1524

AseE Annual and NationalEnvironmental Engineering Meeting

October 18-22, 1971 • St. Louis, Missouri $0.50

COLUMN WEB

STRENGTH IN STEEL

BEAM-TO-COLUMN

CONNECTIONS

W. F. Chen, A. M. ASCE andD. E. Newlin, A. M. ASCE

Meeting Preprint 1524

This preprint has been provided for the purpose of convenient distri­bution of information at the Meeting. To defray, in part, the cost ofprinting, a meeting price of SO¢to all registrants has been established.The post-meeting price, when ordered from ASCE headquarters will be50¢ while the supply lasts. For bulk orders (of not less than 200 copiesof one preprint) please write for prices.

No acceptance or endorsement by the American Society of CivilEngineers is implied; the Society is not responsible for any statementmade or opinion expressed in its publications.

Reprints may be made on condition that the full title, name ofauthor, and date of preprinting by the Society are given.

COLUMN WEB STRENGTH IN STEEL BEAM-TO-COLUMN CONNECTIONS

by W. F. Chenl and D. E. Newlin2

Associate Members, ASCE

ABSTRACT

In the design of an interior beam-to-column connection,

consideration must be given to column, web stiffening. The pre­

sent AISC Specifications require stiffening of the compression

region of column web on the basis of two formulas. The first

formula defines the strength of the compression region as a func-

tion of web and flange thickness and the applied load from the

beam flanges. The second formula precludes instability on the

basis of the web depth-to-thickness ratio. If stability is the

more critical, web stiffening is required regardless of the mag-

nitude of the applied load. Results of previous and the present

investigation indicate that both formulas are conservative.

This report is a further examination of the criteria for

stiffening the column web opposite the beam compression flange(s).

In the experimental program this compression region is simulated

in a manner allowing rapid and easy testing of specimens. A for-

mula is developed for predicting the load carrying capacity of the

compression region for sections in the range of instability. More-

over, the effects of strength and stability are combined into a

single formula. Simulation tests are also made to investigate the

effect of column flange thickness and less common loading condi~

tions on the strength and stability of the compression region.

lAssociate Professor of Civil Engineering, Fritz EngineeringLaboratory, Lehigh University, Bethlehem, Pennsylvania.

2Teaching Assistant, Department of Civil Engineering, LehighUniversity, Bethlehem, Pennsylvania.

i

1. I N T ROD U C T ION

IIi. the present AISC Specification [1] there are two

formulas giving the requirements for stiffening the compres-

sion region of a column web at an interior beam-to-column

moment connection. (Fig. 1 illustrates this region). For-

mula (1.15-1) in Ref. 1 or Eq. 8.21 in Ref. 2 gives the strength

a column web will develop in resisting the compression forces

delivered by beam flanges as

(1)

This equation was developed from the concept that the

column flange acts as a bearing plate as illustrated in Fig. 2.

It distributes the load caused by the beam compression flange

of thickness t b , to some larger length t b + 5k at the Hedge"

of the column web. The distance from the beam flange to the

edge of the column web is .k (Fig. 2). In Ref. 6 the stress

distribution at the distance was developed by curve fitting of

test results on steel having a yield stress of 36 ksi. In

Ref. 5 the formula was shown to be.conservativE:..

The application of Eq. 1 is limited by the" AISC Specifica-

tions to cases where the column web depth-to-thickness ratio,

dclt, is small enough to preclude instability. The limiting

ratio is defined by the formula

d cT

180I(J

Y

-1-

(2)

This formula can be derived using the concept of simply

supported edge conditions for the column web panel with an

elastic solution for the buckling of a simply supported long

plate compressed by two equal and opposite forces [5]. The

test results of Ref. 5 and Ref. 6 show Eq. 1 to be conserva-

tive for all grades of steel and for all sections tested re-

gardless of dc/t. (The test set-up is shown in Fig. 3). The

present AISC Specifications do not permit consideration of

any load carrying capacity in the compression region of sec-

tions with d /t ratios greater than 180/ra-. Development ofc y

a method of determining ultimate loads for such a condition is

one of the objectives of this report.

Of course, strength and stability are interrelated. Thus

an additional objective will be to develop a single formula

for predicting the ultimate load carrying capacity of the com-

pression region regardless of the dcft ratio of the column

section.

