SPLICE PLATE DESIGN RECOMMENDATIONS FOR SPLICE
GAP CONNECTIONS OF BUCKLING RESTRAINED
BRACED FRAMES:
by
‘Tyler M. Josephson
{A projec submitted tothe faculty of
Brigham Young University
in partial fulfillment ofthe requirements forthe degree of
Master of Science
Department of Civil and Environmental Engineering
Brigham Young University
August 2008BRIGHAM YOUNG UNIVERSITY
GRADUATE COMMITTEE APPROVAL,
of a project submitted by
‘Tyler M. Josephson
‘This project has been read by each member ofthe following graduate committee and by
‘majority vote has been found to be satisfactory.
gfe for
Date
_ G8 [08
s/sfos
Dae
i
Dene
a
Accepted forthe Department E Taman
Graduate CoordinatorABSTRACT.
‘SPLICE PLATE DESIGN RECOMMENDATIONS FOR SPLICE,
GAP CONNECTIONS OF BUCKLING RESTRAINED
BRACED FRAMES
‘Tyler M. Josephson
Department of Civil and Environmental Engineering
Master of Seience
Buckling restrained braced frames (BRBs) are an efficient lateral force resisting
system, Braces in these frames ae equal in strength in both compression and tension due
{0 the confined steel core and the debonding agent. Experimental esting has shown that
the beam column connections of BRBs experience significant moment, The beams and
columns may experience local buckling while the gusset pate may experience weld
fracture,
[A new type of connection has been experimentally tested at Lehigh University
and Brigham Young University. ‘The connection consists of splice gap in the beam. Thesplice acts as a hinge which decreases bending moments developed atthe beam-to-
column connection. Results from these experiments showed a significant deerese inthe
connection moments and that yielding was confined primarily tothe gap plates.
‘This project accomplishes two things. The firsts that it uses computer modeling
validation techniques developed at Brigham Young University to determine splice plate
strains, Three designs were developed to capture the extremes of plate thicknesses,
‘The second purpose isto validate an analytical model that uses an equation to
‘estimate the strain the plates will experience,
Results indicate thatthe thinner the plate, the lower the equivalent plastic strain in
the plate. The upper left and lower right plates experience strains 5 to 10 times higher
than the lower lef and upper right plates. The equation in the analytical model appears to
okt true, The obtained results from both the computer validation and the analytical
‘model may be an indication a to how the plates wll performs but further research needs
to be done to more accurately model the plates.ACKNOWLEDGEMENTS.
‘There are few things that made this study possible, Firstly, my committee chair,
Dr, Paul W, Richard's, patience and understanding. Secondly, Iam grateful fora wife
{at has listened to so many things about this research that must surely have bored he to
death, Finally I would like to thank the wonderful UNIX machin in 0m 317A ofthe
CCyde Building that ran all my computer models. There were many times we didn’t get
along and I thought I might have thrown it out the second story window, but in the end
we did it.TABLE OF CONTENTS
LIST OF FIGURES.
1 INTRODUCTION.
1.1 BUCKLING RESTRAINED BRACED FRAMES
12 EXPERIMENTAL TESTS USING BRB,
13. EXPERIMENTAL TESTING OF SPLICE GAP CONNECTION
14 OBJECTIVE,
2. ANALYTICAL MODEL
2.1 BACKGROUND.
2.2 METHODOLOGY FOR CALCULATION.
COMPUTER SIMULATIONS.
3.1 TEST METHOD AND LOADING PROTOCOL.
32. BAY AND BRB DESIGN.
33. PLATE DESIGN
34 MODELING TECHNIQUES.
3.5 EQUIVALENT PLASTIC STRAIN (PEEQ).
13
436 STRAIN DUE TO MOMENT.
4 RESULTS AND DISCUSSION
4.1 PEEQRESUL
Ss
4.1 UPPER LEFT
4.2 LOWER LEFT
4.13 UPPER RIGHT...
4.14 LOWER RIGHT...
4.15 GENERAL OBSERVATION
42. ANALYTICAL MODEL.
42.1 RESULTS AND DISCUSSUION.
5 CONCLUSIONS AND RECOMMENDATIONS.
5.1 CONCLUSIONS
52. DESIGN RECOMMENDATIONS.
53 RECOMMENDATIONS FOR FUTURE STUDY.
REFERENCES
APPENDIX A. CALCULATIONS FOR PLATE DESIGNS.
A.1 CALCULATIONS FOR PLATE DESIGNS.
APPENDIX B. EXPLANATION OF SPREADSHEET.
B.1 CALCULATIONS FOR PLATE DESIGNS.
