGULAR ROTATION - RADIANS PER INCH

FIGURE 10

Comparison of Column Web and Overall Connection Rotation with that for the Ikam : Series H
Figure 11. The stress d ist ribut ion on the com pression
Ranges in the B series wa s uniform on the whole, whil e
In the tension areas the stresses were somewhat higher
In the center. For the C series the distribution of stress
was unifo rm In both Ranges at V.e, whde at 1.5 V tc high
tensile stresses occurred at mid flange . Specimen 0 12
llso showed a generally uniform distribution throughout
Both Cll and D.12, however, appeared to suffer from
eccentric effects as ind ica ted by the higher stresses on one
side of the fl ange, and this probably caused the weld
tearing. Specimen H - I showed a stress concentration in
the center of the beam tension flange, the concent ration
being very pronounced at 1.S VV" Measurements of hOrl'
zontal strain in the column web were taken during the
tests. Figure 12 sho\\s a plot of these strains In specimens
... 1 and ... 2 at a beam load of 1.5 jf ...

'" -<~

...,.- .- ..-,
)... ) ...
_. .. .)
,~

0 0
":"1"",- " -
""'"
~ ~.
01" .. . . 0

\ " .... 0. .. I. FI(.UK[ 1 j

General Vr ew of Four,\X' ay Test Pr o~ress

..
0 0 III

~ .. ~.
l~"
r
0 .. 0
..
I,
DYN JACK a

, . >'.1'
>- ~
DVNAMOMETER
JA

,-.. FIGURE 11
0
..
.. <
"

SlrtSS Dist ribution in &.1m Fla nges Series .A

SMALL CUTOUTS
. . . . ..... o.D v. " ._ ....'..,..
1'400 0(; .... i lO_
...O< oo
'.-1o
~"

-00' _,.... ~
Jlt; " 01oO\1.. ' 0

( )
_N T
III. 00 '0 I , III,

/1
,I
FIGURE ORIENTED TO SHOW
CONNECTION AS FOUND IN
CONSTRUCTION

..... <"
)
.-
~

FIGURE 12
SECTION A -A

fi GURE 14
Distribution of Horizontal Strain in Column Web :
Tests A-l and A2 Test Arra ngement- f our .Way Connect ions

13
2. Four- way Connection Tests ing connecti on. The measurements taken were much the
same as in the two-way tests, Figure 16 showing the in -
This program consisted of three specimens with details
as shown in T able 2 and Figu res 13, 14 and 15. T est AA strumentation plan in Test AA.
is simil ar to T est A-4 of the ''Two-way'' series except Test AA
for two additional 16 WF 36 beams framing into the
For the beam-to-column flange connection in Test AA
column web and directly .welded thereto. In the same
that portion of the column web which was stiffened by
manner Test DD is similar to Test 0-12 of the Two.way
the flanges of the other pair of beams showed little roo
series. T est BB was exploratory in nature and does not
tation compared with the pa rt of the con necti on cons ist-
have its two-way counterpart. The beams fr aming to the
ing of 3" of the beam, the column flange and about 1"
column flanges were 16 WF 36 as before and were
of the unstiffened column web. As expected, the beams
direct welded. The other pair of beams were 12 WF 27,
d irectly welded to the column web and subjected to e9ual
the tens ion flan ges of which were welded to horizontally
opposing moments p rovided a stiff connection. With
placed column plate stiffeners. Their compression flanges
only partial stiffening provided. the connection of the
rested on tee-type seats which acted as column stiffe ners.
beams to the column flange showed considerable flexi
However. these seat plates were 4" away from the ideal
bility (See Figure 17). Local buckling of the beam
stiffener locations.
flan ges was observed at a load of 53 kips (2 .28 V w) in
The specimens were fabricated of the WF sections in-
the beams framing to the column flanges, and at a slightly
dicated in Table 2, the beams being each 4'-3" long and
higher load in the beams framing to the column web.
the columns 9' -0" long.
The falling off of the beam loads was rather slow. When
The testing was done in a five million pound universa l
the beam loads had fallen off by 15')'0 of V., twice work
testing machine which provided ample space for placing
ing load was applied to the co lum n, the whitewash indi-
these specimens and for the lateral supports. The test
cated th at the co lum n suffered cons iderabl e yield ing, but
arrangement was similar to tha t for the two-way tests.
there was no other evidence of failure in the column.
Figure 14 shows the test arrangement and is oriented to
Figure 18 shows specimen AA at the end of the test.
show the positioning of loads as found in a typical build-
Te. t BB
The con nection involving the 16 WF 36 beams, welded
directly to the column fl anges, proved to be re lative ly stiff.
The connection involving the 12 WF 27 beams framing
to the seats and top plates was considerably more flexi-
ble than an equivalent 12 WF 27; however this fl exibil ity
did not prevent the con nection from fully meeting the
established criteria for a satisfactory connection.

Tes t DO
'= t=d The connection involving the beams welded directly to
the column flanges proved stiffer tha n the connection of
the beams to the tee stiffeners (See Figure 17 ) . The stiff
TUT U lfST II TUT DO ness of the latter connection is mainly dependent on the
FIGURE J' thickness of the stem of the tee stiffener, the flan ges of
The Test Series o( Four-Way Beam.ta-Column Connections the column being too far away to offer much resistance.

TABLE 2
PROGRAM OF FOURWAY CONNECfION TESTS
Test Colum n Beam Stiffener
No. Shape Web' Flange' Shape Web' Flange' T ype Dimension
AA 12 WF 65 0.390 0.606 16 WF 36 0.299 0.428 None None
BB 12 WF 40 0.294 0.516 16 WF 36 0.299 0.428 t V2" thick
12 WF 27 0.240 0.400 t
DD 12 WF 40 0.294 0.516 16 WF 36 0.299 0.428 Tee stiffener ST6 WF 32.5 x 22"
Indicates AISC Manual value.
t Horizontal plates that served as top plate and as ~at (plate) .
14
On the other hand, the column web is ably assisted in
preventing rotation at the connection by the flanges of
the split beam tee stiffeners. The two beams that were
connected to the stiffeners had very good load and rota-
tion capacities. The east and west beams connected to the
column flanges just reached the required ultimate load .
., "
and showed a smaller rotation capacity caused by a butt
weld failure starting at a load of 49 kips (2.18 V .. ). The
tirst crack occurred in the west beam at the interface be-
tween the column flanges and the end of the butt weld
to the ~m tension flange, and increased unti l weld fai l-
ure penet rated to
the fillet welds connecting the beam
web to the column flange . The tension flange butt welds
of the north and south beams, connected to the stiffeners,
had very small cracks starting at a load of 55 kips. but
they did not progress any further since, at this load, the
beam compression flanges buckled. Figure 19 shows FIGUIlE 16

s~imen DO at the end of the test. Instrumentation Plan- Test AA

~~ OF USABLEIOATA
~
~
.-/

i /.~368EA"S Ta
~ ~E.?.!!r.:'C~,=- ~'::'!..~~~A.!' _".'!..ENT, '" L
2 .5~
~ ? r
!~ ~- I
V- 'H'
-
~A"S
: \ ,I
ROTAT ION REO' O AT Mp
<f) l::::B ; ON COLU"N FLANGE
~2 .~

u i IT,BEAi S ON SPLIT lEE ST IFFENER

..,
~
z / rHEOR TIC AL BE AJ : /.OTATlJN
,, I (12 'IF 2 7 beam s
!h
:.: ~
'.
OUI.. ~ , r-----
, V--ri, --- CONNECT ION
ROTATI ON
,
,,, ,
BEAMS ON COl UMN "

I-
Z
,
FL AN GE S ,I
THEORET ICAL
,, 1 rL
o~ '.000 ,,, ( I uLTIMATE BEAM I ~
1 MOMENT FOR IZwrZ7 I
I
:E \
,I ,,
I '

AA , BB DO
500
0.001
RAD IANS
PER INCH

o
ANGUL AR ROTATION - RADIANS PER INCH
FIGUIl! 17
Comparison of Overall Connection Rotation with that for the Beam: AA , BB, DO
1~
 3 . Simulated Connecti on Tes ts
After examining the results of the twoway tests it was
realized that practically the same stress and strain state in
a connection could be produced by far simpler and
quicker tests. These tests were of three types descClbed
as follows :
3. 1 Tests to Determine Column Web Buckling
Crit erion
These tests simu lated the lower part of the connection
in which the beam flange was in compression against the
column and consisted of a piece of column compressed
at the flanges between two bars, the size of the bars be
ing made the same as the section of the flange of the
simulated beam.
The size of the bars was kept constant at 7" x Y2",
simulating the flange of the 16 WF 36 beam used in all
the twoway tests. The bars were tack welded to the
flanges at the mid length of the colu mns, whi ch were ap -
proximately 3'-0" long. The specimen was then tested
in a 300 kip universal testing machine with the simu
lated column in a horizontal position (See Figure 20).
Eleven tests were carried out , the detai ls of which are
g iven in Table 3.

BAAS SIMUL AT ING

COLUI!4N STUe aEAM FlNtGES

fiGURE 19

Test DD--failure Details The E Series . Tests to Determine Compression Region Criterion

TABLE 3
PROGRAM OF COMPRESSION CR ITERION TESTS
Test Column Bar Simulated Failure
No. Shape Web' Flange' Width Thickness Beam Load (kips)
E 14 8 WF48 0.405 0.683 7" Y2" 16 WI' 36 137
E 15 8 WI' 58 0.510 0.808 202 .5
E 16 10 WI' 66 0.457 0.748 175 .7
E 17 10 WI' 72 0.5 10 0.808 190
El 12 WI' 40 0.294 0.516 102.5
E18 12 WF65 0.390 0.606 143
E19 12 WI' 85 0.495 0.796 247.5
E 20 14 WF 61 0.378 0.643 137.5
E 21 14 WI' 68 0.418 0.718 164
E 22 14WF84 0.45 1 0.778 22 1
E23 14 WI' 103 0.495 0.813 250
Indicates AISC Manual value.

