You are on page 1of 8

EXPERIMENTAL STUDY ON BOLTED UNSTIFFENED SEAT

ANGLE CONNECTIONS
By W. H. Yang,1 M. D. Bowman,2 Member, ASCE, and
W. F. Chen,3 Honorary Member, ASCE

ABSTRACT: Seated-beam connections are occasionally used to attach a beam member to a column web or
flange, particularly during construction. This paper presents the results of an experimental program that was
designed to examine the flexural behavior of bolted, unstiffened seated angle connections. The sensitivity of
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 04/21/13. Copyright ASCE. For personal use only; all rights reserved.

several critical variables that influence the strength of the connection angle is reviewed. General descriptions of
the test procedures and observations are presented, followed by a detailed discussion of the test variables. The
test results are compared with corresponding theoretical predictions to asses the accuracy of various analytical
models. It was found that seat angle strength is significantly influenced by beam setback, angle thickness, and
the bolts that connect the angle to the flange of the supporting beam. Moreover, the strength of the seat angle
can be accurately predicted using a plastic hinge based analytical model that reflects the failure mode.

INTRODUCTION For many years, unstiffened seated connections have been


Two categories of seated connections are specified in the designed using tabular methods. Tables in the AISC manual
current AISC/LRFD manual on connections (Manual 1994): of steel construction (Steel 1945; Manual 1989, 1994) were
Unstiffened seated connections and stiffened seated connec- generated based on the required bearing length method devel-
tions. An unstiffened seated connection is made with a seat oped in the 1940s. A detailed review of the design require-
angle and a top angle, as illustrated in Fig. 1. The angles may ments for unstiffened seat angles was presented by Garrett and
be bolted or welded to the supported beam as well as to the Brockenbrough (1986). Moreover, limited experimental testing
supporting member. While the entire load is assumed to be by Roeder and Dailey (1989) confirmed that the LRFD pro-
transmitted to the column through the bottom seat angle, the cedures are adequate, but the tests also demonstrated variations
top angle is mandatory for stability considerations. in the strength due to the absence of the top seat angle.
Another application for an unstiffened seat connection is to Development of an analytical model to predict the flexural
provide temporary support for a beam member during con- strength of a bolted unstiffened seated connection is presented
struction. The beam member often has a permanent web con- elsewhere (Yang et al. 1997). Based on this work, it was
nection that is installed as the erection progresses. The erection shown that the strength of an unstiffened seat angle, with no
seats are typically left in place, even though they are no longer bolts used to connect the beam flange and the seat angle, is
required, after the final shear connection has been completed. controlled by the formation of a plastic hinge in the vertical
If a moderate to heavy beam is placed upon a seat angle for leg, and is given as
temporary support during construction, then the flexural resis- 1 4bs ⫺ 2ta
tance of the seat angle may be the limiting strength. (The limit R2 ⫹ R⫺1=0 (1)
(Fy Lta)2 Fy Lt 2a
states for beam web crippling and beam web local yielding
may control for light beam members with thin webs.) The where R = load-carrying capacity of the seat angle; L = seat
construction loads placed upon the seat angle must not exceed angle width; ta = angle thickness; bs = beam setback; and Fy
the strength of the angle prior to completion of the beam web = yield strength of the angle. If bolts are used to attach the
shear connection. Moreover, unforeseen construction loads be- beam flange and the seat angle, then it is assumed that two
come more of an issue when additional members are erected plastic hinges form and the strength of the seat angle is com-
that are connected to and supported by the beam that rests puted from the following equation:
upon the temporary erection seat angle. 16 1 4bs ⫺ 2ta
Another factor that plays a role in the strength of a tem- R4 ⫹ R2 ⫹ R⫺2=0 (2)
porary seat angle is the placement of at least two bolts to (Fy Lta)4 (Fy Lta)2 Fy Lt 2a
secure the beam to the angle during erection. Occupational The experimental program, which is presented in this paper,
Safety and Health Administration regulations (‘‘Structural’’ was conducted to study the failure modes of various seat angle
1994) state in Section 1926.751(a) that, ‘‘. . . the load shall not connections and to provide a means to verify the analytical
be released from the hoisting line until the members are se- model. All of the experimental testing was conducted in the
cured with not less than two bolts, or the equivalent, at each Kettelhut Structural Engineering Laboratory in the School of
connection and drawn up wrench tight.’’ The regulations, Civil Engineering at Purdue University, West Lafayette, Ind.
however, do not indicate the size of the bolts to be used, nor
do they indicate how the bolts are to be tightened.
1
Formerly, Res. Asst., School of Civ. Engrg., Purdue Univ., West La-
fayette, IN 47907.
2
Prof. of Civ. Engrg., School of Civ. Engrg., Purdue Univ., West La-
fayette, IN.
3
George E. Goodwin Distinguished Prof. of Civ. Engrg., School of
Civ. Engrg., Purdue Univ., West Lafayette, IN.
Note. Associate Editor: Amde M. Amde. Discussion open until April
1, 2000. To extend the closing date one month, a written request must
be filed with the ASCE Manager of Journals. The manuscript for this
paper was submitted for review and possible publication on June 29,
1998. This paper is part of the Journal of Structural Engineering, Vol.
125, No. 11, November, 1999. 䉷ASCE, ISSN 0733-9445/99/0011-1224–
1231/$8.00 ⫹ $.50 per page. Paper No. 18659. FIG. 1. Bolted Unstiffened Seat Angle Connection