Within the compression region, the column flange simulates

a shallow continuous beam. The bending stiffness of the column

flange is primarily a function of its thickness. It is the

third objective of this report to investigate the contribution

of the column flange as a shallow beam to the load carrying

capacity of the compression region.

occasiona~ly, the opposing beams of an interior beam-to­

column moment connection will be of unequal depths. This may

-2-

result in a situation where the loads applied to the compression

region are eccentric (Fig. 4). Investigation of the effect of

this type of eccentricity on the strength and stability of the

compression region is also included in this report.

2. A N A L Y TIC A L MET HOD S

A complete elastic-plastic analysis ofa beam-to-column

connection using the finite element method has recently been

reported by Bose [4]. Both buckling and ultimate strength

solutions are obtained.

The finite element approach is important to the under­

standing of the behavior of the connection. However, a

practicing engineer is not likely to attempt the use of it in

the design of steel structures. Additionally, there remain

the questionable areas of boundary condition stress, residual

stress, and degree of accuracy.

In contrast to the very rigorous approach it is current

practice to accept the results of physical experiments coupled

with drastically simplified statical analyses as a basis for

design rules. This is, indeed, a logical approach for practi­

cal use.

It is evident, however, that, to gain the accuracy needed

for more effective and efficient connection design, large amounts

of experimental data would be required. The quantity of tests

-3-

needed to cope with all of the variables affecting connection

behavior makes further pursuit of this approach unattractive.

A compromise between the crude and rigorous approaches,

that optimizes the benefits of experimental tests in combina-

tion with certain idealized theoretical aspects of,the problem,

is the essence of the approximate approach proposed herein.

The problem will be tre~ted as an elastic plate with proper

interpretation of boundary conditions for the inelastic range.

The desired effect is that of increasing accuracy while re-

taining simplicity for design use.

3. D EVE LOP MEN TSTRENGTH

o F B U C K LIN GFORMUL'A

One of the major contributions of the column flanges is

to provide lateral support to the edge of the web panel. The

flanges provide web edge supports because of the very high

bending stiffness of the flange in the plane of the flange.

The flanges provide what accounts to simple support with 36

ksi material because of the early yielding near the juncture

of the web and flange. It was observed that further t4is

local yielding does not spread throughout the compression

panel until just prior to ultimate load when the panel begins

to buckle. With the use of high strength materials this local

yielding can not occur prior to ultimate load and the flanges

therefore closely simulate the role of fixed end supports for

the web panel.

-4-

From observations of the test results in the present

and previous tests, it appears justified to assume that the

concentrated beam-flange load acts on a square panel whose

dimensions are dc x dc . For this condition the critical buck­

ling stress becomes

Pcr 33,400ocr = dct = (d It)2

c

(3)

as developed in Ref. 5.

In Ref. 7 the buckling -load of a fixed end long plate

compressed by two equal and opposite forces is twice the buck-

ling load of the same plate when it is simply supported. It

was also observed in previous tests [5] that sections made

of 100 ksi yielding stress steel having dclt ratios greater

than Eq. 2 did attain stresses approaching twice the critical

stresses predicted by the simply supported theory. This is

illustrated graphically in Fig. 5.

It can, therefore, be stated that cr cr = 33,400/(dc/t)2

is a lower bound for 36 ksi material and 20cr is an upper

bound for 100 ksi material.

The above conditions can be closely approximated for all

(4)cr cr

grades of steel by making ocr a function of cry as follows

f(5p 33,400 (----y)

dct = (d It)2 6c

If th~ e~pression for Ocr in Eq. 4 is adjusted to fit

the most critical test (test point marked No. 21 in Fig. 5)

-5-

(5)d c

pcr

the resulting equation

4100 t 3 10y

will be conservative for all grades of steel and for all

dc/t ratios. Comparision of Eq. 5 with test results is shown

in Fig. 6 using nominal values of yield stress. A good agree-

ment is observed.

4 • D EVE LOP MEN T' OFT H EI N T ERA C T ION FOR M U L A

Fig. 7 is a non-dimensional comparison of test results

and the AISC design formulas for strength and stability.

There are a number of observationa that may be made.

First, when the dc/t ratio exceeds the values (180/;a;)

the specification implies that the section must be stiffened

regardless of the magnitude of the applied load. Some speci-

mens obviously have a load capacity that is even greater than

required.

Second, the test data seems too scattered to permit an

accurate prediction of the ultimate load capacity. It is

evident that the strength formulation does not describe what

really occurs in the column compression region of a beam-to-

column connection.

