You are on page 1of 36

COLUMN SPLICE RETROFITS

PREPARED BY: HAGEN TAM

ADVISORY PANEL: TRACY BECKER


ALI ROUFEGARINEJAD
MICHAEL CHISHOLM

SEAONC UNIVERSITY RESEARCH PROGRAM


UNIVERSITY OF CALIFORNIA, BERKELEY
SPRING 2021
SEAONC UNIVERSITY RESEARCH PROGRAM
The SEAONC University Research Program supports students performing research in a Master Degree
program at partner universities. SEAONC Members, typically with support from their employer, submit
potential research topics to the SEAONC Seismology Committee for consideration. The SEAONC
Seismology Committee reviews all submissions and makes recommendations to the SEAONC Board to
fund those projects that appear to have the biggest potential to benefit SEAONC membership. Funded
projects are identified to the partner university so that a student and faculty advisor can be matched to a
topic. The student works on the project, for no more than one semester, under the supervision of an
advisory panel comprising the academic advisor, and two practicing engineers (both SEAONC members).
The student develops a written report, at project completion, and presents their research and findings to
the SEAONC membership via a live presentation.

2021 PROGRAM PRIMARY SPONSOR:

SEAONC

2021 PROGRAM CORPORATE SPONSORS:

Computers and Structures, Inc.


Core Structure, Inc.
Forell | Elsesser Structural Engineers
KPFF Consulting Engineers
Neil Moore and Associates
Slate Geotechnical Consultants

AUTHOR:
Hagen Tam is a Master’s Degree candidate at University of California, Berkeley.

ADVISORY PANEL:
Tracy Becker is an Assistant Professor at University of California, Berkeley,
faculty advisor.
Ali Roufegarinejad is a Research and Development Manager with Forell Elsesser
Engineers, corporate sponsor advisor.
Michael Chisholm is a Project Engineer with Degenkolb Engineers, SEAONC
Seismology advisor.
Table of Contents
Abstract ........................................................................................................................................... 1
1. Introduction ................................................................................................................................. 2
2. Example Building ....................................................................................................................... 3
2.1 Column Splice Demand and Capacity .................................................................................. 3
2.2 Column Splice of Interest...................................................................................................... 4
3. Pre-Retrofit Column Splice Abaqus Model ................................................................................ 6
4. Retrofit Designs .......................................................................................................................... 8
4.1 CJP retrofit......................................................................................................................... 8
4.2 Fin plates retrofit ............................................................................................................... 8
4.3 Box Plate Retrofit ............................................................................................................ 10
4.4 Extension Plate Retrofit ................................................................................................... 11
5. Retrofit Finite Element Analysis Results.................................................................................. 12
5.1 CJP Retrofit ..................................................................................................................... 13
5.2 Fin Plates Retrofit ............................................................................................................ 15
5.3 Box Plate Retrofit ............................................................................................................ 18
5.4 Extension Plate Retrofit ................................................................................................... 20
6. Conclusion ................................................................................................................................ 22
7. Future Work .............................................................................................................................. 22
8. References ................................................................................................................................. 24
Appendix A ................................................................................................................................... 25
1

Abstract

Pre-Northridge column splices with PJP welds are prone to weld fracture under tensile stresses.

Column splices typically occur every other floor, possibly affecting several columns in a building.

Like pre-Northridge beam-column moment frame connections, the causes of fracture are the low

toughness properties of the weld material and poor detailing in the pre-Northridge eras. The

standard retrofit involves removing the existing partial joint penetration (PJP) weld and replacing

it with widely accepted details with complete joint penetration (CJP) welds at the flanges and web.

This approach is relatively costly and time-consuming if required in the majority of the columns

and at multiple levels in a building. Alternate retrofit methods and their efficacy will be presented

in this paper, including but not limited to stiffening fin plates and the full CJP replacement.

To study the efficacy of the retrofit options, displacement and rotation demands at the top and

bottom of the column are obtained from an existing Perform-3D nonlinear response history

analysis of a building with pre-Northridge column splices. The displacement demands are applied

to detailed component-level Abaqus finite element models of the existing and retrofit connections.