1224 / JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1999

J. Struct. Eng. 1999.125:1224-1231.


EXPERIMENTAL PROGRAM
A total of 77 seat angle specimens for 42 different cases
were tested. The tests focus on those cases where the strength
of the seat angle controls the behavior of the connections.
Other cases that may control connection strength, such as the
limit states of beam web crippling, beam web local yielding,
and welded seat angles, are not addressed in this study. The
primary test variables included in the study were angle thick-
ness, angle length, beam setback, and beam-to-angle attaching
bolts.
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 04/21/13. Copyright ASCE. For personal use only; all rights reserved.

Test Setup
The test setup consisted of a test beam with end supports
and lateral braces, a braced test column, a loading frame, and
FIG. 2. Instrumentation and Dimensions
a longitudinal bracing system. A 2,743 mm (9 ft) long W460
⫻ 193 (W18 ⫻ 130) section of ASTM A36 steel was used for
the test beam. It was supported by a seat angle specimen at ducers were recorded with an OPUS 2.0 computerized data
the test end and a concrete block at the opposite end. A steel acquisition system. A detailed discussion of the test setup and
roller was used to simulate a simply supported end condition the instrumentation is provided by Yang (1997).
at the far end. The test beam was braced at the loading point The bearing stress distribution (or the reaction point) on the
and at the test end to provide lateral stability. The lateral braces seat angle is a key issue in computing the load-carrying ca-
were designed so that they would not carry any vertical load. pacity of seated beam connections. However, it is difficult to
The brace at the test end consisted of a pair of angles con- measure the bearing stress distribution directly. To overcome
nected to the column, but not attached to the beam web. This this difficulty, an indirect approach with two load cells was
prevented lateral movement of the beam without inhibiting utilized. One load cell was placed on top of the test beam at
vertical displacement. It should be noted that the test setup the loading point, while a second load cell was placed under-
utilized a seat angle only, and did not include a top connection neath the test beam at the far end. Using this arrangement, the
angle on the beam flange or web. The primary purpose for magnitude and location of the end reaction on the seat angle
this omission was so that the load and deformation response can be calculated using the following two equations of equi-
of the seat angle could be studied. The limited tests by Roeder librium (Fig. 2):
and Dailey (1989) clearly demonstrate that the top angle does Rangle = RLC #1 ⫺ RLC #2 (3a)
carry some load and contributes to the overall connection re-
sponse. Moreover, for applications when only a seat angle is RLC #1
utilized, such as during construction, it is important to under- x 0 = L ⫺ (L ⫺ a) (3b)
RLC #1 ⫺ RLC #2
stand the behavior of the seat angle alone.
A 1,524 mm (5 ft) long W360 ⫻ 463 (W14 ⫻ 311) section where Rangle = magnitude of end reaction on the seat angle;
of ASTM A36 steel was used for the test column. The vertical RLC#1, RLC#2 = load cell output at the loading point and far end,
leg of the test angles was attached to the web of the column respectively; and x 0 , L, and a = dimensions noted in Fig. 2.
with two 25 mm (1 in.) diameter A325 bolts. The bolts were LVDTs #1, #2, and #3 monitored the vertical deflection of
inserted into 27 mm (1 1/16 in.) diameter holes and tightened the test beam at the far end, loading point, and test end, re-
by the turn-of-nut procedure using an impact wrench. The sur- spectively. Because LVDT #3 was very close to the seat angle,
face of the test angles were coated with a whitewash to ob- and because the deformation of the test beam was negligible,
serve the onset of yielding. Flaking of the whitewash can be the reading of LVDT #3 represented the vertical movement of
used to infer locations where large strains occur. the seat angle that was in contact with the bottom flange of
The loading frame consisted of an W-section cross beam the test beam.
and two laced columns that were made with channel sections.
The frame was braced to prevent out-of-plane displacement Variables and Material Properties
using angles. A vertical force was applied to the top of the test
beam via a 535-kN (120-kilo pound) capacity hydraulic jack The major test variables in the experimental study were
attached to the cross beam of the load frame. beam attack, angle thickness, angle length, beam-to-angle at-
A longitudinal bracing system was originally designed to tachment bolts, and level of bolt tightness. To examine these
provide horizontal constraint at the far end of the test beam. variables a number of different test cases were evaluated with
However, pilot tests indicated that a 25 mm (1 in.) horizontal various combinations of the test variables. A total of 77 seat
movement of the test beam in a longitudinal direction could angle specimens with 42 different cases were tested.
be developed for the unrestrained beam. Hence, significant Beam setback is used herein to indicate the distance from
compressive force could be engendered in the test beam due the face of the column web to the end of the test beam at the
to restraint at the far end. Therefore, for most of the tests, the beginning of the test. The following six different beam setback
restraint at the far end of the test beam was released, and the values were used: 12.7, 19.0, 25.4, 31.8, 38.1, and 44.5 mm
longitudinal bracing system was only used to hold instrumen- (1/2, 3/4, 1, 1 1/4, 1 1/2, and 1 3/4 in.).
tation and to prevent excessive horizontal displacement when All of the test angles had equal horizontal and vertical leg
collapse occurred. sizes. Five different angle thicknesses and two different angle
sizes were examined: 9.5 and 12.7 mm (3/8 and 1/2 in.) thick-
Instrumentation nesses were used for angles with 127 mm (5 in.) long legs;
whereas 15.9, 19.0, and 25.4 mm (5/8, 3/4, and 1 in.) thick-
The instrumentation used in the experimental program in- nesses were used with 152-mm (6-in.) leg sizes. Also, three
cluded two load cells and four LVDTs. The location for the different angle widths were used: 152, 203, and 267 mm (6,
instrumentation is shown in Fig. 2. Test data from these trans- 8, and 10 1/2 in.). Tension tests were conducted per ASTM A
JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1999 / 1225