16
 In all these tests Yielding began first in the co lumn
fdlet immediately beneath the bar. Yielding was seen to
progress into the web by means of Jines radiating from
this point and other semicircular lines orthogonal to
these.
The yielding continued some distance into the web un-
til the column web failed by buckling. At a load within PL ATES SI MUL ATI NG
20% of the failure load, a slight bending of the co lumn COLU MN STUB BEAM FL AN GES
flanges was noti ced. Table 3 presents the maximum loads
obtained in the tests. Figure 2 1 shows El at failure.

FIGURE 22

The F Series. Tests to Determine Tension Region Criterion

machine. The dimensions of both the plate and the col-

umn flange were varied to study their respective influ-
fiGURE 21
ences. The effect of changing the co lumn flange thick
Test E-l ness was further studied by repeating ce rtain of the tests
with the colum n flanges machined to nbout half the origi-
3 .2 Tests to De te rm ine Conn ecti o n Te n si on Crite rio n nal thickness. The plates simulating the beam flanges
These tests simulated the upper part of the connection were also changed in size, keeping the column section
in which the beam flange is in tension, and consisted of constant. Table 4 summarizes these tests. The plates were
two equal plates welded to the flanges of the co lumn , the butt welded to the centers of a column of length about
size of the plates being made the same as the section of 3'0", as shown in Figure 22, and the specimen then
the flange of the simulated beam. Tension was applied lined up in the testing machine with the co lumn
to these plates by means of an 800 kip universal testing horizontal.

TABLE 4

PROGRAM or TENSION CR ITERION TESTS

Test Column Plate Failure load

No. Shape Web' Hange' Width Thickness (kips) Method of failure
1'1 8 WI' 31 0.288 0.433 7" %" 100 Crack in column fillet
r2 8 wr 31 0.288 0.433 7" iI
1 1 Ii " 9S Crack in center of weJd
F'3 12 WF6S 0.390 0.606 8 112" 1'8" 149
F4 14 WF 68 0.418 0.718 81,'2" Y,q" 167
FS 14 WF 84 0.4S1 0.778 11 V2" <%t 212
F9 12 WF 6st 0.390 0.606! 8V2" 1'8 " 82 Crack in column fi ll et
F IO 14 WF 84t! O.4S I 0.778t! 11 1/ 2" Va" 12S Crack in cente r of weld
F 12 12 WF 6S 0.390 0.606 8Vz" I YJ " 189
F I ~ 14 WF 68 0.418 0.718 8 1dt" 1 V2" '99
F14 8WF67 0.57S 0933 7" %" 2S6 Crack at outside of weld
F IS 14Wrl76 0.820 1. 313 II y," VB" 444
IndIcates AISC Manual value.
t Column ftan8~ machined to ~6H for test F-9.
U Column Range machined to Va (or test FIO.
H

17
 The first yield lines were noted in the column fillet
immediately beneath the plate at a load of abollt 40%
of the ultimate load. The yield ing proceeded
a. into the column web
b. underneath the column flange parallel to the pl.te
and
c. on the column flange sta rting from the center of the
weld in lines parallel to the column web.
By the time failure occurred, yielding had progressed
2" into the web in tests F 1, F2, F3, F4, F-~ , F9 and
F-IO and had progressed acrOSS the web in tests F-1 2,
F 13, F14 and F- l~ . All specimens excep.t F. l, F9,
F-14 and F- D failed by the occurrence of a crack in the FIGURE 23
center of the butt weld, the fracture taking place after Tests Fl and F l~
noticeable flange bending. Fl and F'9 cracked in the
column fillet while F 14 and F l~ suffered a tearing out 3 .3 Ecce ntric Stiffene r Tests
which started from the outside of the column flange and
In four-way connections the column s may be stiffened,
proceeded to its ce nter. The tear pulled out part of the
opposite the compression Ranges of the flange-connected
column flange material. Table 4 presents the max imum
beams, by the support provided by the compression
loads obtained in the tests. Figure 23 shows F- ~ and F l~
fl anges or the seating plates of the beams which frame
at failure.
into the column web. In a connection such as specimen
BB (Figure 1~). where the flange.connected and web
connected beams are of different depths, their compres-
sion flanges are not opposite, and the degree of such

TABLE 5

PROGRAM OF COMPRESSION TESTS WITH ECCENTRIC STIFFENERS

Eccentricity Failure Load
Test Column Stiffener (in) (kip)
EO 12 WF 40 3~" x 1ft" x 101'4" 0 172
E-2 2 146
E3 4 11 3
E3. 3Y-t" x %" x 101'4 " 4 116
+ 3" x 1,4," x 8" Tee
E4 3%" x %" x 10%" 6 104
E-l none 10 2.~

E9 t4 WF 61 4%" x 0/8" x 12Y2" 0 282

E6 2 232 . ~
E7
.. 4 167.6
E8 6 142.8
E24 nOn<: 13n

No stiffening u~d .

18
stiffening IS questionable. To determine the degree of The results of the eccentric stiffener tests are given in
such stltfenlng J series of tests were carried out on pieces Table 5. As can be seen from both series the stiffeners
of 1 ~ WF 40 and 14 WF 61 columns approximately of eccentricity 2" provided about 6570 of the stiffening
'-I' 0" long. The columns were compressed between bars action of the concentric stiffener whereas the stiffeners of
for casts of 0, 2" . 4" and 6" eccentricity as shown in eccentricity 4" and greater provided less than 20'10 of
Figure 21 by means of a 300k universa l testing machine, the concentnc stiffening action . Figure 25 shows E8 at
the tests being similar to the compression criterion tests fai lure.
in Part 3 I. Included In the tests on the 12 WF 40 was
one (E3.!) in whi ch the compression region of test BB
was simulated that is, a tce seat was added to a stiff
ener of 4" eccentricity.

I '-I
[ ccunltlC
SlIH[HEflS

lARS Si lollA. ATIHG I 1

IAIOI rLAHG(S

FIGURE 24 FIGURE n
Te~ ts with Fccentri C Stiffeners Te!o( E8

19
 Part B- Discussion of Test Results
1. Connection Requirem ents beam. SpeCimen A L with its unstiffened , thin-web col-
1n a beam-Io-column welded connection theIe arc sev- umn section is a not able example where column web
eral regions which are subject to local overstress and buckling was the principa l cause for the high rotations
therefore it appears pertinent, before discussing the be- at low moments. In border line cases, as for example
havior of the tested connections, to define a satisfactory A-4, the buckling of the column web did not become ex
connection. It is defined as onc which is capable of de- cessive and the deformations are due to a combination of
veloping the theoretical maximum moment of resistance high inelastic strains in the column web in areas of both
of the beams (the "plastic moment") when working te~si?n an~ co.mp ression and to some web buckling. Thus
axial load is on the co lumn. A desirable additional qual- thiS Investigation clearly demonstrates the importance of
ity of a satisfactory connection is that it maintains its mo- the co lumn web opposite the compression flanges of the
ment capacity for a conside rable rotation at the ultimate beams.
load. The rotation required at plastic hinges (namely, From strain gage readings it was calculated that the
the "hinge angle") for a variety of practical structures vertical plate stiffeners of Series C in the elastic range,
has been determined in Reference 4 and its particular each transmitted only about %Gt hs of the (orces coming
application to this investigation is treated in Section 1.2 (rom the beam flanges and the web transmitted 1'8ths .
of the Appendix. Placing these stiffener plates closer to the column web
might have improved the distribution. However, since
2. Two. way Co nn ection Tes ts the prime purpose of this type of connection is to afford
A sign ificant feature of these tests was the ability of a conve ni.e?t four-way con nection , the plate usually needs
the connections to develop the strength of the beams. In to be poSitIoned flush with the edge of the column flange.
all cases except two, (A.l and A-4) where co lumn web Although there were high stress concentrations at the
cripp ling was responsible for failure, the beams were not centcrs of the butt welds in the Series A and H tests it
only able to rea ch their predicted ultimate load, but were was noted that no weld fai lures occurred until after ex~es
able to sustain this load over conside rable rotation. sive rotation had taken place.
Local buckling is a fa cto r which might influence the 3. Fou r way Conn ection Tests
value of the plastic moment of a beam section and of its
rotation capacity. Haaij er lGJ has determined the propor - All three specimens satisfied the criteria by both pos
tions of sections that will not buckle before the onset of s~sslOg the strength to develop the theoretical beam plas-
strain harden ing. The width to thickness ratio of the tIC moment and by showing sufficient rotation capacity at

beam flange, bi t, should not exceed 17, and the ratio, peak loads.
dill'" (beam depth to web thickn ess ) shou ld not ex Test AA, as shown in Figure 26, was stronger than its
ceed 70 for pure bending. The beam section chose n two-way counterpart, Test A-4 . This evidently shows that
(16 WF 36) was just within these values, with the re- the stiffe ning action provided by the two beams framing
sult that loca l buckling coincided with the beginning of
strain harde ning and was not detrimental to th e strength
o( the co nnection .
In comparing the theoretical and experimental moment-
rotation curves (Figures B, 9, and 10) in the elastic range,
the connections are not as stiff as the 16 WF 36 beams.
This fl ex ibility is of cou rse due to strains in the co lumn.
These were g reatest in Specimen A-I, with A.4, B-6,
BB, C-9 and C. LI also showing noti ceable deviation
from the theoretical curve .
The structural adequacy o( a particular type of welded
beam-t~-column connection ca n be ascertained in part by
compa rmg the moment and rotati on capacity of the beam " 1---~--~~--~---4--~
with the local resista nce and the local stiffness o f the co l
umn. The co lumn must have strengt h to res ist the beam o}---~--~,~--~,~--~--~
moment, but it need not necessarily be as st iff as the beam. Dt:FtECJIOH- JNa/,s