Retrofit options are compared on the basis of reducing or eliminating the potential for crack

initiation and weld fracture by exploring stress concentrations. This paper will present a summary

of retrofit options applied to column splices in projects around the Bay Area and the effectiveness

of these various options in addressing the fracture problem at column splices.


2

1. Introduction

Pre-Northridge beam-column connections are a well-known deficiency in older steel-framed

buildings; however, pre-Northridge steel moment frame buildings also have a lesser known

potential failure mode: column splice fracture. These pre-Northridge column splices fracture in a

brittle manner due to low toughness weld material and poor connection detailing (Roufegarinejad

et al. 2015). Currently, the most widely accepted typical retrofit involves removing the deficient

PJP weld and replacing it with a CJP at the web and both flanges, as appropriate. Though this

method is widely accepted as an adequate retrofit, it is expensive both in terms of cost and

fabrication time. Further, these CJP welds are often more conservative than is actually required

and could be replaced by better quality, properly detailed PJP welds, especially for shorter

structures (Shen et al. 2010).

Given Shen’s findings that PJP welds may be adequate, researchers worked to demonstrate the

suitability of PJP welds in practice. Shaw et al. (2015) modeled column splices with five different

configurations of varying column sizes, web connection type, and flange weld thicknesses. All

designs showed the ability to achieve significant inelastic deformation capacity, even though they

were designed to remain elastic as a force-controlled element. In particular, they found that the

splices without a weld access hole (WAH) and without a welded web splice performed well. Under

all configurations, the expected weld toughness was greater than the toughness demands from the

loading protocol, suggesting these configurations may be used in general practice. Although

Shaw’s research focused on new construction, these findings suggest that the full CJP replacement

for these column splices may not be the optimal solution. Often, the CJP retrofit is too expensive,

too conservative, or both and has the potential to be replaced by adequate and more cost-effective

retrofit options.
3

This paper will summarize the joint research project between UC Berkeley and SEAONC that

examines this column splice deficiency and the adequacy and effectiveness of several retrofit

schemes for an example building. Analysis of the example building is completed using the

nonlinear response history analysis (NLRHA) capabilities of Perform-3D. The displacements and

rotations from the NLRHA are imposed onto finite element Abaqus models of several different

retrofit schemes. These schemes were loosely based on previous retrofit solutions applied to mid-

to high-rise buildings in the Bay Area by Forell Elsesser Engineers and Degenkolb Engineers.

2. Example Building

This study will primarily look at a seven-story steel structure in San Francisco that was recently

seismically retrofitted. The existing lateral load resisting system in the building consists of a steel

moment-resisting frame where all columns are moment-connected in both directions. These

columns were spliced at the second, fourth, and sixth floors.

2.1 Column Splice Demand and Capacity

Forell Elsesser Engineers supplied the Perform-3D model of the building that was used for an

NLRHA in the original retrofit. The main purpose of the Perform-3D model was to establish

typical demands on column splices to be applied in Abaqus, a finite element software. To perform

a nonlinear response history analysis, a suite of seven ground motions, scaled to the BSE-2E level,

was run through the Perform 3D model. Afterwards, the moment and axial force response histories

were exported at the column splices at each corner column. Corner columns were selected because

they were subject to the large uplift forces from the overturning moment. The bending moment

and axial force response series were used together to determine the column splice with the greatest
4

tensile stress in the flange for any one time-step. This stress of 37.4 ksi is calculated using AISC