J. Struct. Eng. 1999.125:1224-1231.


370-94 [4] (‘‘Standard’’ 1994) to determine the material prop- the fillet in the center of the vertical leg of the angle. Yielding
erties of the various angle sections. The yield strength for the then spread along the entire width of the seat angle. From the
various angle sections ranged from 293 to 352 MPa; the tensile load-deflection curves in Fig. 3 (Case 12: BTL-T4-S3-1), it
strength range was 428 to 539 MPa. can be seen that at the initial stage of loading the displace-
The beam flange was attached to the horizontal leg of the ments increase almost linearly; the horizontal displacement is
seat angle with two high-strength bolts. Three different tight- very small as the initial load is applied. When about 15% of
ness levels were considered: Fully tightened by turn-of-nut the peak load is reached, there is an increase in the slope of
procedure, snug-tightened with a small crescent wrench, or the load-vertical deflection curve. At this point (‘‘a’’ in Fig.
finger-tightened. Most of the specimens were fully tightened 3) the section at the first hinge location (vertical leg of the
(35 tests), with only two tests conducted for each of the other angle) has not yielded and the fibers at the second hinge lo-
two conditions. cation (horizontal leg of the angle) are not significantly
An additional 38 tests were conducted to examine the stressed. When about 25% of the peak load is reached, yield-
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 04/21/13. Copyright ASCE. For personal use only; all rights reserved.