If, on the other hand, one assumes that the compression

region of the column is effectively a square web panel with

-6-

-----_ __.._. __ ._ _-_ _-_ __ __ ._.- --------------------~~~----

dimensions d c x dc' the data is considerably less scattered

(Fig. 8). In Fig. a this interaction between strength and

stability can be conservatively described by the straight

line in the form:

(J 3/2

P = (1.75 cry) dct - ( lao ) d~ (6)

It was found that Eq. 6 is in good agreement with results

of all tests except on those specimens made of 100 ksi material.

For these specimens the axis intersection points in Fig. 8

would have to be at least 2.2 or 2.3.

If the constant 1.75 was adjusted so that it was a

function of the yield stress, this would accomplish the de­

sirable effect of shifting the interaction line upward for

higher strength steel. Changing 1.75 to 1.75(;0-/136) givesy

an unconservative prediction for 50 ksi materials. Changing

1.75 to 1.70 (to/ t3'6) provides the desirable effect. They

interaction equation takes the form:

(JZ5/4 cr- 3/2a2P = d t - (-tao-)( 1.44) c c

which reduces to, for example

p 61 d t - 1.2 a2 for cr 36 ksic c y

P 92 d t - 2.0 d2 for a 50 ksic c y

P 128 d t - 2.9 d 2 for cry 65 ksic c

and P 220 d t - 5.6 d 2 for cr 100 ksic c y

-7-

(7)

(7a)

(7b)

(70)

(7d)

t

When the formula is solved for t it takes the form:

d 2c ;a; + 180 Cl Af

125 d c to:; (8)

this can be similarly reduced to, for example

(8a)td c C

lA

f51 + l.7d for cry = 36 ksi

c

where Cl is tne ratio of beam yield stress to column yield

stress and Af

is the area of the beam flange delivering the

concentrated load, P. Thus, Cl Af = P/cry • Then t becomes

the required web thickness in the column compression zone

regardless of de/t.

The predicted ultimate loads from Egs. 5 and 7 for recent

Lehigh University tests are tabulated on Table land compara-

tively plotted against test values in Fig. 9. Fig. 9 shows

that the interaction equation (7) is as accurate as the sta-

bility equation (5) and for 100 ksi material the interaction

equation (7) provides better accuracy than the stability equa­

tion (5). In the range where the stability equation is not

applicable, i.e. dc/t < l80/~, the interaction equation (7)

is compared with AISC predictions in Fig. 9. Where the

stability formula is applicable, AISC makes no prediction.

-8-

5. DES C RIP T IONa F S PEe I A L T EST S

5.1 TEST PROGRAM

-Fifteen tests were performed to explore web behavior as

influenced by various types of loading conditions and column

flange variations. Figure 3 gives a schematic of the web

crippling test set-up.

The first series of tests simulated the compression

zone of a column loaded by moments from two opposing beams

of unequal depth. This is illustrated schematically in Fig. 4.

To observe the effect of increased column flange thickness,

tests were performed on specimens with and without COver plates.

These plates were I inch thick, 20 inches long, and slightly

wider than the specimen flanges (to permit fillet welding all

around) .

In a separate test series the role of the column flange

as a continuous stiffening beam was studied. The column

flanges were slotted from the outside edges to the web on

both sides of the load points. The ends of the cuts near the

web were pre-drilled to avoid sharp notches. On one test the

distance along the column between the slots was equivalent to

t f + 5k. On another specimen the distance was d - 2 t f where

t f is the flange thickness and d is the specimen depth (see

Fig. 12).

-9-

5.2 TEST PROCEDURES

The test set-up permits rapid testing of specimens. It

is basically the same one used by Graham et al [6], and is

shown in Figs. 2 and 3. Essentially a simulation test, a

short length of column is placed horizontally between the

loading platens of a testing machine and is then compressed

at the top and bottom faces of the column. The bar is tack­

welded to the column flange to simulate a beam flange framing

into it at that point.

The specimens were tested in the 800 kip mechanical

machine at Fritz Laboratory. The instrumentation consisted

of dial gages to monitor the deflection in the direction of

the applied load and another dial gage to monitor the lateral

deflection of the column web. This lateral deflection indica­

ted the stability of the column web.

6. RES U L T S

Table 1 summarizes the measured properties of all test

specimens not only those described in the previous sections

(WIO to W-21) but also including the tests reporte~ in Ref. 5

(W-3 to W-9). Table 1 also summarizes the test results and the

theoretical predictions.

6.1 ECCENTRIC LOAD TESTS (Simulation Beams of Unequal Depth)

Fig. 10 shows the load-vertical deflection diagram (p-~

curves) for three specimens (Test 1, Test 2, and control test)

-10-

with three different eccentricities (10 in., 5 in., and O).