The desirable strength and rotation capacity is supplied FIGURE 26

partly by the co lumn and partly by the end portion of the Beam Load vs. Beam l)eAections: Test AA

20
OntO the column web strengthens the connection more in results in the following two sets of tests. Test E 18 on
than it is weakened by consequences of the triaxial a 12 WF 6~ column failed at a simulated beam flange
stresses. In both tests DD and 012 the split beam tee load of 143 kips, whereas test A4 in whicJt the 12 WF 6~
stiffeners effectively prevented any buckling of the con sect ion was used in an actual connectio n failed at a com-
nection . Test BB cannot be compa red with a two-way test puted beam flange load of 110 kips together with a com
since it had no two-way counterpa rt. puted beam web load of 40 kips.
Test BB showed much st raining in the web of the
4. Effect of Axial load 12 WF 40 column at a beam flange load of 110 kips
In both the two and the four-way tests the column axial whereas the simulated test with no beam web force failed
load had little effect on the strength and rotation capacity at a simulated beam flange force of 116 kips (See Test
of the connection. The columns showed no particular E 3a, Table ~) .
signs of di stress when subjected to an axial load of 1.65
times working load except that specimen SO showed 5.2 Tests to Determine Tension Criterion
straining in the web of the 12 WF 40 column. Since the The simulated tension region tests ignore :
strain lines were not found throughout the crosssection it 1. the dfe'll of the column axial load
may be presumed that residual stresses may have been at 2. the effect of the compression region of the connection
least partly respon sible for the appearance of these strain on the tension region.
lines. Further, at the end of each test, with the final beam
For similar reasons to those in Sect ion 5. l both of these
10:lds still applied , twice column working load was ap-
eflects should be negligible. This is borne out by the reo
plied with no evidence of marked di stress in the column .
suits of tests F-2 and H I. Test H - t , in which an actual
5. Correl ation of Tests connection was subject to axial load, suffered a weld fail-
ure at a beam flange tension load of approximately 100
5.1 Tests to Determine Compression Criterion kips while test F-2. a simple tension test suffered the same
These Series E tests give much information about the failure at 95 kips. All of the tension failures occurred
actual resistance of the web of a column to local forces because of excessive straining in a region close to the
applied at the flanges and they are intended to simulate column fillet and the center of the weld, as a result of the
the compression region of a connection . However they outward yield ing of the co lumn flanges. The shear
neglect : stresses resulting f rom the narrowing of the tension plates
due to the Poisson effect may have influenced the mode of
1. the effect of the column axial load failure in tests F-14 and F-15. These two specimens
2. the effect of the tension region of the connection on were under much higher unit tension than the other F
the compression region specimens.
3. the effect of the compression from the beam web.
5.3 Eccentric Stiffener Tests
The discussion in Section 4 indicates that column axial
Both series of tests showed a rapid decline in the ef-
load has negligible effect whereas the stress concentra-
fectiveness of the stiffener for eccentricities greater than
tions caused on the tension and compression regions are
2". In the tests on both the 12WF40 and 14WF61
so far apart that any interaction would be small. If the column stubs the stiffeners with 2" eccentricity proved
tension region of the connection does not fail then we 65% as effective as the concen tric stiffeners while those
can assume that its effect on the compression region is with 4" eccentricity were only 20ro as effective as the
negligible. The compression from the beam web does concentric stiffeners. Stiffening with still greater eccen-
have some effect and this probably caused the difference tricity had virtually no effect. For design purposes it
W orkmg load corresponds to an average axial stress of 14 . ~
would probably be advisable to neglect the resistance of
ksi . stiffene rs having eccentricities greate r than 2".

21
 Part C- Analysis and Design of Connections
1. Ana lysis of Connection s stiffer than the beam at AI,. the necessary rOlation will
As stated in Part B a satisfactory connection IS defined occur in the end of the beam.
as one which is capable of developing the theoretical
1.1 Ana lysis of Comp ression Reg io n of Con nection
ma.ximum moment of resistance of the beams when work-
ing axial load is on the column, It is also desirable for This ana lysis, the idealized "pproach, ide.hzes the
the con nection to have suffICient rotation capaci ty as ex- beam as placing on the comprcs ion region of the con-
plained in Part B. nection a concentrated for ce at the level of the beam
The analysis then should determine those items which Range. The force of the conne(tlon from tt~e beam web
arc necessary at the joint to ensure development of the is neglected.
plastic moment at the connection and, if poSSible, ade- The critical Item in this regIOn In an unstifTened con
quate rotation capacity. Potential items for investigation nection is the buckling of the co lumn web. From experi.
menta l evidence as discussed later (for illustration see
acc:
Appendix 1.3 and 1.4) a conservative estimate of the
1. The strength of that region of the connection adjacent
strengt h of the compression region of a connection could
to the beam compression flange when no column stiff-
be obtained by assuming that the resistance supplied by
ening is required.
the column web in resisting the beam flange force is
2. The strength of that region of the connection adjacent u.w(I+)k).
to the beam tension flange when no column stiffening
is requ ired.
3. The increase in the strength of the connection due to
COLU MN FLANGE NO OF FLANGE FILUT
the presence of stiffeners.
4. The possibility of column failure due to a combina
Ir'-F=rt
tion of axial and local st resses.
5. The effect of the pair of beams framing into the co l-
COLU MN
umn web on the connectio n of the other pai r of beams
onto the column flanges . WEB
6. The rotation required of connect ions and their capac-
ity to rotate. BEAM COMPRESSION

Items I, 2 and 3 will be discussed in Sections 1.1 and FLANGE

1.2 of this Part and also in the Appendix. Items 4 and 5

have been discussed in Part B, their effects havi ng been
deduced from the observation of tests. It has been ex-
FIGUIlE. 27
plained that the effects of column axial load can be
AnalYSIS of Compr~slon Rtglon of Conntion-
neglected and that the stiffening action of the second Idtaliz~d Approach
pair of beams strengthens the connection more than the
triaxial stresses set up in the column web wC2ken it. This implies that. as shown In Figure 27. there is a
A conservative procedure would then be to analyze the distribution of stress on a 2.5: I slope to the column
connection as if the second pair of beams were not pres- "k-line" so that the resistance of the column web is equiv-
ent. Item 6 has been investigated both anaJyticalJy and alent to a uniform resistance supplied over the length
experimentally. The rotation required of connections (t +5k) . Hence, for a connect ion without stiffeners
ca n be found from Reference 4. This of cou rse varies
with the beam loa ding, size and span but in Section 1.2 Q, = u, W ( + )k ) (I)

of the Append ix there is alculated a sample value of Now the force supplied by the beam fl ange when the
th, required rotation which will be greater than that re- beam is under plastic moment is bl UII so the miOimum
quired by most connections. For purpos<s of compari- column web thickness required is 8ive~ by
son th is value has been plotted on Figures 8, 9, 10 and
17 which show moment rotation curves of tested con nec-
or
bl u. = u. W (I + )~) (2)
tions. Inspection of thes< figures shows that all tested
bl
connea:ions do have sufficient rotation capacity. More- w----
- ,+)k
(})
over, if the connection is made stronger, SO that it is much
22
 bl (onslder '1I IlTeners of e((enlnuw:s of 2" or Il:ss as 6O~o
In (a.sts where ft' > stiffeners are not c<:quired
dilltlvc as compared to concentnc stiffeners. In this
5k 'T IJlter case , equatIOn (4) becomes
In the comp ression region of the connection
bl bl UJI =. aJl U' (I + 5k) + 0.6 "Jl/,b
In uses "here < --- and stiffeners are r('<lu ired
fl'
,+ 5k whlth reduces to
formulJ (:!) IS modIfied to Include the resistance of the~c 1. 7
I,=-[bl II (/ + '1:)[ (9 )
stdTentrs b
hI (11/ :::: ('~ II (I + ')k) + 0'1/ A" ( I)
where again
In the case of horizontal plate st iffeners having a total b,
width e~ua l to the width of the beam fl ange A" may be I ,,-
,? t6 (6)
approxllnated as
A" :::: ', b Two other methods of analysis of the compression
regIOn of the connecll O'l have been suggested m the
Hence, the reqUired sti ffener thickn ess
Appendix but the above analysis for both concentnc and
hI u (I + 5k ) (5 )
eccentric stiffeners is advocated for use.
b
1.2 Analysis of Tension Region of Connection
As a fu rther limitation (See Section 1. 1 of the Ap-
The mechanism of failure in thiS reg ion is as follows :
pendiX) ,
a co lumn Range acts as two plates. each of whi ch is fixed
b,
I ,,- (6 ) along three edges and free along the other together With
, p 16
a central rigid portion, the whole being loaded by the
Tests C-9. C- II and D-1 2 indicate that the ve rtical beam tension Range. The load remains more or less
plate stiffeners ca rry about half the st ress that the column uniformly distributed until the "plates" reach their ulti
web does , Making this assu mption, formula (4 ) becomes mate ca rrying ca paci ty. At this stage , the " plates" deBect
in the case of vertica l plate st iffeners, at their outer edges causing excessive straining in the
cent ral portion of the butt weld. in the co lumn flange
hI a, _ a, IV (I + 5k ) + ~
2
2 I, (I + 5k) adjacent to the weld and in the column fillet. Failure
then OCcu rs by cracking in one of these regions. The
so that "plates" are under bending action so their ultimate ca
hI pacity depends on the square of their thickness. AnalYSIS
1,::::--- - '" (7)
,+ 5k in the Appendix (Section 1.6) illustrates that a conserva-
tive estimate of the capacity for each "plate" for wide
As a further limitation (50 Section 1.1 of the flange co lumns is 3.5 U.,/ e 2. The central rigid part o(
Appendix ). width 'm' adjacent to the column web will be highly
strained and hence will carry a (orce corresponding to its
(8)
area at yield stress. Hence
In those: cases in which the beam Range width is much Q, = u.,lm + 7U.,I? ( 10)
less than the column Bange width these C type stiffeners
The force in the beam tension flange when plastic
would not be as dfective as assumed and it would be
moment is applied to the beam is bl u,. To give 20'1.
inadvisable to rely on their stiffening action when the
conservatism in this region o( the connection correspond.
co lumn web is greatly deficient according to formula (3) .
ing approximately with the average conservatism in the
Eccentric Stiffening compression region one obtains
Since the testing done on eccentric stiffeners was very hI u, = 0 .8 [u, 1m + 7 u, I;] (11)
limited. observations concerning their action cannot be
This reduces to
conclusive. Very light columns were used ; hence the
results should if anything be conservative.
The tests have indicated that the horizontal plate ( 12)
stiffeners of eccentricities greater than 2" had very little
stiffening action. A conservative design procedure then Ie being the requ ired column Bange thickness.
would be to neglect the action of such stiffeners and 10 If beam and column sizes are taken (rom the AISC
23
Manual then the value of m jh for all those con nection s in N_.... ....u .. .........1,.. _
whi ch formula ( 12 ) is approximately satisfied varies ........". _ '.e!
from 0 . 1~ to 0 .20. Making the conse rv at ive as
su mption m/b = 0.15. (1 2) reduces to
I, = 0.4Vbr ( 13)
z
120
o
In Cases where Ie> 0.4 ybi st iffeners are not required ;: .. "26
J9I
in the tension reg ion of the connection. trl
z 3.2.Jl'O