strength of the seat angle with no bolts used to attach the seat ing starts at the first hinge location (‘‘b’’ in Fig. 3). As a result,
angle to the beam flange. Although this condition is not rec- a noticeable decrease in stiffness of the connection is observed.
ommended for conventional practice, it was studied, neverthe- The degradation of stiffness continues until the peak load point
less, to provide a comparison with the strength of seat angles is reached and a failure mechanism is formed. After the peak
that are attached to the beam flange. For this condition, the load point, the connection can deform continuously without
mechanism of friction and gravity alone was used to attach significant loss of strength until the end of the test. The bolt-
the beam flange to the seat angle. tightened test was always terminated due to excessive defor-
mation, and no collapse of the test beam occurred during load-
TEST RESULTS AND OBSERVATIONS
Test Procedures
The loading rate of the hydraulic jack was adjusted manu-
ally using a pressure regulator to control the output pressure
of the pump. The load was applied continuously from the in-
itial loading state until failure. The data acquisition system was
programmed to record test data at 1.5-s intervals. The initial
elastic loading rate was about 4.5–9.0 kN/s (1–2 kip/s). When
the load-deflection curve began to deviate from linear behav-
ior, the loading rate was gradually reduced to displacement
control to collect additional data. The test was terminated
when one of the following four conditions occurred:

1. Peak load point passed and the load started to drop while
the vertical displacement, as inferred from LVDT #3, still
increased.
2. Excessive vertical or horizontal displacement of the test
beam observed, i.e., the readings of LVDTs #3 or #4
surpassed 25 mm (1.0 in.).
3. Collapse of the test beam.
4. The reading of Load Cell #1 reached the maximum ca-
pacity of the hydraulic jack.

General Observations
Based on their general behavioral trend, the tests fall into
two broad categories: (1) Bolt-tightened (39 tests) and (2) no-
bolt cases (38 tests). Bolt-tightened indicates that two bolts
were used with nuts tightened to attach the horizontal leg of FIG. 3. Typical Load-Deflection Curves: (a) Load versus Verti-
the seat angle to the beam flange, whereas the no-bolt case cal Displacement; (b) Load versus Horizontal Displacement
indicates that no bolts were used to secure the angle to the
beam. Twenty-four of the 39 bolt-tightened tests produced an
angle failure strength that ranged from 197 to 498 kN (44.3
to 112.0 kips); 15 specimens did not fail when the capacity of
the loading actuator (516 kN) was reached. Thirty-four of the
no-bolt tests produced an angle failure strength range of 72–
443 kN (16.2–99.6 kips); four specimens did not fail when
the capacity of the loading actuator was reached.
The bolt-tightened specimens were typically stronger and
deflected more than no-bolt specimens with comparable thick-
nesses and angle lengths. Fig. 3 illustrates typical load-deflec-
tion curves of seated connections from these two categories.
Both specimens consist of 127 ⫻ 127 ⫻ 12.7 angles (5 ⫻ 5
⫻ 1/2) that are 267 mm (10.5 in.) wide with a 19-mm (3/4-
in.) beam setback.
For a bolt-tightened seated connection, yielding—as in-
ferred by whitewash flaking—was observed to initiate near FIG. 4. Deformed Shape of Bolt-Tightened Specimens

1226 / JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1999

J. Struct. Eng. 1999.125:1224-1231.


strength can be correlated approximately with the angle width.
Consequently, the test results for the 12.7 mm (1/2 in.) thick
angles are scaled down by a factor of 8/10.5 to compare them
graphically with the results of the other angle specimens. It
can be seem that the load-carrying capacity is very sensitive
to the value of the beam setback for both no-bolt and bolted
cases. For example, if a seated connection with a 19-mm
(0.75-in.) beam setback is used as a criterion for comparison,
then a 6.4-mm (1/4-in.) increase in the beam setback would
cause the load-carrying capacity to decrease by an average of
21.4%; a 6.4-mm (1/4-in.) decrease in beam setback would
cause the load-carrying capacity to increase by an average of
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 04/21/13. Copyright ASCE. For personal use only; all rights reserved.