“
2
2
22
23
BB
24
4
24
22
32
32
32
34
35
35
39
39LIST OF FIGURES
Figure 1-1 BRB hysteretic behavior (Prinz 2007).
Figure 1-2 BRB detail (Coy 2007). 6
Figure 1-3 Lehigh Flange Splice connection (Coy et al, 2007)
Figure I-4 Common BRB connection failures (Prinz 2007),
Figure 1-5 Lehigh Experimental Setup (Fahnestock etal. 2007).
Figure 1-6 Flange splice detail (Coy et al, 2007). : . 9
Figure 2-1. Strains develop from both axial loads and moments n
Figure 2-2. Varies used in equation 2-1. 1
Figure 3-3-1 Boundary conditions and terminology for test frame
Figure 3.2 Left connection, cross setion, and terminology 0
Figure 3-3 Right conection, cross section, and terminology 18
Figure 3-4 Design 1 detail 19
Figure 3-5 Design 2 detail 19
Figure 3.6 Design 3 detail 20
Figure 3-7 Modeling techniques for splice plates (Prinz 2007) 20
Figure 38 Beam to shell modeling techniques at
Figure 4-1 Results for upper lt plate 26
Figure 42 Resuks for lower lt plate 0Figure 4-3 Results for upper right plate.
Figure 4-4 Results for lower right plate.
Figure 45 Analytical results forthe positive pushover
Figure 46 Analytical results for the negative pushover
Figure A-1 Caleulations for plate design 1...
Figure A-2 Calculations for plate design 2.
Figure A-3 Calculations for plate design 3...
31
36
37
38
Figure B-1 (top) the geometric user inputs, (bottom) the methodology for calculation 391 INTRODUCTION
1.1 BUCKLING RESTRAINED BRACED FRAMES
(One ofthe great challenges in structural engineering is designing an economicel
structure tobe stiff enough to handle the lateral loading caused by a seismic event.
Earthquakes are unpredictable in motion and magnitude. A building must be stiff enough
in all directions tobe able to handle the demands imposed by an earthquake, Sufficient
stiffness canbe achieved by bracing the building. Any numberof braces canbe placed in
te building to schiove the desired lvel of siffiosss. Braces are wally paved in
‘orthogonal directions to each other thus abe to ress the demand in any direction,
Currently there ae many types of braces. A popular choice today is buckling restrained
braces (BRBS),
Buckling Restrained Braces (BRBs) were firs used in the United Stats in carly
2000 and were implemented asthe main lateral force resisting system due to several
advantages BRBs have over eccentric braced frames and concentric braced frames (Ko
and Field, 2001). While the latter types of braces perform well n tension, they buckle
under compression loading leaving the frame to resist the axial forees. BRBS are
designed to restrain the axial ste! from buckling under compression, Because the
compression loaded element can't buckle, « BRB bas equal capacity in both tension and
compression. Figure I-A shows an idealized hysteretic behavior of a BRB.
1‘There are three key elements ina BRB. ‘The first is an inner stel core tha will
lake the axial load, Sutrounding the stel core is mortar which is encased in a square
steel tube. A de-bonding agent is applied between the stel core and mortar. The
combination of the mortar and confining stel tube prevent buckling ofthe inner stel
core, (See Figure 1-B),
1.2. EXPERIMENTAL TESTS USING BRBs
A special opportunity was afforded tothe University of California at Betkeleyin
constructing @ new building that would ned to resist seismic loads fom nearby seismic
fault (Lopez 2007). Buckling Restrained Braces were chosen asthe lateral force resisting
system, Howover, befor the Seismie Provisions (2006) ofthe AISC code came out,
there were no official provisions on how to design and implement these types of braces.
Recommended Provisions (Sabelli 2008) required designs tobe in aocordance with two
types of qualiying tests. The frst type included at lest one axial brace test an the
second type included using the brace ina sub-assembly, Forthe sub-assembly test
engineers were able to perform experiments on eam that would be nesy identical to
the one used inthe field, Tree tests were performed using the same frame setup. Each
test increased the percent interstory dit the las fest was loaded tothe whimate strength
of the frame.
Results showed that while the brace capacities would be adequate in meeting the
‘demands of the loading, significant locelized connection failures occurred, Failures
included buckling of the fre end of the gusset plate (which led to crack initiation of.