342 Draft Eqn. C5-14, reproduced here,

𝑃𝑃𝑈𝑈𝑈𝑈 𝑀𝑀𝑈𝑈𝑈𝑈,𝑥𝑥 𝑀𝑀𝑈𝑈𝑈𝑈,𝑦𝑦


𝜎𝜎𝑈𝑈𝑈𝑈 = � �±� �±� �
𝐴𝐴𝑔𝑔 𝑆𝑆𝑥𝑥 𝑆𝑆𝑦𝑦

where PUF is the axial force, A is the gross area, MUF,x and MUF,y are the bending moments in the

x- and y- directions, respectively, and Sx and Sy are the strong- and weak-axis section moduli. The

tensile stress demand was compared to the critical fracture stress determined from the weld fracture

toughness, crack length, and several other parameters. This critical fracture stress σcr is calculated

by AISC 342 Draft Eqn. C5-12 and C5-13 (similar to Barsom & Rolfe (2006)), as

𝐾𝐾𝐼𝐼𝐼𝐼
𝜎𝜎𝑐𝑐𝑐𝑐 =
1.6𝑎𝑎 𝑎𝑎
�2.3 − 𝑡𝑡 0 � �4.6 ∗ 𝑡𝑡 0 � �𝜋𝜋𝑎𝑎0
𝑓𝑓,𝑢𝑢 𝑓𝑓,𝑢𝑢

with the fracture toughness KIC from Table C5.3. Given a Charpy V-Notch (CVN) toughness of 8

ft-lb from previous testing done by Roufegarinejad et al in 2015, Table C5.3 returns a KIC of 80

ksi√in. The crack length a0 is the unwelded thickness of the flange, and tf,u is the flange thickness

of the column above the splice. With a crack length of 1.22 in. and flange thickness of 2.47 in., the

critical fracture stress is just 12 ksi. The pre-retrofit column splice connection is clearly deficient

in resisting the seismic demands.

2.2 Column Splice of Interest

The column splice of concern is located on the second floor and connects a W14x500 below and

a W14x342 above. The flanges are connected with a 1.25 in. PJP weld from the outer edge of the

smaller column while the webs are bolted together as shown in Figure 1. To investigate the

behavior at the column splice, the finite element model must include more than just the splice
5

itself. The column was modeled from floor to floor to capture the stresses at the column splice.

The translational and rotational displacements at the floor above and floor below were exported

from Perform-3D and applied in Abaqus. Displacement were applied, instead of forces, because

strains are what ultimately lead to failure of the column splice. These displacement demands are

summarized in Table 1.

Figure 1: Example building existing column splice condition (Roufegarinejad 2021)


6

Table 1: Displacement demands for the Abaqus column assembly

H2
H1 V H1 Rotation H2 Rotation V
Translation
Translation Translation (Strong-axis (Weak-axis Rotation
(Strong-
(Weak-axis) (Axial) Bending) Bending) (Torsion)
axis)

Units in. in. in. rad. rad. rad.

Floor
-6.9752 8.7873 -0.0426 -0.0090 -0.0088 0.0004
Above

Floor
-4.2168 5.2052 -0.0258 -0.0284 -0.0212 0.0002
Below

3. Pre-Retrofit Column Splice Abaqus Model

The pre-Northridge column splice of the example building is modeled in Abaqus, as shown in

Figure 2. The model consists of the column from the floor below to the floor above, and the

respective displacement demands from Perform-3D were applied at each end. This finite element

model reflects the pre-Northridge column splice and is composed of 4 structural elements: the

W14x500 below, the W14x342 above, and two 5/8 in. plates bolted on either side of the web. In

agreement with the as-constructed condition, one face of the web of each wide-flange column is

aligned such that one of the plates is flush with both columns.

The PJP welds at the column flanges are modeled as tie constraints between the two columns, as

shown in red in Figure 3. The bolts are modeled with tie constraints between opposite faces of the

bolt holes on the tension side of the column, shown with larger yellow circles in Figure 3. This

assumes the strength of the bolts are sufficient and the yielding would occur in the 5/8 in. splice

plates. To model this bearing, a “hard” contact with normal behavior is defined at the unwelded

regions of the column face. This essentially means that when the material is put into compression,
7

they will experience compressive bearing stresses normal to the contact surface. All elements

within 12 in. of the splice location were meshed to a quarter inch.