42.1%.
Differences between bolted and no-bolt specimens are also
shown in Fig. 6. Significant differences in the strength for the
FIG. 5. Deformed Shape of No-Bolt Specimens
two cases are evident. However, a similar trend on the effect
of beam setback is apparent for both cases: An increase in
ing. At the end of the test, the yield region appears to have beam setback results in a decrease in strength. Only limited
extended completely through the compression zone of the ver- test results are shown in the upper right corner of the figure
tical leg, and yield lines had formed in both legs of the seat for the thicker bolt-tightened specimens and the 25-mm (1-in.)
angle. A view of the typical deformed shape of several bolt- no-bolt specimen.
tightened specimens is illustrated in Fig. 4. The AISC Connections Manual (Manual 1994) indicates
For a no-bolt seated connection, yielding initiates near the that 13 mm (1/2 in.) is a typical nominal setback. The setback
fillet in the center of the vertical leg, and then propagates along position could change somewhat during construction for a va-
the length of the angle. From the no-bolt load-deflection riety of reasons. One possibility is that some shifting could
curves shown in Fig. 3 (Case 16: NBL-T4-S3-1), it can be occur as the column is adjusted for final plumbness. The sen-
seen that no obvious linear portions are observed at the initial sitivity of the setback variable, as illustrated in Fig. 6, indicates
stages of loading. Vertical and horizontal displacements both that efforts should be taken during construction to ensure that
increase nonlinearly soon after the load is applied, and both excessive changes in the nominal setback do not occur.
displacements are about the same magnitude throughout the
loading process. The test often ends with the sudden collapse
of the test beam. The postpeak unloading portion of the load- Angle Thickness
deflection curves were only occasionally recorded due to the
Fig. 7 illustrates the relationship between average strength
rapid nature of the no-bolt failure modes. Only one yield line
and angle thickness. [Again, the results of 12.7 mm (1/2 in.)
forms in the no-bolt test, rather than two yield lines, as noted
thick angle specimens are scaled down by an angle length
for the bolt-tightened case. The yield line for the no-bolt test
factor of 8/10.5 for comparison purposes.] Because the eccen-
forms on the vertical leg of the angle that is bolted to the
tricity of the load is directly related to angle thickness (i.e.,
column web. Also, the width of the yield zone is much nar-
eccentricity = setback ⫺ thickness/2), there is a fundamental
rower than the yield zone developed in the bolt-tightened
difference between connections made with thin and thick an-
cases. The horizontal leg of the angle experiences no plastic
gles. For example, when the beam setback is bs = 19 mm (0.75
deformation. Fig. 5 shows typical deformed shapes of several
in.) and the connection is made with a 9.5 mm (3/8 in.) thick
no-bolt seated angles.
angle, then the load from the supported beam would be trans-
mitted to the angle vertical leg (which is bolted to the column
DISCUSSION OF TEST VARIABLES web), primarily through the bending action of the angle hori-
Beam Setback zontal leg, as illustrated in Fig. 8(a). In contrast, however, if
the connection is made with a 25 mm (1 in.) thick angle, then
Fig. 6 illustrates the relationship between average seat angle most of the load would be transmitted to the vertical leg
strength and beam setback. All of the angle test specimens in through direct bearing action, as shown in Fig. 8(b). In the
Fig. 6 are 203 mm (8 in.) wide, except the 12.7 mm (1/2 in.) latter case, the bending action of the horizontal leg of the angle
thick specimens, which are 267 mm (10.5 in.) wide. The angle would be a secondary effect.

FIG. 6. Effect of Beam Setback on Connection Strength

JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1999 / 1227

J. Struct. Eng. 1999.125:1224-1231.


vailing for no-bolt seated connections than for bolt-tightened
cases.
Second, connection ductility can be greatly improved if the
beam-to-angle attachment bolts are installed and tightened.
The experimental results indicate that most of the fully tight-
ened seated connections can deflect more than 25 mm (1.0 in.)
without a significant loss of strength; whereas connections
without beam-to-angle attaching bolts deflect less than 12 mm
(0.5 in.) before collapse occurs (Fig. 3). Most of the tests for
the bolt-tightened cases were terminated due to excessive de-
formation, whereas the tests for the no-bolt cases often ended
with a sudden collapse of the test beam.
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 04/21/13. Copyright ASCE. For personal use only; all rights reserved.