From the curves it is seen that the ultimate load is essen-

tially unaffected by the eccentric load condition. Eccen­

tric load has the effect of adding an additional amount of

stiffness to the web. Design based on the assumption of equal

depth beam condition (zero eccentricity) will be conservative.

6.2 INCREASED FLANGE THICKNESS

Fig. 11 shows the load-deflection diagram (p-A curves)

for two such tests, one for a plain beam with no cover plate,

and the other for the same section with 1 inch cover plate.

Referring to the plain beam test, it is seen that from no

load to approximately half of ultimate load the curve is al­

most linear, and the upper half of the curve to ultimate load

is at a lesser slope indicating the occurence of some yielding

and redistribution of stresses from the· increasing load. The

maximum design load, as determined by the quantity (tb + 5k)tcry ,

shown in Fig. 11 by the horizontal line is reached soon after

the' occurence of some yielding on the load deflection curve

with considerable reserve capacity remaining.

The addition of the one inch cover plate made a significant

difference. The p-A curve is essentially linear all the way to

ultimate load indicating no occurence of significant yielding.

The load corresponding to (tb + 5k)toy formula (one inch cover

pla~e is included in k dimension) leaves only 4.8% of ultimate

-11~

load as reserve capacity as compared to 43% in the WID x 29

control shape without cover plate. For tests of the W12 x 27

shape these figures are 4.8% and 33% respectively. Although

the flange thickness is tripled, the ultimate load in increased

only by a factor of 1.3.

6.3 SLOTTED FLANGE TEST

Fig. 12 shows the load-deflection diagram for three such

tests, one for a control test with no flanges slotted, and the

other two for the same shape with the flanges slotted and the dis­

tance along the column between the slots being d-2 t f and tf+Sk

respectively. From the curves it is seen that the ultimate load

is decreased only slightly by slotting the flanges. Slotting the

flanges has the effect of adding a small amount of stiffness to

the co1umb web during the early stages of loading but weaking the

web stiffness when the column web is significantly yielded. Spe­

cimens at the end of the test are shown in Fig. 13.

7. DEFORMATION CAPACITY

When the column web has the requisite strength the

desired rotation capacity of the connection is supplied jointly

by the column web and the end portions of the beam. In Ref. 3

it is developed that the required rotation at the ends of a

fixed ended beam uniformly loaded along its length, so that it

will be able to form a mechanism, is Mp~/6EI. In the practical

case of a Wl6 x 36 beam spanning 24 feet the required rotation

is calculated ~n Ref. 3 to be 7.2 x 10-3 radians. If this

-12-

deformation was to be absorbed entirely by an interior column

web, the required compression deformation would be 2 x 8 x

(7.2 x 10-3 ) or about 0.12 inches.

All the tests reported herein the specimens can develop

the required deformation.

8. SUMMARY AND R E COM MEN D A T ION S

8 • 1 PARAMETERS

It has been' shown that the parameters most pertinent to

the strength and stability of the column compression zone in

a beam-to-column connection are four fold. They are web

thickness, t, column depth, dc' yield stress, cry' and the

role of the column flanges as supports for the web panel. The

column flanges vary in their support effect from a lower bound

of simple edge supports for structural carbon steels to an

upper bound of fixed edge supports for high strength steels.

8.2 FORMULAS

The formulas developed or under consideration in this

study are summarized below.

d c vcr;Strength governs when t > --rao-

d I(J. Stability governs when t < c 180y

-13-

Ultimate Load Form Design Form_._---

Strength Strength

p = (tb + 5k)tcry

Cl Af AISC

t ~ tb

+5k (Eqs .1&2)

Stability Stability

P = 0 Stiffener Required

Strength & Stabilitycr 5/4 cr 3/2

P = (~) dct - (~) d;Strength & Stability

d~ ~ + 180 C1 Aft < -----""---:----~-

- 125 to dy 0r--~--~~~---~--~-~~-~-~~--_~_.~~~_~~_~ ~_;....~~~~

Interac­tion(Eqs. 7&8)

Strength

Stability

4100 t 3 ,,;-a-p = y

d c

Stability

d > 4100 t3

o - Cl

Af~

combina­tion ofAISC andBucklingFormula(Eqs.2&5)

The present AISC formulas are conservative. This has

been shown previously and is reconfirmed in this report. The

AISC formulas are perhaps incomplete in that they offer no

estimate of the load carrying capacity of the compression

zone when dc/t exceeds lBO/va;. The stability formula given

by Eq. 5 can predict the load carrying capacity of the column

web accurately when dolt exceeds l80/~.