_gusset-to-column welds) an flexural yielding ofthe beam-to-column connection, Theconclusion was thatthe connection between all the elements (the gusset plate, the beam,
‘and column) was too rigid. The brace was unable to reach its ultimate axial loading,
Therefore the “leftover” loading was undesirably resisted by the frame itself
Similar testing of BRBs was performed by Christopolus atthe University of|
‘Washington (Adams 2005). The main objective of the research was to evaluate the
performance of BRBS in full sale testing. Full sale testing made it possible
determine how well the braces would perform within a frame system. Five full seale tests
were conducted on atypical bay ofa braced stee! building. Each test varied in the design
‘of the gusset plate (square or tapered), the type of bolted connection (sip-critieal or
bearing), and orientation ofthe BRB core section (in-plane with the coluran webs or
perpendicular to the plane ofthe column webs). All braces had a nominal yield strength
0 220 kips with fat core plate cross scetions,
‘The results were similar to those atthe U.C, Berkley. Tensile fracture ofthe BRB
core was never reached in any of the specimens. Instead, severe buckling and yielding
‘ceurred inthe beams and columas in regions closest to the gusset plate, ‘These failure
modes led to out-of-plane rotations of the beams which n tr caused the brace to buckle
and rotate out of plane,
1.3, EXPERIMENTAL TESTING OF SPLICE GAP CONNECTION
[A proposed solution to solve the rigidity issues ofthe BRB connections was to
create a “plastic” hinge inthe beam by splicing it near the gusset plate. The hinge would
limit the moment that could develop in tbe bear and column, Free rotation encourages
all the loading to go through the long axis ofthe BRB. Fahnestock etal (2007) at Lehigh
3‘University designed a hinged connection that was used ina scaled laboratory experiment
‘The hinge was composed ofa splice inthe beam just after the beam gusset connection,
Plates were bolted on either side ofthe web, The working point remained atthe
intersecting point ofthe brace line of loading andthe ceaterlines ofthe webs ofthe beam
and column, The BRB was pin connected tothe gusset plate. (See Figure 1-C)
‘The building design was of atypical Ls Angeles office building on sf sol see
Figute I-D). Due to limiting laboratory constraints, the height ofthe building was
limited to four stores and sealed by a factor of 35; any scale factor much less than this
‘would not provide accurate results. Loading, forces, beams and columns were scaled
accordingly
Four seismic loads were used for testing; frequently occurring earthquake (FOE),
‘design basis earthquake (DBE), maximum considered earthquake (MCE), and an
tershock earthquake of 80% ofthe DBE. The 1971 San Fernando Earthquake was
chosen forthe FOE sealed at 1.33, The Nosbridge Earthquake was used for both the
DBE and the MCE scaled at 89 and 1.33, respectively.
[A variation ofthe hinged connection was tested at Brigham Young University
(Coy 2007). In liew of placing the plates across the web, plates were bolted on the
exterior and interior of the top flange (see Figure I-F) Plates were fastened tothe web
only to keep the connection in plane during loading. The placement ofthe lange plates
shifted the working point to act through the middle ofthe top lange. ‘The gusset plate
was designed to maintain a common intersection of lines of action between the column,‘beam, and brace. A bolted connection of BRB and gusset plate was used in place of
pinned connection,
‘The experiment was run similar to a sub-assemblage setup using non-scaled
beams, columns, and braces
Results indicated the hinges were able to act as a moment release while allowing
the BRB to reach closer tots ful axial capacity. Damage was limited to the splice area
‘hile litle to no yielding oecurred inthe gusset plate, the column, or the welds
connecting the tw.
1.4 OBJECTIVE
‘The purpose of this paper isto analyze thre different design variations of the
splice plates using ABAQUS as the finite element software, The results willbe used to
‘compare equivalent plastic strain values and validate an analytical model for computing
plate strains.Force
Tension
Buckling
Restrained
Brace
Displacement
Compression
igure I-A lacaied HR hysteretic behavior (Pyne 2007)
3. Outer Tube 2. Conerete Mortar
4. Steel Core (covered
Jnunbonding material)
Figure 1-8 BRB detail (Coy 2007)
64r
oSS5 | 0508
4 foooptooos|
BS] iger —
4 we
Figure 1-C Lehigh Flange Splice connection (Coy tal. 2007)1
Topof sel elevation: 57-6"
eval
Isp sin S00"
Laval
“Tepot sts elevation 15:0"
teva 1
‘op of sta seaton 620"
Grin at
eee
[Figure 1D Lehigh Experimental Setup (ahnestock eal. 2007)Beam Local |_ Gusset weld
Buckling Fracture
a
Column
Column Local
Bucking
Figure 1-6 Common BRB connection failures (Prinz 207)
Figure -F Flange splice detail (Coy a. 2007)2 ANALYTICAL MODEL
2.1 BACKGROUND
Strains develop inthe splice pats as the bay sways. Under smal drifts the
strains ae due to axial compression or tension. As the story dit increases, flexural
‘moments develop tension inthe extreme outer fiers and compression inthe bottom
fibers (due to the bending ofthe plates) causing additional the axial strain. The spice
plates experience both an axial-based and a moment‘ased strain, (See Figure 2-A,)
2.2. METHODOLOGY FOR CALCULATION
The analytical model assumes thatthe stains remain within elastic range. The
(2)
Where the angle isthe angle of how much the spice plates bend, A spreadsheet
was developed in Microsoft excel o calculate the angle 8. (See 0 for an explanation of
the spreadsheet.) Lis the ength ofthe splice inthe beam. The distance from mid
thickness of the top flange tothe extreme fiber ofthe splice plate is y. Figure 2-B shows