Bolt Tightness
FIG. 7. Effect of Angle Thickness on Connection Strength
ASTM A325 high-strength bolts were used to connect the
seat angle specimens to the web of the test column and to the
bottom flange of the supported beam. The bolts connecting the
seat angle to the column web were fully tightened using turn-
of-nut tightening per the Research Council on Structural Con-
nections specification (‘‘Specification’’ 1988). The tightness of
the bolts connecting the test beam to the seat angle was a test
parameter in the experiments. The following three different
tightness levels were studied: (1) Turn-of-nut fully tightened;
(2) snug tight; and (3) finger tight.
Fig. 9 compares the load-deflection curves of seated con-
nections tightened using the three different methods. From this
comparison it can be noted that the ultimate load-carrying ca-
pacities of the seated connections are not very sensitive to the
bolt tightness. A snug tight connection has almost the same
load-carrying capacity as a fully tightened one; a finger tight
connection can reach 95% of the strength of a fully tightened
connection. Because further tightening of the bolts from the
snug tight condition to the fully tight condition does not im-
prove the performance of the connection significantly, then it
is appropriate to suggest that the snug tight condition for the
two beam-to-angle attaching bolts should be permissible.
However, it should be noted that if the seat is used as part of
FIG. 8. Illustration of Connections with Thick and Thin An-
gles: (a) Thin Angle; (b) Thick Angle the final connection and the bolt holes are oversized, the bolts
must be fully tightened.
The slenderness of a thin angle vertical leg under compres-
sion is much larger than that for a thick angle. While the load- Beam End Restraint
carrying capacity of seated connections with thick angles is a Two duplicate tests were conducted to study the effect of
largely a problem of material strength, the effect of stability the restraint at the far end of the test beam provided by a
could play a role in the performance of seated connections longitudinal brace. Fig. 10 compares the load-deflection curves
with thin angles, particularly for the no-bolt cases. As a result, of the two duplicate bracing tests. For the bolt-tightened spec-
seated connection strength increases very rapidly with an in- imens shown in Fig. 10(a), the flexible longitudinal brace sig-
crease in the angle thickness. Consequently, the ultimate ca- nificantly increases the connection stiffness, but not the
pacities of some of the seated connection test specimens made strength. The brace also changes the deformed shape of the
with 19 mm (3/4 in.) and 25 mm (1 in.) thick angles could seat angles by causing the plastic hinge location to shift up-
not be reached due to the limited capacity of the loading ward along the vertical leg, as illustrated in Fig. 11. For the
system. no-bolt specimens, however, the longitudinal brace can in-
Beam-to-Angle Attachment Bolts
The effect of the beam-to-angle attachment bolts on the per-
formance of unstiffened seated connections can be observed
in the load versus displacement curves shown in Fig. 3. Two
general observations can be made.
First, the load-carrying capacity can be increased by at least
125% if the beam-to-angle attachment bolts are installed and
tightened. This increase in load-carrying capacity is primarily
accomplished by changing the failure mode from a single-
hinge mechanism (which is statically determinate) for the no-
bolt case to a double-hinge mechanism (which is statically
indeterminate), for the bolt-tightened case. The change in the
failure mechanism is evident when comparing the deformed
shapes of bolt-tightened cases (Fig. 4) with those of no-bolt
cases (Fig. 5). Additionally, the effect of stability is more pre- FIG. 9. Effect of Bolt Tightness on Connection Strength

1228 / JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1999

J. Struct. Eng. 1999.125:1224-1231.


cantly influence the seat angle connection strength. In the fol-
lowing sections, the AISC LRFD design procedures, which are
included in the connections manual (Manual 1994) and an
analytical model proposed by Yang et al. (1997), are compared
with the experimental data.

Assessment of LRFD Procedures


In this section, the strength given by the AISC LRFD pro-
cedures (Manual 1994) are compared with appropriate ex-
perimental results. It should be noted that the testing program
primarily concentrated on the case of a flexible seat angle and
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 04/21/13. Copyright ASCE. For personal use only; all rights reserved.

stiff beam, with no top angle attachment. The tests do not


mobilize other failure modes (web yielding and web crip-
pling), which are evaluated by the AISC procedure. Moreover,
several cases were tested where two bolts were not used to
connect the beam flange to the connection angle. For these
cases, no comparisons are made with the AISC LRFD proce-
dures, because the AISC procedures assume that two bolts are
used to attach the beam flange to the seat angle.
To provide a means of evaluating the AISC LRFD proce-
dure (Manual 1994), the predicted strength was compared with
the experimental results for 11 applicable bolt-tightened spec-
imens. The results of this comparison are shown in Fig. 12 as
the ratio of predicted to observed angle strength. Significant
variations in this ratio are evident: For thin and short seat
angles, the LRFD predictions are high (1.098 maximum); for
thick and wide seat angles, the LRFD prediction is very con-
servative (0.537 minimum). The average ratio for the 11 cases
FIG. 10. Effect of Longitudinal Restraint on Connection was 0.795, and the standard deviation was 0.173.
Strength: (a) Bolt-Tightened Case; (b) No-Bolt Case It is believed that the effect of the beam-to-angle bolts are
implicitly included in the basic assumptions of the LRFD pro-
cedure. However, experimental observations indicate that the
failure mechanism is different than that assumed in the LRFD
procedure. While the LRFD procedure assumes that the critical
section is at the toe of the fillet of the horizontal outstanding
leg that is connected to the beam (Garrett and Brockenbrough
1986), the experimental tests demonstrate that the vertical leg
that is bolted to the column is more critical. Yielding begins
in the vertical leg of the seat angle and then extends com-
pletely through the compression zone of the leg at the end of
the test. Although a secondary hinge does occur near the toe
of the fillet on the horizontal leg for bolt-tightened cases, the
rotation direction of the plastic hinge is the opposite of that
assumed in the LRFD procedure.
Additionally, it was observed that bolted unstiffened seated
connections can adequately yield without fracture. The con-
nection is capable of rotation, without distress, far beyond that
required to accommodate the end rotation of the beam to
which it is connected. This observation, which was first re-
ported by Roeder and Dailey (1989), implies that it is neither
FIG. 11. Shifting of Hinge Location on Vertical Leg of Angle:
(a) Braced and Bolted (Case 1: BTS-T3-S3-1); (b) Unbraced and necessary nor appropriate to assume that the limit states of
Bolted (Case 2: BTS-T3-S3-2) excessive bending stress on the seat angle and local web yield-
ing of the supported beam are reached simultaneously as in