All material is assumed to have a bilinear behavior with strain hardening after yielding, with a

hardening modulus of 375 ksi (1.3% of E). More accurate material property information and force-

displacement curve may be included in future studies for increased accuracy. The expected

strengths were determined following ASCE41-06 procedures in Table 5-2 and Table 5-3 for

ASTM A36 steel (ASCE41-17 Table 9-1 and 9-3). Groups for the column sizes are determined

using ASTM A6 Table A. To apply the deformations at each end, the points on each face of the

column are tied together using a multi-point constraint. Then, the displacement demands from

Perform-3D are applied to the midpoint of each face as boundary conditions.

Figure 2: Abaqus assembly of column splice (rotated 90°)

Figure 3: Unmeshed Abaqus assembly tie constraints


8

4. Retrofit Designs

Two retrofit approaches are explored. Each approach is modeled in Abaqus to determine the

adequacy and effectiveness of the retrofit.

4.1 CJP retrofit

As a control, the CJP retrofit where the web and both flanges are welded together with CJP welds,

is modeled to create a best-scenario retrofit. The bolted splice plates are removed, and each flange

and the web are welded together with a CJP. In the actual retrofit of the example building, Forell

Elsesser Engineers used this retrofit to minimize the potential for crack propagation.

In the retrofit of a similar building in San Francisco, Degenkolb Engineers used an additional PJP

weld to complete the existing connection. For retrofit of wide-flange columns as seen in Figure 4,

the retrofit was primarily limited by access and shoring. This retrofit included installing fin plates

to the extreme fibers of the existing column splices as shoring, removing of a portion of the web,

and adding a new PJP weld at the interior flange face. This process creates a hybrid CJP weld with

the new and existing PJP welds. Once the new PJP was added, a plate was added to replace the

removed web sections.

4.2 Fin plates retrofit

The fin plates retrofit was designed by adding fin plates to the exterior edges of the column to

increase the section modulus, and decrease the stresses felt at the flanges. These additional fin

plates increase the area, moment of inertia, and section modulus of the splice section, thereby

reducing the tensile stresses on the PJP weld. A fin plates retrofit detail for the wide-flanges in our

example building has been designed and modeled in Abaqus for analytical testing.
9

To design the fin plates retrofit, large steel plates were added to the extreme fibers of the columns

at the splice section to increase the section modulus at the splice. The design details can be found

in Figure 5. Using linear section analyses, the stresses at the splice location were determined. The

size of the fin plates was increased until the section modulus was sufficient to decrease the stress

at the PJP weld to below the critical fracture stress. These calculations are found in Appendix A

of this report. Once the minimum size of the fin plates was determined, each plate was checked

for local buckling limits provided in AISC 341-16 T. D1.1. Finally, the fillet welds were sized to

resist the total force transferred into the fin plate. Going through this process resulted in 2 in. by 7

in. plates with 20 in. long, 1/2 in. fillet welds on either side of the plate, as shown in Figure 5.

Similar to the original PJP weld, these fillet welds were modeled in Abaqus as tie constraints.

Figure 4: Reinforcing PJP wide-flange retrofit details


10

Figure 5: Fin plates retrofit design

Due to the unexpected stress concentrations in the Abaqus results of the fin plates retrofit, two

different configurations of fin plates were also examined. These configurations are referred to as

the box plate retrofit and extension plate retrofit. Both of these options were modeled in Abaqus

and further examined in the following section.

4.3 Box Plate Retrofit

In the box plate retrofit, additional plates with the same height and thickness as the fin plates are

added from flange to flange in line with the fin plates. These box plates aim to reduce the shear

lag issues by turning the wide flange into a box column at the splice location. The box plate retrofit

design can be seen in Figure 6. Linear bending stress calculations were done to check the demand-

capacity ratio (DCR) of the retrofit in Appendix A. Based on these calculations, this retrofit should

reduce the tensile stress on the PJP weld to 9 ksi, which is well below the critical fracture stress.
11

Figure 6: Box plate retrofit design

4.4 Extension Plate Retrofit

The other redesign for the fin plates retrofit included the addition of plates as extensions on either

side of the web and both flanges as seen in Figure 7. The flange extensions were four 2x5 plates

with a 1 in. cut to avoid the imperfections and extrusions from welding. The web extensions are

the same size as the fin plates but shifted to be in line with the web of the column above. Similarly,

these extension plates aim to minimize shear lag issues while still increasing the section modulus

to sufficiently reduce the tensile stresses at the weld. Calculations for the DCR assuming a linear

stress distribution are included in Appendix A. These calculations predict a stress of 10 ksi, which

is still below the critical fracture stress.