The interaction formula given by Eq. 7 has the advantage

of being a conservative fit to pertinent test data. Another

-14-

advantage is that it makes possible a one-step analysis of

the compression zone of a connection to determine whether or

not a stiffener is required.

9. ACKNOWLEDGEMENTS

The work is a part of the general investigation on

T1Beam-to-Column Connections," sponsored by American Iron and

Steel Institute and Welding Research Council (WRC) at the

Fritz Engineering Laboratory, Lehigh University.

Technical advice for the project is provided by the WRC

Task Group on Beam-to-Column Connections, of which J. A.

Gilligan is chairman.

The writers are especially thankful to'Messrs. W. E.

Edwards·, A. N. Sherbourne, G. C. Driscoll, and members of

the WRC Task Group for their comments, to Messrs. J. A.

Gilligan, T. R. Higgins, and L. S. Beedle for their review

of the manuscript, to G. L. Smith for conducting tests and

reduction of data, to P. A. Raudenbush for her help in typing

the manuscript, and to J. Gera for preparing the drawings.

-15-

10. REF ERE NeE S

1. AISCSPECIFICATION FOR THE DESIGN FABRICATION, AND ERECTION

OF STRUCTURAL STEEL FOR BUILDINGS, American Instituteof Steel Construction, February 1969.

2. ASCE and WRCPLASTIC DESIGN IN STEELS - A GUIDE AND COMMENTARY,

The Welding Research Council and the American Societyof Civil Engineers, ASCE Manuals and Reports on Engi­neering Practice No. 41, second edition, 1971.

3. Beedle" L. S.PLASTIC DESIGN OF STEEL FRAMES, John Wiley and Sons,

New York, 1958.

4. Bose, Somesh K.ANALYSIS AND DESIGN OF COLUMN WEBS IN STEEL BEAM-TO­

COLUMN CONNECTIONS, Ph.D. Dissertation, Departmentof Civil Engineering, University of Waterloo, Canada,March 1970.

5. Chen, W~ F. and Oppenheim, I. J.WEB BUCKLING STRENGTH OF BEAM-TO-COLUMN CONNECTIONS,

Fritz Engineering Laboratory Report No. 333.10, LehighUniversity, Bethlehem, Pennsylvania, September 1970.

6. Graham, J. D., Sherbourne, A. N., Khabbas, R. N., andJensen, C. D.

WELDED INTERIOR BEAM-TO-COLUMN CONNECTIONS, AISCPublication, Fritz Engineering Laboratory ReportNo. 233.15, 1959. Also, Bulletin No. 63, WELDINGRESEARCH COUNCIL, New York, August 1960.

7. Timoshenko, S. P. and Gere, J. M.THEORY OF ELASTIC STABILITY, 2nd edition, Mcgraw-Hill,

New York, 1961.

-16-

p

cr

11. NOT A T ION

area of one flange (of the beam framing in);

ratio of the beam yield stress to the column yield stress;

column web depth between column k-lines or between toesof fillets;

depth of beam;

distance from outer face of flange to web toe of fillet,Fig. 2i

concentrated load;

thickness of the beam flange;

column web thickness;

normal stresSi

yield stress in ksi;

vertical displacement, Fig. 3.

-17-

IJ-ICDI

Tab'Ie 1

SECTION PROPERTIES AND PREDICTED CRITICAL LOADS

, . Inter-Measured Dimensionst AISC Buckling action

cr cr P p. P Testsy y d cr cr cr PultTest (Nom.) Measured c t k Eqs.I,2 Eqs.S,2 Eq. 7No. Section ksi ksi in. in. in. kips kips kips kips

W-3* WID x 39 100 121.9 8.1S 0.344 0.91 ,0 20S 246 253

W-4* W12 x 4S 100 118.2 9.87 0.344 1.11 0 169 204 260,W-S* W12 x 31 36 39.8 10.59 0.270 0.70 0 46 40 61