crease the strength somewhat [Fig. 10(b)]. The increase in


strength is due primarily to the lateral force developed by the
bracing restraint. This force tends to offset some of the mo-
ment in the vertical leg of the angle caused by the vertical
reaction. Once the angle slips, the horizontal force is released
and there is a sudden reduction in the load-carrying capacity
[Fig. 10(b)].

ASSESSMENT OF ANALYTICAL METHODS


The experimental tests demonstrate that the installation of
beam-to-angle attachment bolts greatly affect the strength of
unstiffened seat angle connections. Also, it was found that
beam setback and angle thickness are variables that signifi- FIG. 12. Variation of Strength Ratio for Bolt-Tightened Case

JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1999 / 1229

J. Struct. Eng. 1999.125:1224-1231.


the current LRFD procedures. Instead, the two limit states ening. Therefore, the average stress level at the hinge locations
should be evaluated individually, and the one that produces a is higher than the yield strength Fy. In contrast, for the no-bolt
lower load-carrying capacity should govern. cases, the strain level of most fibers at the plastic hinge lo-
cation is less than the onset of strain hardening (i.e., at the
Assessment of Rational Flexural Model yield plateau) at the time when the collapse of the test beam
occurs. Therefore, some degree of strain hardening contributes
The analytical procedures by Yang et al. (1997) are based
to the ultimate strength of the bolt-tightened connections, but
on a rational flexural strength model that accounts for the for-
not to that of the no-bolt connections. Because only the yield
mation of plastic hinges in the unstiffened seat angle [(1) and
strength Fy has been used in the prediction of the rational
(2)]. To evaluate the analytical model, the predictions based
model, it is logical that the predictions for most of the no-bolt
on the rational model were compared with the experimental
cases are rather accurate, whereas those for the bolt-tightened
results for the fully bolt-tightened specimens (Fig. 12). For the
cases are conservative.
11 specimens, the average strength ratio was 0.761 with a
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 04/21/13. Copyright ASCE. For personal use only; all rights reserved.