In cases where Ie < 0.4 Y Tt and sti ffeners arc requi red
one ha s equilibrium configurat ions exact ly the same as

SI.
. U1

2"6.26.
....
. ,61 . n 110
2",l21
those in the compress ion region of the connection. Hence lOl,'U,lH
IS)
' ,193
stiffening requ irements will be given by equations (5).
.,."'."'.In
. 1110 9 U
(6). (7) and (8) . . /$' . '
,. ',n
10&.120 100
1.3 Relative Strength s of Tension and Compressi on ' ''2.1)6 '9 : ~ . $
_._ ' !.!-!~ _____________________________ . ]J__ _
Regions of the Connection ".'01 $.,1Ir.I.92 ') . .0

Equation (3) states that a connect ion will be on the : ~:.1 ."3
320.1$
~," , "1
n
verge of n(>(~di ng stiffeners in the com pression region if ---:~~---------. ~~;------- i );------------

.,
0.5J 090
hi ~
w= - - - z~..J
OU
1 + 5k
CI:i ~~
or zzt: ."
bl = w (I + 5k ) (14 )
wOO:
.... u
-'- I L 080
from equations (13) and (14) this connect ion will or
will not need stiffeners in the tension region acco rding
:0
<o!
it
'2 ..
it

COLUMN SIZE
to whethe r
FIG U RE 28
I, oS 0 .4 V w (I 5k ) + Cri ticl l Parts of Connections between Standard WF BI:Jms
i.e., and Columns
Ie
--=
V wk
oS 0.4 V 5 +I/k (1 5) 2. Comparison of Test Results with Analys is
2.1 Compression Region of Co nne ction
Since for all practical connections in which (12) is As explained in Part 8, the connect ion tests gave some
approximately satisfied what different results from the analogous compression
tests becau~ the former involved the additional compres
0 .2 < I/ k < 0 .8
sion supplied by the beam web. As can be $Ce n fr om
then by taking I/ k = 0.2 it ca n be seen that this connec-
tion will need stiffeners in the ten sion region if
T able 7. the assumption of a length of (I 7k ) of co l- +
umn web at yield stress res isting the force applied through
I, the simulated beam Range in the comp ression tests (Series
-=
\ 1 wk
<09 t ( 16) E) is conservative. Also, as seen from Table 6, the use
of the compress ion design criterion
and by taking Il k = 0 .8 it can be seen that this connec- bl
tion will not need stiffeners in the tension region if w=-- 0)
1 + 5k
I, advocated in the last section leads to co nservative result s
-= >0.96 (17) when compared with connection tests. The results from
" wk Table 6 are summarized as follows:
figure 28 shows a plot of the values of 1,/ V wk for all 1. For test A-l , formula (3) requires that the co lumn
8", 1Oil, 12" and 14" nominal depth columns of the web be 0.666" thick. The actual thi ckness was
wide flange series. It can be secn from this figure that 0 .284". and the column web (a iled at a load slightly
in most (ast's the (ritial region of the connection depends in excess of working load as shown in Figure 6.
only on the column parameters. For values of leI Y wk 2. For test A-2, the formula requires a web thickness of
between 0.91 and 0.96 the need for column stiffening 0.428" and as would be expected the thi ckness of
wi ll depend on the beam. 0,~87" proved satisfactory.

24
 Connection A4 requires a web thickness of 0.470". 8. The formulas showed the C, 0 and DD co nnections
With an actual thickness of 0.417", the connection to I>< adequate and so they proved to 1><. By the lime
attained over 80% of the required moment. the beams had failed, however, there was some
4. The formula shows A5 to be entirely adequate with buckling in the column stiffeners.
out stiffeners and It so proved to be. The theoretical restraint provided by horizontal stiff
5. The formul. shows HI to I>< slightly inadequate but eners in a connection is given by (Til bl, (refer to formula
it did take the maximum moment reached in the test, (4 .
thIS moment I><ing 9570 of the plastic moment. There
Compa rison with tests show:
was some straini ng in the column web. but failure
a. Test El in which an unstiffened 12 WF 40 was com-
did not appear to be imminent in the comp ression
region pressed failed at 102.5 kips whereas test E-o in which
the same column was stiffened with two ~" hori -
6. The formula shows AA to be inadequate, but, prob
zontal stiffeners failed at 172 kips. The difference of
ably because the stiffening action of the second pair
of beams was not considered in the analysis. the con- 69.5 kips compares favorably WIth the calculated d,f
nection proved satisfactory. ference of 63 kips.
7. For B6, B8 and BB, the formulas show thin stiff b. A similar examination of tests E9 and E20 on a
eners to be reqUired In the tests there was no eVI- 14 WF 61 show an experimentally determined differ-
dence of overstress in the stiffeners actually supplied, ence of 144.5 kips compared to the calculated differ
except for a few strai n lines in the B8 stiffeners. ence of 115 kips.

TABLE 6
COMPAR ISON OF COMPRESSION REGION CRITERION
WITH CONNECTION TEST RESULTS
Req'd Manual Measured Req'd Actual
bl k wI w I,
Specimen in.2 10, 10.
1V
10. in. 10 . "In. Remarks
1.. 1 2.99 0.812 0.666 0.288 0.284 Column web buckled
1..2 2,99 1.312 0.428 0.575 0.587 Column web O.K.
1..4 2,99 1.188 0.470 0.390 0.417 Column web weak
1..5 2.99 1.500 0.378 0.580 0.580 Column web O.K.
B6 2.99 0.812 0 288 0.284 0.258H 0.437 Stiffened connections
B8 2.99 1.125 0.294 0,300 0.25* 0.250 satisfactory
C,9 2.99 0.812 0.288 0.284 0.382t 0.437 Connections 0 K. but
C,II 2.99 1.125 0.294 0.300 0.40* 0.250 some stiff. buckling
012 2,99 1.125 0.294 0.39** 0.606 Connection O.K.
H I 2,99 0.812 0.666 (0,288) 0.600m (I)
AI.. 3.02 1.188 0.474 0.390 0,395 Connection O.K.
BB 2.89 1.125 0.294 0.316 0.25 * 0.5 Connection O.K.
DO 2.91 1.1 25 0,294 0.317 OAO** 0.6 Connection O.K.
II) Column web 0 K up to O . 9~M" when fadure occurred in tension region of connection .
~termlOed by-slendernHs limitation, Equation (6) .
~termined by slenderness limitation, Equation (8) .
~:at 4" above compression Range. Stiffening also included a vertical plate ~neath the horizontal stiffener
bl
t lXtermined by t'qu3tion u' = ---
I+H
bl-1I/ (/+ ~~)
U ~1C'rminC'd by t'qu;lIion I,
b
tU ~6" doubler pl3te :added 10 web.
bl
t Dt-termined by t'quation I. = - - - , - IV
I + ~.
25
 There is some inconsistency in the above compression the plate was strained into the strain hardening range
region analysis since a length of column web of (I + 7k) and failure was probably caused by shearing stresses at
is assumed to be effective in the simulated tests whereas the ends of th e weld due to drawing down of the plate.
an effective length of only (I +
~k) is assumed in the A further indication of this is that the weld failure began
connection tests. Formula (24) given in the Appendix at one end of the weld. This type of failure would not
is possibly a more rational approach to the analysis of the occur in an actual connection since the beam flange is not
compression side. This formula stressed above yield stress.

w =
bl + 3.lkw. (24)
The second comparison, between actual column flange
thickness and that re9uired by e9uatlon (13), is mainly of
1+ 7k
statistical interest . The last column shows the ratio of
is consistent when applied to the connection tests and to
tension plate stress at column failure to tension plate
the simplified tests. In the simplified tests of course
Yield stress and illustrates that in all but three tests (F4,
w, = O. However formulas (3) and (24) give nearly F14 and F D) the tension plate was much stronger than
the same results when applied to practical connections
would have been sufficient to cause column failure at or
and are simpler.
prior to tension plate yield .. Considerable conservatism
2.2 Te nsion Region of Conn ection in equation (13) is illustrated in the cases of F-4 and
F 14. This is probably due to the 20'70 conservatism
The only connection specimen in whi ch the primary
introduced in e9uation (II) .
cause of failure was in the tension region was test H -l
where failure occurred at approximately 9~'7o of the 3. lim itation s of This Investigat ion
beam plastic moment. The actu al column flange thi ckness
in thi s case was 0 .43 3" while that requlCed by formula The investigation considered two- and four-way interior
(13) is 0.69" . Hence in this case formula (13) appears beam -to-column connections in which every beam of the
conservative. connection was loaded e9ually and gradually to failure.
Table 8 compares the tension tests with the analysis by Some modification of the reported behavior might have
means of two methods-first through the ultimate capac- been observed if the following variations had been
ity e9uation (10) and then through the final design included :
e9uation (13) . a. Repelilive Loading. A sufficient number of cycles of
The comparison with equation (10) shows conserv- loading and unloading could cause premature failure
atism in all cases except test F-15 . However in this case but this is unlikely since much of the load in a build-