these variables graphically
10Figure 2-AStralas develop from both axial loads an moments
igure 2-8 Variables sed in equation 2-1
n3 COMPUTER SIMULATIONS
3.1 TEST METHOD AND LOADING PROTOCOL.
Analysis was performed on a single one story bay. Figure 3-3-A, Figure 3B, and
Figure 3-C show the testing setup and the terminology used in describing the setup and
results, A pushover analysis was performed inthe postive and negative direction. This
analysis considered displacing the bay in only one direction at atime, either inthe
Positive or negative direction. Each direction consisted of ineten
nts of displacement
‘The bay was restricted against out-of-plane movement. Data was obtained from the
middle ofthe splice pate.
The loading protocol for the pushover analysis was incremental tory drift,
isplacements oF 0.5%, 1%, 2%, 4%, and 6% dif. A story drift of 656 was considered to
be a reasonable maximum a building would experience,
3.2 BAY AND BRB DESIGN
‘The bay used for testing the different splice plate designs was the prototype bay
design by Coy (2007). The bay measured 20 fet long (measured from the center ofthe
columns) and 13 feet high. Columns were W12X72s and the beam was a WI6X77, ‘The
yield strength ofthe columns and beam was 50 ksi, Strain hardening was used forthe
Rcolumns and beam, The splice plates were rectangular and had a yield strength of 36 ksi
‘The brace had yield strength of 46 ksi. The entire model used kinematic strain hardening
The BRB design was the same one used by Coy (2007). The brace axial capacity
\vas approximately 200 kips which isa typical brace strength,
3.3. PLATE DESIGN
‘The three plate variations were designed as outlined by Coy (2007). The four
criteria were shear failure, global buckling, tensile yielding of the gross cross section, and
{ensile rupture inthe net section. See Appendix A for calculations, Theee plate
thicknesses were used; 518", 3/4", and 7/8", The widths ofthe plates were chosen based
on the minimum width needed to achieve a cross section based on all four design criteria
"The Jength was based on the minis suuber Uf bulls meee to satisfy bolting
requirements; bols were not modeled. Plates were designed forthe same thickness for
all 4 postions (upper lef, lower left, upper right, and lower right). (See Figure 3-D,
Figure 3-6, and Figure 3-P),
3.4 MODELING TECHNIQUES
‘The models were created using all purpose thin sell elements with three
translational and three rotational degrees of freedom at each node. The plates were
connected to the flange through tie connections. The distance used to space out the plates
andthe flange was calculated based on plate centerline dimensions. (See Figure 3G).
‘The top nodes ofthe non-brace-side column are tied 3 rigid body to a reference point
in the center ofthe column flange. The displacement loading was applied at this
Breference point. Rigid bodies were ereated with the bottoms ofboth columns and ted to
their respective reference points. These reference points were Fixed. (See Figure 3-H).