W-6* WID x 29 36 41.6 8.91 0.308 0.73 0 81 73 90

W-7* WI0 'x 54 50 57.8 8.0S 0.380 1.02 106 --- 'ISS 215

W-8* W 8 x 67 36 30.9 6.60 0.S75 1.22 137 --- 180 250

W-9* W12 x 120 100 97.7 9.95 0.700 1.57 585 --- 978 980

W-I0** HID x 62 36 33.7 7.82 0.504 1.33 130 --- 168 237

W-12 W12 x 45 50 54.0 10.02 0.385 1.00 0 155 151 166

W-1S W12 x 36 100 110.6 10.74 0.324 0.82 0 130 123 235

W-17 WIO x 29 36 42.2 8.91 0.310 0.73 0 82 74 95

W-20 W12 x 27 36 40.7 10.62 0.269 0.69 0 45 39 '64

W-21 W12 x 45 50 56.8 10.02- 0.385 1.00 0 165 159 168

*Reported in Ref. 5**Welded Sectiontdc ' t, and d (depth) were actually measured and k was computed from the expressions (d - d c )/2

(\

t

IL

""'-CompressionRegion

. Fig. 1 Schematic of Typical InteriorBeam-To-Co1umn Moment Connection

pprtbl/ 2.5:1

(Jytttttttt--i to- tb + 5k

Uyt t t • t • t t..J1IIII'-----......

Fig. 2 Simulation of the Compression Region

-19-

+ e- Eccentricity

Fig. 4 Schematic-·of'Eccentricity Problem

SOOk Me'chine

Fig. 3 Test Set-up

-2-0-

120

a­KSI

80

Simple ,Support

p 33,400(j=-=-~-

det (de/t)240

• 100 ksi

o 50 ksi

D 36 ksi

Clamped Support

u= 66,800(de/t )2

o 20

delt

40

Fig. 5. Comparison of Theory Developed in Ref. 5. Using Tests of Specimens with daft RatioGreater than or Close to l80/;cr;

-21-

p=

de t :p;;

15•

4•

crer _ 4100Iffy - (de/t)2

4 []l20

• 100 ksi 5

0 50 ksiD 36 ksi

2

6

8

12

10

a 20 40

delt

Fig. 6 Comparison of Stability Formula with Tests ofSpecimens with d It Ratio Greater than or Closeto 180/~ c

y

-22-

--------------------_.._---~ ..-

177 0

2.0 00 6

8 100 0

0 9 20

0 • 0($)12

05 15

0 DO 21 CJ •1.5 D .3

4cr D

0 •Uy

DO

1.0AISC Eq. 1.15-1

P(j =(tb +5k) t

• 100 ksi(\J

50 ksiSafe I 0l()

0.5 Region - 36 ksicT

0

I.LJ

(de) = 180 uC/)

t a lOY «

0 0.5 1.0 1.5 2.0

(dc/f)

(de/t )0

Fig. 7 comparison of Test Resultswith AISC Formulas

-23-

2.5

1.75 2.01.51.00.5o

2.0 (j P I(Jy = = (de/t)2dct (jy.

[]8 (del t)o1.75 0

1.5 0 90 7[] • 0

(j

(jy0

D

1.00

o JI7 D

Eq. 6/ 12 ~I .34• 15•

0.5 • 100 ksi

0 50 ksi

[] 36 ksi

Fig. 8 Comparison of Test Results with InteractionFormula. Ref. 6 Points not Numbered(dc/t)a = 180/rcr;

-24-

TESTPult

KIPS

DAise ,Eqs. I,2

• Stability , Eq~. 2 ,5

o Interaction, Eq. 7

100 200

PREDICTION Per' KIPS

20~

7--0 0

Test No. ....4...........--_08 --0-----0 3 --.~--

15 --l.2.:::::o:i 0

21----....... )41.,

o

100

200

Fig. 9 composite Comparison of maximum Test Load, with MaximumPr'edicted Load ,as. Determined by th-e Various Formulas

-25-

PKIPS

240

200

160

120·

80

40

Test 2

Test I

Test 2--- Control Test

RControl Test Test I

WI2 x36

50 11

100 ksi

o

~, INCH

Fig. 10 Eccentric Loading Results

-26-

,..- - ..---------- --------------.------------------",,--------------c--c-----c--c----_

PKIPS

0.60.50.40.3

~,INGH

With Cover Plate

0.20.1o

80 Control Test(Plain Beam)

60p P

III =2000/0

WID x2940

P PActual Control

20 Test Test

100

Fig. 11 Contribution of Increased Flange Thickness, t

-27-

stotted Flange Test20

80 P Ptf, =0.565

11

WI2x45 J:2.00 11

60

PControl

40 Test

120

140

100

P Control TestKIPS

o 0.2 0.4 0.6

A,INCH

0.8 1.0

Fig. 12

-28-

a) Side View Slotted Flange Test

b) Top View, Width = d'

Fig. 13 Slotted Flange Specimen After Test

-29-