TABLE 7

COMPARISON OF FORMULA, Q, = u, w (I + 7k) WITH COMPRESSION TESTS

Bar Column Web Computed Test
Thickness Yield , U II w' k' Q, Q.
Test Column in . ksi 10 . 10 . kip kip
E 1 12 WF 40 Yz 40.2 0294 1.12 ~ 99 102 . ~
E-14 SWF4S Yz 34.4 0.40 ~ 1.063 110.1 137
E-n SWF~S Yz 36.2 0 . ~10 1.1S8 162.6 202.~

E 16 10 WF 66 Yz 40.0 0.4 ~ 7 1.2~ 169.0 17~ . 7

E-17 10 WF 72 Yz 3~ . 0 O . ~IO 1.313 173 190
E-IS 12 WF 6~ Yz 37. 2 0.390 1.188 129 143
E19 12 WF 8~ Yz 37.8 0 . 49~ 1.3n 190 247.~
E20 14 WF 61 Yz 36.2 0.378 1.2~ 127 137 . ~

E-21 14 WF 68 Yz 38.3 0.418 1.313 D~ 164

E22 14WF84 VZ 39.3 0.4l1 1.3n 180 221
E-23 14 WF 103 Vz 38 . ~ 0.49~ 1.438 201 2~0

AJSC Manual values.

26
 ing is dead load and any variation of total stress woul d From formula (13)
be of small magnitude. required t, = 0.4 VTt
b. Un"f1,al Loading of Opposing Beams. In this case
shear stresses would be induced in the column web. = 0.842"
However, when the beam loadings are approximately actual Ie = 0.921"
th e same as is usually the case at interior co lumns the
above design formulas would be valid. They might Hence no stiffening is needed in the tension region of
require modification in the ext reme case of a beam the connection. The computation for tension stiffening
framed into only one column flange. could have been omitted by inspection of Figure 28 which
shows that the compression region of the connection for
c. Wind Loading. This would tend to cause moments in
a 12 WF 99 is the critical one regardless of beam
the same direction and hence high shear stresses in the
dimensions.
column web.
4 .2 Connection i" Which Stiffening is Required in
4. Advocated Design Methods
Compression Regior. Only,
There follow examples of connection design using the
Consider a two-way connection in which 16 WF 58
proposed formulas.
beams frame onto the flanges of a 10 WF 89 column.
4 .1 Connection in Which No Stiffening is Required From formula (3)
Consider a two-way connection in which 16 WF 50 required w = 0.670"
beams frame onto the flanges of a 12 WF 99 column .
From formula (3) required
But actua l w = 0.615"
H ence stiffening is required in the compression region
bl
tu=--- of the connection. The required size of horizontal plate
1+ 5k stiffeners is given by equations (5) and (6) .
w = 0.546" From equation (5)
actual w = 0.580" .
required I.
bl-w (I
= b
+
5k)
Hence no stiffening is needed in the compression re o
gion of the connection. = 0.053"

TABLE 8

COMPARISON OF TENSION REGION ANALYSIS WITH TENSION TESTS

Av. Plate
Yield Stress Ultimate CapaCity, Q. Stress Flange Thickness
Test Column Column Plate Computed at Test Computed
No. Stub Flange u. from (10)" Test Vlt., 0'0 from (13)" Actual 0'0/0'"
F-l 8 WF 31 37.0 38.9 81 100 19 0.94 0.43 0.49
F-2 8 WF 31 37.0 38.9 68 95 31 0.72 0.43 0.80
F-3 12 WF 65 36.0 31.6 123 149 28 0.86 0.61 0.89
F-4 14 WF 68 34.2 31.6 155 167 32 0.89 0.72 1.01
F-5 14 WF 84 34.2 31.9 191 212 21 1.27 0.78 0.66
F-9 12 WF 65 36.0 31.6 55 82 15 0.86 0.31 0.47
F-l0 14 WF 84 34.2 31.9 80 125 12 1.27 0.38 0.38
F-12 12 WF 65 36.0 31.8 167 189 15 1.35 0.61 0.47
F-13 14 WF68 ,4.2 31.8 200 199 16 1.37 0.72 0.50
F-14 8WF67 33.5 38.9 242 256 45 0.99 0.93 1.16
F-15 14 WF 176 36.0 31.9 456 444 44 1.24 1.31 1.38
Dimensions of the sptcimen are given in Table 4 .
Measured from coupon tests .
Adjusted for variation in yield stresses from 33 ksi.
27
 But from equation (6) If vertical plate stiffeners are to be used equation (7)
, ..... -b,
, ,p 16
gives
bl
:> 0.2~"
required I
, = -'
,+ ' k II

Hence 10 compression region of connection use ~" = 1.1 79"

horizontal plate stiffene rs, welded along three edgcs.
From Equation (8)
If vertica l plate stiffeners arc required equatIOn (7)
gives d,
', # -30
("
required t = - - - - /V
, I + ~k = 0.40~" < 1.1 79"
= 00~6" Hence use 13~ (l" vertical plate stiffeners Rush \I, Ith the
But from equation (8) toes of the column Ranges.

d, 4 .4 Ecc e nt ric Stiffening

I, :> 30
Consider the S;1me connection as in Section 4.3 with,
:> 0362" in addition, two 16 WF 36 beam, framing into opposite
Hence use 8" vertical plate stiffeners. sides of the web of the 12 WF 6~ column. If the tension
Ranges of the beams are at the same level then the seating
From formula (13) plates of the 16 WF 36 beams can be used as sbtleners
of approximately 2" eccentricity for the 18 WF 105
required It = 0.934"
beams.
But actual I, = 0.998" The required thickness, I .. is gIven by equation (9) ;
Hence no stiffening is required in the tension region of 1.7
the connection. I, == T[bl- W (I-+- ~k )l
4 .3 Connection in Which Stiffening is Needed in == 1.164"
80th Tension and Compression Region s
This satisfies equation (6) .
Consider a connection in which 18 WF 105 beams
Hence use 10/,0' seating plates for the 16 WF 36
frame onto the flanges of a 12 WF 6~ column. Equa.
beams.
tions (3) and (13) indicate that stiffeners are required
in both the tension and compression regions of the
connection.
If horizonul plate stiffeners are to be used, equation
Conclusions
(~) gives Results of this investigation show that stiffening may
I, = 0.685" be omitted in many beamtocolumn connections. Sum
which satisfies equation (6). marized in the Synopsis are recommendations for design
defining the cases for which stiffen ing may be omitted,
Hence use I YIo" horizonta l plate stiffeners in both and also suggesting the proportions of stiffeners for ca~s
tension and compression regions of the connection. when they are needed.

28
 Appendix
1. Theo retic a l Anal ysis
1.1 Limi ting Sle nd e rness of Stiffen ers V"'t(al SII/ftner - CTyp"
Thickness, tt
The slenderness limits (or the stiffeners are difficult to
establish because-
a. The restraint provided by wdds at the ends of stiffen-
Slm~y suppo:.e,d~ ![
ers is not known.
b. The stress distributions in the stiffeners are not known. I d,
The assumptions made in the following analysis prob
As indicated in the Figure, consider the stiffener sim-
ably lead to conservative limits. The calculations for the
ply supported along the edges welded to the column
limiting slenderness of stiffeners are taken from formulas
flanges.
and figures in R<f<rence 6.

Horizonlal Sli/ftn" - B Typ" u~= (~'~ y') (;J

12

For aer:= all = 33 ksi

5imf0 sVfpor ted
d,
- = 30 (8)
I,
b,
1.2 Rotation of Connections
Fixed _ free Examination of Figure 13 of Reference 4 shows that
ThicKness> t, the "hinge angle" or rotation at plastic moment required
at the ends of a fixed ended bn.m uniform ly loaded .Iong
x its length, so that it will be able to form a mechanism, is
given by
Simply 5uppor led H=(i .ppL
1
( 18)

M.L
As indicated in the Figure, consider the stiffener fixed or H = 6EI ( 19)
along the edge welded to the column web and conserva
tively assume it simply supported along the edges welded Taking a practical case of a 16 WF 36 beam of 24'
to the column Oanges. span the required rotation is calculated to be
Using formula (3 . 1~) of Reference 6 and the (onstants H = 7.2 X 10- ' radians
DI = 8,000 ksi Here a partirular case is taken but the above value ot
the rotation will be greater than that required of most
D" = 16,000 ksi
connections. Considering a 12" gage length spanning
D, = 31,000 ksi across the column the average rotation required across this
length is 1.2 X 10-' radians p"r inch. This value is
G, 2,400 ksi (Ref. 6) plotted on all figures showing connection rotation char-

u~= 7510 (:J acteristics.