standard deviation of only 0.078, less than one-half of the


standard deviation of the LRFD procedure. It was found that SUMMARY AND CONCLUSIONS
the LRFD-predicted strength is less conservative as the angle The tests and analytical predictions in this paper refer to
leg cross-sectional area (L ⫻ ta) decreases. This negative ten- situations where the seat angle is attached to the supporting
dency was caused by neglecting the reduction effects of axial column with bolts. The failure mechanisms and maximum
and shear forces on the plastic moment capacity of section, loads would be notably different if the seat angles were welded
which was discussed in detail by Yang et al. (1997). all around to the supporting columns.
For no-bolt cases, the predictions by the rational model Based on the experimental results reported herein and com-
compare generally well with the experimental results. For the parisons between experimental and analytical models, the fol-
17 applicable cases, the average strength ratio was 1.013 with lowing conclusions can be stated when seat angle strength con-
a standard deviation of 0.164. However, there are some be- trols the connection behavior.
havioral shortcomings in the rational predictions for the no-
bolt cases. For example, the predicted to experimental strength 1. The strength of bolted and no-bolted seated connections
ratio tends to increase for cases where there is a small beam decrease significantly as the beam setback increases.
setback or a thin seat angle. The reason for this discrepancy 2. Seated connection strength increases very rapidly as the
is that the rational models are based on a first-order plastic angle thickness increases.
hinge analysis, which does not take into account the second- 3. The load capacity of seated connections was increased
order effect of stability (i.e., P ⫺ ⌬ effect). As mentioned at least 125% when beam-to-angle attachment bolts were
previously, the effect of stability is more prevailing in the no- installed and tightened. The increase is primarily accom-
bolt cases than it is for the bolt-tightened cases. The smaller plished by changing the connection failure mode from a
the beam setback, the smaller the eccentricity; the thinner the single-hinge mechanism for the no-bolt case to a double-
angle, the larger the slenderness. Therefore, the first-order ra- hinge mechanism for the bolt-tightened case.
tional model tends to overestimate the load-carrying capacity 4. The ultimate load-carrying of bolt-tightened seated con-
of no-bolt connections with thin seat angles and small set- nections is not very sensitive to the bolt tightness.
backs. 5. Limited testing of the restrained connection demon-
Finally, the difference in the average value of the predicted strated that preventing longitudinal movement of the
to experimental strength ratio for the two general cases should beam increased the strength of no-bolt connections
be mentioned: (1) 1.013 for the no-bolt case, and (2) 0.761 somewhat, but the restraint did not significantly influence
for the bolt-tightened case. A comparison between the pre- the strength of bolt-tightened connections.
dicted strength and experimental results is illustrated in Fig. 6. The AISC LRFD procedures for predicting the strength
13 for various beam setback locations for 13 mm (1/2 in.) of unstiffened seated connections do not directly include
thick by 267 mm (10.5 in.) wide seat angle members. This the effect to the beam-to-angle attachment bolts. How-
25% discrepancy for the two general cases is due to the cor- ever, the LRFD procedures are conservative for most
relation between strength and ductility, as noted previously cases, although the strength distribution is not uniform
when the effect of beam-to-angle attachment bolts was ana- for seat angles of different sizes.
lyzed. For the bolted cases, abundant ductility is available to 7. The experimental results confirm the correctness of the
ensure that the strain of most fibers at the two plastic hinge failure modes utilized in the plastic-hinge based analyt-
locations can reach a level far beyond the onset of strain hard- ical models presented by Yang et al. (1997) for both
bolted and no-bolt seated connections. The rational
model demonstrated a more uniform strength distribution
than the LRFD procedures.

ACKNOWLEDGMENTS
The research reported herein is based in part on work sponsored by
the Department of Labor, Occupational Safety and Health Administration,
which is greatly acknowledged.

APPENDIX. REFERENCES
Garrett, J. H., and Brockenbrough, R. L. (1986). ‘‘Design load for seated-
beam in LRFD.’’ Engrg. J., 23(2), 84–88.
Manual of steel construction, allowable stress design. (1989). 9th Ed.,
American Institute of Steel Construction, Chicago.
Manual of steel construction, load and resistance factor design. (1994).
2nd Ed., Vol. II, American Institute of Steel Construction, Chicago.
FIG. 13. Effect of Beam Setback for Bolted and Unbolted Seat Roeder, C. W., and Dailey, R. H. (1989). ‘‘The results of experiments on
Angles seated beam connections.’’ Engrg. J., 26(3), 90–95.

1230 / JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1999

J. Struct. Eng. 1999.125:1224-1231.


‘‘Specification for structural joints using ASTM A325 or A490 Bolts— ‘‘Structural steel assembly.’’ (1994). Section 1926.751, 29 CFR, Chapter
Load and Resistance Factor Design.’’ (1988). Research Council on XVII, Occupational Safety and Health Administration, Washington,
Structural Connections, Engineering Foundation. D.C.
‘‘Standard test methods and definitions for mechanical testing of steel Yang, W. H. (1997). ‘‘The behavior and design of unstiffened seated-beam
products.’’ (1994). A 370-94, ASTM, West Conshohocken, Pa. connections,’’ PhD thesis, Purdue University, West Lafayette, Ind.
Steel construction, a manual for architects, engineers and fabricators of Yang, W. H., Chen, W. F., and Bowman, M. D. (1997). ‘‘The behavior
buildings and other steel structures. (1945). 4th Ed., American Institute and load-carrying capacity of seated-beam connections.’’ Engrg. J.,
of Steel Construction, New York. 34(3), 89–103.
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 04/21/13. Copyright ASCE. For personal use only; all rights reserved.

JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1999 / 1231

J. Struct. Eng. 1999.125:1224-1231.

You might also like