Doe to the symmetric nature ofthe bay, data was only obtained on the side ofthe fame
Perpendicular tothe plane of the webs ofthe beam and columns. The brace wasn't
allowed to rotate about the 3 axis. (See Figure 3-3-A), ‘This simolted the buckling
restraint provided by the concrete core. Because litle yielding in the beam was expected,
the middle ofthe beam was replaced with one-dimensional beam element, thus
reducing the number of nodes and computation time
3.5 EQUIVALENT PLASTIC STRAIN (PEEQ)
Equivalent plastic strain (PEEQ) data was obtained for each position at the splice
ap. PEEQ isthe sum of the all the strains in a material during a loading, Because iti a
summation of strains it will never decrease and therefore will be a good representation of
‘cumulative damage in the spice plates after completion ofthe loading cycles,
3.6 STRAIN DUE TO MOMENT
“To compute the strain de tothe moment in the plates elastic strain data was
obiained ftom ABAQUS in the 2 direction. Elastic strain data was obtained from the top
plate and from the bottom plate onthe left and right sides, Equations 31 and 3-2 were
used to calculate te strain components de the axial load ad the moment from the
ABAQUS computer model
anus32)
‘The above equations use data forthe top and bottom plate of only one side, left ot
Fight, The total strain, €,, .asus OTE nats +8 the value obtained ditecty from the
ABAQUS analysis. ‘The axial strain, €,,., 8 the same for both plates. The result is two
‘unknowns and two equations which can easily be solved.Reference itor
displace losing
\ No rotation
shout ad,
no owot
pore
‘placement
igure 3A Boundary conditions and terminology for tt frameFigure 3-8 Left connection, cross section, and terminologySection 68"
ver eage
a Slave Sutace
ester Surface
Center tine
Strains symmetric ot
ths ne
Figure 3.€ Right connecton, crus section, and terminology
18Figure 4. Design 1 detail
Pe -o06e
Figure 34 Design 2 detailFigure 3 Design 3 detail
Splice Pstes
‘Figure 3-G Modeling techniques for spc plates (Prine 2007)romana 609
Figure 3H Beam o shell modeling tecigues
214 RESULTS AND DISCUSSION
4.1 PEEQ RESULTS
4.1.1 UPPER LEFT
For the postive and negative pushover analysis, the PEEQ was highest forthe
thickest plate and asthe plate thickness decreased so did the PEEQ (See Figure 4-A). For
{he 0.5% and the 1.0% sory drift, the PEEQ was negligible, The PEEQ for the 25, 4%,
and 6% story dit increased between drifts by a constant 0.03 to 0.04 for both the
positive and negative pushovers. The outer edges ofthe plates experienced the last
‘mount of stain while the middle ofthe plate experienced constant stain, The negative
‘vite produced a “sag” in the graphs representing a decrease in strain nthe mide ofthe
plate. ‘The sag was more noticeable asthe plate width ineased.
LOWER LEFT
"or the postive and negative pushover analysis the PEEQ was highest forthe thickest plate and as
‘he plat thickness decreased 50 did the PEEQ (See
Figure 4-B). For the 0.5% and the 1.0% story drift the PEEQ was negligible. The
positive pushover resulted in higher strains on the outside and lower strains in the middle
‘The negative pushover caused an increase in strain going from the inside ofthe plate to
the outside. The 5/8” plate showed the most extreme case of such occurrence where the
2‘outer edge had a strain evel six times that ofthe strain on the inside, A thicker plate
resulted in a
‘more uniform strain distribution for both positive and negative pushover analysis, As the
plate thickness decreased, the strains were more extreme and less consistent.
4.1.3, UPPER RIGHT
For the 0.8% story dit, the PEEQ was negligible (Sce Figure 4-C). There were
two main differences between the postive and negative pushover analysis. The first was
for the 3/4” plate; where strains inthe negative direction were about 60% higher than the
strains inthe postive direction. The second diffrence was in the S/8"plate. The
maximum strain forthe positive and negative pushover were the sam, but differs
significantly forthe 6% drt. Inthe postive pushover for drifts of 496 and 6° strains
‘were almost the same while in the negative direction were not. The same phenomenon
occurred inthe 5/8” plate inthe lower let position, Strsns were fairy constant
throughout forthe 7/8” and S/S” plates. Strains for he 5/8” plat increased from the
Inner to the outer edge of the plate
4.14 LOWER RIGHT
For the 05% story di, the PEEQ was negligible (See
Figure 4-D). The strains are lower on the inner and outer edges of the plates then
drifts, The middle exhibits
in the midale of the plat forthe negative and po
23constant strain distribution, As the plate gets thinner there is a noticeable “sag” in the
strain forthe negative pushover analysis. The positive pushover analysis resus in higher
strains than the strains forthe negative pushover analysis,
4.15. GENERAL OBSERVATION
Inall three plate designs fr al four positions, the thicker the plate, the higher the
snaxinmum PBEQ, Also for all three plate designs and al four postions, the postive
pushover almost ahvays resulted in higher maximum strains. The upper let and lower
Fight positions appeared to behave in the samme manne forall thee plate designs. The
strains were comparable inthe negative diretion and in the positive direction. ‘The lower
left and upper right had similar maximum strains occurring atthe outer edge of the plate.
4.2 ANALYTICAL MODEL
42.1, RESULTS AND DISCUSSUION
[Figure 4-E and Figure 4-F show the results ofthe analytical model and the finite