1.3 Elastic Distribution of Stress on Column Ok' Lin e

E. W . Parkes5 developed a theory giving the stress
For Ucr = UJI = 33 ksi di stribution just inside the Bange of a column (in this
b, case the column 'k' line) fo r either a tension or com-
_=
I,
1~ . 2 pression loading on the flanges while the stresses are still
in the elastic range. For purposes of our case we will
make the idealizations that-
To round figures !:!.I, = 16 (6)
1. The load applied to the column flange can be con
29
 sidered as a line load perpendicular to the co lumn column web in the development of the compression cri-
web. terion in Sedion C. This resistance is represented by the
2. T he moment of inertia or the beam fl ange about the corncrs of the rcctangle In Figure 29 wlll ch show YIeld
ax is through its own centroid pa rall el to the flange point stress dlstnbuted over a distance (I + :>k) for the
can be considered as in fi nite. 'A' Senes Tests ,nd over a Jistan ce (I + 7 k) for the T
3. The distance between the co lumn 'k' line and the Series Tests.
centroid of the column flange can be considereJ l S As dlustrated in the figure it does so happen tlut the
negligible compared to the depth of the column . non dimensionalizlng stress, 0 0' 3S used by Parkes (auscs
4. As far as stress ana lysis is concerned the web or the th e rJt io (ll/ l u o to have values \'e ry close to 0. I for all the
column an be considered as innni(cly wide so rh:H fhe specimcns tested ex cept the colu mn sel tlon 12 \XlF 6'; JS
stress distribution at mid width is uniform. uscd III test A-4 Heme the 3ctual Inei.tsti c strC"iS dl \trl
but ion at failure for all the test cases cx(ept A I I ) f(;pre
P.lCkes a.nalyzes the case mentioned above and J.bo the
sen ted c10sdy by the pl ot on Fig ure 29 "Iud> ",duJe, the
realisti c case where the above idealizations do not apply.
ho n zo ntal lin e at YIeld stress r e pre sc ntJll ~ the In CIJ \ tll
For the case of all wide flange columns as used in prac-
reS lstant e and the obl Ique IlO e rrpresc ntlllg the eiJstl(
tice however the deviation in the clastic stress di stnbu -
resistance. Since the arC'a under till S (une IS ,gre.lter th,ln
tion between the idealized and the realistic cascs IS less
the area und er the curves represe nting the assumed r<:
than 570' Being based on the idealized case then the
sistJn ce of the column webs then the assumptIon of J
non dunensionalized Curve as drawn in I\gure 29 repre
sents to 5% the elastic stress distribution along the col -
dI stribution of yield stress over J dist.ln(e of (t + ') k)
umn 'k' line for all wide flange shapes used in pra ctice. or (I + 7k) as the case may be is conservative.
It is also interestIng to note the stress distribution at
The sca le of Figure 29 has been made so that the area be-
neath this CUlVe represents the ultimate load as obtaIned various stages of loadIng. In the elastic stages of the tests,
from tests. For purposes of plotting this figure Parkes the distribution of stress is simIlar to that shown by the
used the non dimensiona Jizing parameters Xo and OD elastic curve. After a little yielding has occurrcd, a pla-
which were functions of the colum n dimemions. The teau will develop at yield stress. This plateau will be
cu rve, of course, is not the stress distnbution at faIlure (orne wider as the load increases until at fai lure thc dis-
since yie lding will have taken place. However, by the use tribution is as shown.
of the app rop riate vertical scale facto r this curve wi ll rep-
rese nt the stress distribution unti l the first yie lding occurs. l ,S Alte rnativ e Desig n Fo rmul a s for Comp ressi on
Reg i on of Conne ctions
a CORNER C7' RECTANGLE ENQ..OSIHG ASSUhEO CQ.JJMN The idealized met hod of design has been desc ribed in
WEB RESlSTa.NG( IN ' A' SERIES " e , fT, DISTRIBUTED
Section C. Two other approaches are however worthy of
0+---,-- OIlER 11+!)l1
note:
~ o c.oRHER C7' RECTAHGU: ENClOSING AS5lMEO COU,NN
r:;; Q,f-- \-f- -
WEB RESISTANCE: IN '[ ' $(IIII[S,'e
OYER It ... n).
,17, DlSTRl8UT[O

1,5 1 Plastic Ana ly sis App roach

r E~ 0 -"'to 0 REPft[S[NTS OHE TEST

~ i cu!---l\-
This approach assumes a stress distribution in the
beam, loaded to its capacity AI,.. as shown by Section a-a
~ z
4PPfIOX',,"fE INELASlIC: OISTRlII!VTIOH.AT fAlL~ in Figure 30-a. The corresponding stress dist ribution in
r6 D
the column web at the end of the flange-to-web fi ll et is
~ O'r--+-~~-~~~~~~-+-
show n by S<ction b-b, Th is procedure resu lts in the fol -
.i~ lowing ana lysis:
on ,~~._~~~=-~--~---
l ___ ) 4 :. 6 a. Umlilftlled Column}. (Serin A) . Assume the beam
.!. . OISTAHCE FAOM BUM CCNl'MSSIC)foI FLAHG[
.. NON OIIIIEHSIOI'UUllH(i DISTANCE is developing its plastic moment, M,. For the com
flCURE 29
pression fl ange the pressu re aga inst the co lumn will be
approximately as show n in Figu re 30-a.
Stress Distribution on Column 'K' line Ad,acent 10 8t'2m
Compression F1angt A.
Then
1.4 Probable Inelasti c Distribution of Stress on
Q'=2 u ,
Column 'k' Line
The area under the elastic cu rve d iscussed above ca n
Ix compa red with the assumed resistance offered by the

30
 c. Coillmm wilh Verliral SlifJenerl (SerifJ C and D) .
~COLUMN 'k' LINE The presence of the stiffeners modifies equation ( 21)

ASSUMED STRESS
to Q, == Q, Q, +
Since the stiffener plate is at the toe of the Range it
, will not be as effective in resisting the beam compres -
sion as is the column web. Strain readings on web and
stiffener indicate that the stresses in the stiffeners are
approximately one-half those in the web .
Assuming the latter
,
3~,
2v,
Q, == - 2 I, (I + ('k)
STRESS OISTRIBlJTlON ,
IN BEAM AT M, 't:,. Hence
0- PLASTIC ANAL Y515 APPROACH

, A,
- 2- "If == (1" fl.'
[ ~ t 3k ] + ", I, (I + 6k)
i_COLUMN 'k'Ll NE
I
,L... ASSUMED STRESS
t Ilere f ore 1,:= _
I[ A, _ "' (dt6k) ] (23)
DISTRIBUTION ON
I COLUMN 'k' LINE 2 1+ 6k
, b
The stiffener thickness is again restricted by the in
equality, I, ;;. d,/30.

1.52 Mod ified Plastic Analysis Approach

The pceceding analysis assumes that a.l failure a length
STRESS DISTRIBUTION
IN BEAM AT M, of (J/2 +
3k) of web is at yield stress. How<vcr in most
connections the beam web is thinner than the column web
b-MOOIFIED PLASTIC ANALYSI S APPROACH so that near the horizontal centerline of the connection
where the effect of the beam Aange force is n<gligibl< th<
FIGURE 30
column web merely resists the beam web force and so is
An31ysis of Compression Region of Connection
not at yield st~ess.
If we assume as we have done in the Series E tests and
If the compression region of the connection IS just as shown in Figure 30b that the length of co lumn web
sJti sfactory without stiffeners effective in resisting the beam flange force is (I 7k) +
Q, == Q, and th at the beam web force outs ide this region is re-
sisted by the colu mn web immediately adjacent to it then
or
d
+,k J II,
a, (20)
equilibrium over the length of (I 7k) giv<s +
" . IV [
1 2
a. U1l11ifJened Connection .
II I, 7k
therefore w==---
J + 6k
(21 ) bJ (1" + fIJ" -2 "" == tv (I + 7k) (1"
b. Col/lamJ wilh H Orlzonla} Plale SuDenerJ (SerieJ B). bl+l .5kw,
or tv == (24)
The presence of the stiffeners modifies equation (20) 1+ 7k
to Dy following the same procedure as that in Section
C we have the resuits

b. Horiz onlal Plale StiffenerJ.

1 1
therefore I, == -2b [A, - (d + 6k) "'] (22) 1, ==- [bl+3 .5kw, - 'V(I +7kl]
b,
(25)

I. is again subject to the limitation that I, ~ b./16 where" I, is aga in subject to the limitation th1t
as show n in Part 1 of the Appendix. I, ;;. b,/ 16 .
31
 A B
c. V~rli(a/ P/a/ ~ SliD~n~rJ.
bl + n k U'. I I
-w, (26)
,- ,+ 7k
1 - -<m Centrol RIgId Porfton

where I, > d,/30.

Table 9 compares the results of these two approaches
with the approach in Part C for the connections tested ,
~ b
1.6 Analysis of Tension Region of Connection
Figure A illu strates the action of the column flange in "m" DISTANCE
the tension region of the connection. The column flange m = W t 2(k-f c )
can be considered as acting as two plates both of type
ABCD. The beam Bang< is assumed to place a line load
on each of these plates. The effective length of the plates D
is assumed to be 12 Ie and the plates are assumed to be COLUMN FLANGE
fixed at the ends of this length. The plate is also assumed
to be fixed adjacent to the column web. Analysis of this FIxed
plate by means of yield hne theory(7) leads to the result A / B b, m
that the ultimate capacity of this plate is
q=2' - "2
h .l1.-!ll
z z

= 41/3 + /3~
line Load p . 121,
where (t
2 - 'II ).
I- h
and 'I = /3 [ V /3" + 8). - /3] / 4 Free

/3 = PI'! (reler to figure A) CONSIDERED PLATE

). = hi'! (refer to figure A) a ITS LOADING

For the wide flange columns and beams used in prac

tical connections, it has been found that (1 varies within
the range 3.~ to ~. FIGURE A

T~BL E 9
COMPA RISON Of THE THREE COMPRESSION SIDE CRITERIA
Web Thickness, W Stiffener Thickness, I,
Specimen Idea lized Plast ic Mod. Plas. Actual Ideali zed Plastic Mod. Plas. Actual Remarks
A I 0.666 0.)04 0.624 0.284 Column web buckled
A2 0.428 0.440 0.4~0 0 .~ 87 Colum n web 0 K_
A4 0.470 0.4)3 0.480 0.417 Co lumn web weak
A~ 0.378 0.420 0.412 0.) 80 Column web OX.
8 -6 0 .2~ * 0.326 0.297 0.43 7 Stiffened connections
B-8 0.2) * 0.261 0.2)* 0.2)0 satisfacto ry
C9 0.382 0.429 0.340 0.437 Connections O .K. but
C- II 0.34** 0.39** 0.34** 0 .2~0 some stiffen er buckling
0 12 0.34** 0.34 0.34** 0.606 Connection O.K.
H1 0.666 0 .~ 04 0.624 0.600 (I)
AA 0.474 0.44) 0.479 0 .39~ Connect ion O.K.
BB 0 .2~ * 0 .2~ * 0.2)* O. ~ Connection O .K .
DO 0.34** 0.34** 0.34** 0.6 Connection O .K.
(I) Column web O.K. up to O.95t\1 p when fai lure occurred in len.sion region of conneClion .
~termined by .sienderne.s.s limi!ation.s. Equation (6) .
Determined by .slenderne.s.s hmilation.s, EqualJon (8) .
Seat 4" above compre.s.sion flange. Stiffening also included a pl ate perpendicu lar 10 the .seat- .stt Figure t) .
32
 As a conservative approximation, take (1 = 3.5 Force carried by central rigid portion = UII I "'

Then p. = 3.5 Clllle2 Hence Q, = ulIl m + 7 Ullle2

Hence capacity of two plates is given by 2P. =7 ulIl/

2. Two-woy T I.
2 .1 Summary of Coupon T ls

E CI/'L 'I,v 'I,L 0 .. 11. E,I.

Shape Mark ksi ksi ksi ksi kSl in /i n.
y\.' plate ~9 E/8/31 30,000 35.6 34.8 59.2 1.5 X 10 - '
59 E/5/31 29,500 35.8 34.2 59.6
59 E/2/31 30,200 35.6 34.6 60.0
\/{' plate 68 E/6/31 30,000 33. 1 32. 1 56.0
%6" plate 48/9/31 29,900 38.2 37.2 62.5
48/3/31 31,700 38.2 37.8 61.3
\/2" plate 68 E/6/1( 29,800 24.1 32.8
68 E/6/2( 30,600 26.7 33.6
12 WF 40 38 G/II/ 35.2 36.9 37.3 62.0 1.66 X 10- 2
38 G/U/ 34.3 36.3 36.5 61.7 1.7
38 G/31w 42.8 44.0 42.8 65.4 2.02
38 G/41/ 36.6 38.3 37.6 61.9 1.9
8 WF 31 54 E 31/11/ 34.7 39.4 37.8 63.4 1.72
54 E 31/21/ 36.3 38.1 63.0 1.94
54E 3 1/3 1w 35.4 39.7 38.3 63.0 1.98
16 WF 36 53 E 939/11/ 33.5 40.8 40.0 61.7 2.16
53 E 939/21/ 38.2 39.5 6 1.8 2.22
53 E 939/31w 41.4 43.5 42.7 64.5 2.17
53 E 939/41/ 39.6 39.2 6 1. 2 1.94
8WF67 54 E 67/11/ 32.4 32.2 61.4 1.18
54 E 67/21/ 28.5 35.2 34.6 61.9 1.25
54 E 67/31w 38.8 37.7 60.6 1.94
54 E 67/41/ 34.1 33.2 61.3 1.44
12 WF 99 55 E/21/ 31.3 34.6 34.5 62.5 1.31
55 E/41/ 34.3 36.7 35.8 63.7 1.41
12 WI' 65 42 E/llf 37.2 36.4 62.0 1.61
42 E/21/ 36.4 36.1 62. 1 1.55
42 E/31w 40.6 38.8 61.5 1.43
42 E/41/ 37 .1 36.1 62.2 1.48
For WF membe'rs E is in range U,OOO < E < 30,000 ksi.
c = compression coupon.
1/ = tension flange coupon.
l UI = tension web coupon.

2.2 Calculations for Design of Specimen

Coillmn}. Assume L/ r =
72. P/A = 17,000 - 0.485 (L/ r)2
Then from AISC Manual Column Working Stress = 14.5 ksi
33
Slruclural Shape DelaiiI
Area as
Column Area measured Pw kips 1. 6~ Pw 2 Pw Test No.
S WF 31 9. 12 9. 01 132 218 264 AI , B6, C9
8 WF 67 19.70 19.94 286 472 ~72 A2
12 WF 40 11. 77 11. 31 171 283 344 B8,Cl l ,D I 2
12 Wl F 6~ 19.11 18.66 278 4)9 ~)O A4
12 WF 99 29.09 28.4~ 422 696 800k" M
AISC M:lOual valliC' .
Test ing ml chi ne capacity = 800 kips.

AnalYJis 01 Beams and Beam -Io-columll Flange Welds: W eld requi red when 1," == 0.6 klJJS per ~~G" leg of
All di mension of sections as measured on specimens fill et is Ylr.".
Total th roat area of \1.1" fill et welds actually used is
Beams: 16 \\'I F 3:'; (T ,V == 20 ksi
g reater than tota l area of web, so the Y4 " weld should be
Bending: able to ca rry any forces that the beam web can carry.
"'to == 01(; 5 == V . L Butt weld of fl anges can ca rry :t ny fo rces the beam
fl anges can ca rry.
V _ "" S
20 X )6.4
w - L = 23.) kips
48 Shear:
", ~ VI' (in plmffc range ) == 18 flit/Ill
V, = - - (I II (avg . for 16 \'(' F }6 ) == 39.6 k:;i =
18 X 0.29 X 1).91
'-
83 kips> 5n ki ps
39 (, X 5(,.4
V, = =46.5 kips 1'" U ledlCled) = 52.5 kips
4~
V ( 111 e1aJlic rtlnge ) == 13 wd
1(1

7. I'll
= 13 X 0 29 X I 5.9 1
V, -
- -L-
= 60 ki ps > 23.5 kips (O.K )
39.6 X 63 .76 .
48 = 52.5 kIps Infl uence of shear on VI' may be neglected if

EI4JI;( Analysis oj If/ eld! al Beam W orking Load:

Use butt wel ds on flanges and fillet welds on web.
Web fillet welds carry both shear and bending forces Z Aw
i.c.,
of web. L < \1 ,
U," == 20 ksi d" = 1) .9 1 - 0.86 = 1) .05"
63.76 14.91 X 0 29
Maxi mum bending stress in web
48 < I. 732
15.05
= -- X 20 = 189 ksi 1.33 < 2.S (O.K .)
15 .9 1
Bending moment of web == u 510 lAteral Buckling: elastic range
18.9 X 0.29 X (1 5.0)' Ld 96 X 15.91
207 inkips 500 < 600 (0.1( )
6 bl 7.09 X 0.43 1
Length of each weld = 13" whence U alto.. == 20 ksi
Mc 6M 3 X 207 LOC4/ Blfckling el ast ic ra n~e-See Secti on 18 (b) of
,~ = - 1- = 2 L' = (1 3) ' = 3.67 kips/ inch AISC Specification

I, = -
V"
L
= - --
23.)
2 X 13
=0.922 kIpS/
. .
Inch Actu al -
b
I
7.09
= --
0.43 1
= 16.45 < 32 (O.K .)

I= 'oj 1m' + I,' - 3.79 kips/inch To rea.ch strain hardening btl < 1 7( 8)

34
 Therefore beams are not critical fo r local fl ange buck 23 MATERIAL OIMENSI('f>;S 4f ''''' PROP(RTI[:; - aw . ra , . "t a luet
ling jn plastic rangl!.
T o reach strain ha rdening di 'l! ~ 55 (8)
d 15.9 1
_ == - - == 54.8
IV 0.29

Therrfore beams are not critical for local web buckling

in plastic range. 1 0.9

VL' I A. I)
~.l!f.1l.
D~flerljol1J: S el lltic == -- - assum mg com plete restrai nt. col umn ( 8 G)
3 EI (C 91

8 --'--
V L' v, == 46.5 kips
, - 3 EI L == 46"
E == 30 X 10'I.si
J= 44~ 9:') JO"
8 09
46.5 X ( 18 ) '
3X 30 X 10" X 44896

S u,t. = -
o.lr"
52.~
-X 0.127 == 0. 141 ' I'deaI'Ize d
. " assuming
e WF 67 ' A 21
CO Lumn

r= 903"

~.
46.5
u - E and AI - 4> relationship.

In nondimensia nal form : 1----.-,..

At yield
v
--- 1 - -
0
V~ - - S" ~ (A- 'U
Coi llmn I

At ulti mate
v. !P"I
--== I '213-
V, i5.5
Poro ll ol floI'I , "

8~am R Ol aIIOIlJ:

T he rotation of the beam can be expressed as a change

in slape o.f the point o.f Io.ad applicatio.n with respect to. 11.1111'
the connection assuming the latter to develop complete !2 WfH (A- 5)
restraint CoLli',,"

Applying the moment area theorem :

Po'OlI o L flon , ..
12.10-

VfJL2 2 EI
T herefore
8 jLe : d =2E1 and V 1 1
- -
-L'! 0/I 12 .00

12 Wf 40 (I - I )
3 E1 Col"".n (C III
But V, == L' 8,
o.)()"
"0 ' 011 . ' flGn , ..
3 EI >8, ILt4-
Therrfore 8fJ -- -L'!0" ' -
2 EI 2L

== 0 31 BfJ radians
35
2.4 Shape Properties z
- = (1.283) (7.766) = 9.96
16 WF 36 B.amJ:
4 (0.184) (7.)42 ) = 1.39
3.40 0.145 (1.1))) (3.978) = 4.)9
1).94
_ 1).94 Z = 63.76 in.'
Yo J = - 6- = 6.08"
2. 22

7.955
t '0 = 7.96" - 6.08" = 1.88"
B IjVF 31 Columll: A = (2) (0.4 30 ) (8.09) = 6.96
(7.22) (0.284) _ 2M
9.0 1011
A
"4 = (3.4) (0.377) = 1 283 B ljVF67 Column: A = ( 2) (0.941) (8.36) = 1) .7 )
(0.108) (3.4) /2 = 0 .184 ( 7.148) (0.)87) = 4.19
(0.14) (7.96) = 1.1 )) 19.940"
2.622 12 WF6J Columll : A = (2) (0.)94) (11.88) = 14. 10
A = 10.488 " (10.942) (0.41 7) = 4.)6
18 .66 0'1
I .. = (1.283) (7.766)' - 77.40
4 (0.184) (7.)42)' - 10.46 12 WF99 Column: A = (2) (0.)94) (11.88) = 22. 10
(0.14) (7.9)))'/ 12 - 6.11 (10.9)8) (0.)8) = 6 . 3~
(1.1))) (3.978)' - 18.27 28.4) 0 "
112.24
I .. = 448.96 in.'
12 WF 40 Column : A = (2) (0.498 ) (8M) = 8.03
In 44M .96 .
(10.944) (0.300) = 3.28
S= -
C
=- -=
7.9))
%.4 In.' 11.31 "

3 . Four-way Tests
3.1 Summary of Coupon Test s
E a,. a,. (I f u.u. ..
Shape Mark ksi ksi ksi ksi ksi in/in.
';4" Plate 233 / P 30,900 40.6) 39.87 62.00 0.0172)
2),100 41.02 39.74 6U4 O.ot77)
12 WF6) 233/ Wl 30,400 42.)7 41.81 67.74 0.01)
233jWl 29,700 38.37 38.)4 67.74 0.0067)
233/ Fl 30,100 40.63 40.07 6).)7 0.0187)
233/ Fl 30,600 44.28 40.46 64.86 0.200
12 WF 40 233jW2 31,200 47.17 44.16 W.8) 68.93 0.021
233/ W 2 30,700 )0.00 48.86 43.60 70.87
233/F2 31,300 43.47 41.77 37.86 68.00 0.017)
233/ F2 29,400 42.90 4Ul 37.67 68.4) 0.0187)
16 WF 36 233jW3 29,)00 )0.)8 48.9) 63.63
233jW3 30,600 47.00 4) .66 61.64
233/ F3 30,400 41.86 40.2) 61.18 O.ot8~
233/ F3 30,200 40.)8 38.98 )9.99 0.021 )
12 WF 27 233 / W 4 31,200 43.70 43.70 38.81 61.62
233 / W4 31,100 4).14 41.89 37.83 61.02
233j F4 31,100 40.36 38 . 6~ 34.74 61.24 0.017)
233/ F4 29,800 39.36 38.17 33.79 60.03 0.0207)
IV-Wd>
F - flange.
36
3 .2 Calculations for Design of Specimens 19 = 24 X 0.24 X (n + 0.8!})
ColumnJ: 19 - 4.68
As in S<cti on 2.2 of Appendix - 11 = 2.5 Inches reqUi red bearing length
5.75
Column Working Stress = 14.5 kSl
From Table 25 in the AISC Structural Shop Draft ing
SI,"(/",al Shape D elaiiI: Tex tbook, Vol. 2, the chOICe is:
Area as 4" wide seat ; Y4" fill et welds ; L = 7" ;
Test Column Area measured p.. 1.65 p.. 2 p.. Plate thi ckness Y2"
AA 12 WF 65 19.11 \9.00 276 455 552
B8 12 WF 40 11. 77 11. 70 170 280 }40 T op Plait Weld DtIigl1 :
DO 12 WF 40 11. 77 11.49 167 276 }}4 57.2
Required plate thi ckn ess - - - - = 0.3"
AISC Manual val ue. 20 X 9.75
Use Y2" Plate
IInalyIiJ of Beami and Beam-lo-coilimn Flange W eldI: The length of weld available is
All dimensions of sections as measured on specimens 9. 75 +
2 X }.75 = 17.25"
a., = 20 ksi Using butt welds on the plate, the full strength of the
plate can be developed.
Bending :
W eld Connerling T op Plait 10 Beam Flangn :
Af.., = a..,S = V.., L
Fillet welds of YI" size can be applied to toe of
u,S 12 WF 27 flange, and Y2" fillet welds can be applied on
V, -
- -
L edge of top plate.
Working stress for Y2" fillet is 4.8 kips/ in.
Z = plastic modulus Working stress fo r YI" fi llet is 2.4 kips/ in.
Length of weld ava ilab le = 6" overhead fi llet.
The calculations are similar to those in Section 2 of 6.5" fi llet on top of flange .
this Appendix. Lateral buckling, local buckling, shear,
deflections and beam rotations were investigated and cal Safe load = 6.5 X 4.8 + 12 X 2.4
culations are similar to those found in Section 2. = 60 kips> 57.2 kips
IInalYIiI of WehlJ fo, Sperimen BB
(Actually used 12.5 inches of Ys" fillet .)
12 WF 2 7 BeamI : Cherk on T tt Seal :
Us< working load and allowable working stresses for From Lawson's chart on Page 12} of the Aireo "Man-
the design of welds, seat, stiffener. etc. . ... ual of Design for Arc Welded Steel Structures'- 181
V .. = 19 kips AI .. = 19 X }6 = 684 m.klps
684 . R= ;i2~};;
.0;;4,,;D~L;,;'
T = C = - - = 57 .2 ki ps \/ L' + 16e'
11.95
AISC Specificat ion Section (26h) : where D = :x 6 " L = 8" t = 2.75"
R R = 2}.O4 X t . X 64 = 20.} kips > 19 (O.K.)
-,----:--:- = 24 ksi
I (11 +
k) \/64 + 16 (2.75 ) '

37
3 .3 Material Dimens ions and Properti es
In the figure below the average values of all the dimen -
sions of th e wr secti ons used in the tests are shown. The
ca lculations of the section properti es are simi lar to that
presented in Section 2 of the Appendix. In the Table
~I ow the different section properties arc shown:

SECTION PROPERTIES
Sec.llon Plast IC
T est Deam Area Modu lus Modulus
AA 16 \XI I' 36 1028 55 .59 62 .73
BB 16\X1F36 1029 54 20 61. 52
12 WF 27 783 32.60 36.56
DD 16 WF 36 10.24 54 .06 61.37

036" I
TEST ... .

t
]C
121)-

,z'
0395
602'

0282"

... 1Qt ..,

159'

12 "' 65 I' ''' 36

"Jr
TEST BB

"-
Jr"
~6,r
12 W' 40 7" -I 034 "0 4 ' 1 IZ "' 17
o so
IS'*" 36 I

TEST 00

"I 0317"

.... . - -j
IZ'ttF 40
WF" _SECTIONS. ACTUAl.. DIMENSIONS

FOUR-W.'" CON NEC!!QNS

38
 Bibliography
I . A. N. SHERBOURNE. C. D. )ENSEN- "Dmm JY/ elded Beam Beam! and F,ami'! oj S'rucluraJ Steel" (Dissertation), Le-
Column Connenionl"- Fritz Laboratory Repo rt No. 233.12, high University, 1958.
Lehigh University, 1957. 5. E. W. PIIRKES- " T ht Slreu DiJlriblltion Near a Loading
2. R. N. KHA88AZ , C. D. )FNSFN- "Four-u'JY IY' e/ded fnlerior Poinl in a Uniform Flanged Beam"-Phil. Tran;. 0/ the
Beam Column Connt'(/;on!, '- Fritz Laboratory Report No. Ro)'o/ SorielY, Vol. 244A, 1952.
233 .13, Lehigh University, 1957. 6. G. HIIIIIJER- " Plale BllckNng in the Strain Hardening
,. L. S. BEEDLE, B. THURLIMANN and R. l. KFTIF.It- " PlaJlic Range"-Proc. ASCE, 83 (EM-2) p. 1581, April 1958.
DeJign in SIrJUiura/ SueJ"-lehi,gh University, Bethlehem, 7. R. H . Wooo--"Yield Line Tht'orj'-Research Paper No.
Pa ., and American In stitute o( Steel Construction, New York, 22, Building Research Stat ion , 1955.
19~~.
8. L. GROvER- "Manllal 0/ Dt'Jign lor ;1rc W t'ldt'd Stut
4. G. C. DRISCOLL. }R.- "Ro/alion Capacity Requirement! lor Strllctllres", Air Reduct ion Co., New York, 1946.

Acknowledgments
The enllre prop:rlm was carried oul at the Fritz Engineering them by members of the rese:Jrch staff of the Fritz Laboratory,
Laborat ory of Leh i!;h University, of which Professor W. J. Eney p:Jfticu larJy to Or. L. S. Beed le, Chairman of the Structural
is Director, with funds supplied by the American In stitute of Metals Divi sion, to Or. Bruno Thiirlimann and Or. G . C. Dris-
Steel Con struction. The authors are indebted to members of the coll, Members of the Structural Metals Division, to Mr. I. J.
Research Commi ttee on Welded Interior Beam-Io-Co lumn Con- Taylo r, Mr. O. Darlington and Mr. R. Clark who were respon-
nections who gave inv:Jluable advice, techni cal and ot herwise, in si ble for the instrumentation, to Mr. K. H arpel and his assistants
the o rg:Jnlz:Jtion :Jnd execution of this project. This committee who helped organize the actual tests, and to all those members
included Messrs. F. H . Dill, Chairman, L. S. Beedle, Edward R. responsible for the preparat ion of this manuscript.
Estes, Jr., T. R. Higgin s, C. L. Kre id ler and H. W. Lawson with The helpful advice and guidance given by members of the
Mr. Jonathan Jones acting as an Advisor. Steel Structures Committee of the American Welding Society is
The authors also wish to acknowledge the assistance given greatly appreciated.

39
c
.z-
.::.
PUb. No. Tl05 (2M-5I9O)