12

Figure 7: Extension plate retrofit design

5. Retrofit Finite Element Analysis Results

An implicit nonlinear static pushover, finite element analysis using standard solver in Abaqus was

run to determine the von Mises stresses on each element. Von Mises stresses are studied because

they represent the equivalent stresses on the material prior to yield. Ideally, the column above the

splice would yield prior to fracture of the weld, thereby preventing the brittle, premature failure.

If the von Mises stresses are lower than the critical fracture stress, then it follows that the weld

will not experience premature fracture. Figure 8 shows the column splice of the unretrofitted. The

inherent crack created by the PJP, circled in red in Figure 8, is widening, creating a stress

concentration at the tip of the weld of 52 ksi. This is well above the fracture stress capacity of 12

ksi calculated from AISC 342 (2020), and the weld will clearly fracture. The same displacements

and rotations given in Table 1 are applied to the retrofit models as well.
13

Figure 8: Von Mises stresses on pre-retrofit column splice after imposed displacements

5.1 CJP Retrofit

The FEA results of the CJP Retrofit yielded expected results. The peak stress at the weld is 49.5

ksi and the stresses in the weld and the extreme fibers of the columns are similar to those found in

the pre-retrofit model. However, the new CJP no longer has the unwelded portion of the flange

that creates an inherent crack, and thus, does not have the stress-concentration and prying effect

on the weld. The CJP also uses better weld material with a higher fracture toughness and fracture

strength that can sustain the equivalent stresses (FEMA 355b). The von Mises stresses at the

tension corner of the CJP retrofitted column splice is shown in Figure 9, while Figure 10 shows

the von Mises stresses at a section cut of the splice. The tensile stresses are shown in the bottom

right corner with only stresses larger than the critical fracture stress colored.
14

Figure 9: Von Mises stresses on the CJP retrofit

Figure 10: Von Mises stresses on CJP retrofit on splice section, only stresses above 12 ksi are

shown in color
15

5.2 Fin Plates Retrofit

Because the displacements were taken from the Perform3D model prior to any retrofits, the forces

on the column will not be the same. As discussed previously, the addition of fin plates will stiffen

the section and result in new deformations in the column. Thus, to maintain consistency and have

comparable results, the fin plate FEA results are shown at the time step where the overall bending

moment in the splice is similar.

The FEA results of the fin plates retrofit yielded some unexpected results. The peak stress at the

weld was slightly reduced, from 52 ksi to 49 ksi, as seen in Figure 11. This was unexpected as the

fin plates were designed to reduce the stress at the weld to a level below the critical fracture stress

of 12 ksi. In Figure 12, the von Mises stresses greater than 12 ksi are shown at the splice section

with the bottom right in tension. There is a substantial portion of the PJP that exceeds 12 ksi, which

is problematic.

Unlike in the design assumptions, there was no linear stress distribution with peak stresses at the

extreme fiber of the fin plates. This problem stems from the complex geometry of the fin plates to

avoid the weld imperfections. This coincides with Shaw et al. (2015) where specimens without a

WAH performed better. The complicated geometry of the connection also creates a shear lag that

results in the tensile stresses not transferring completely into the fin plate; rather, the fin plate has

larger tension along the connection with the column flange, resulting in compression in the free

edge of the fin plate.

To better illustrate this issue, the deformations were isolated in each direction and applied to the

model in separate analyses. The column above applies a tensile force at the fixed end of the fin

plate such that the plate bends as indicated in Figure 13. Here, the vertical stresses, S33, resulting

from only strong-axis bending are examined. The deformations are scaled by 100 for clarity. It is
16

important to note that this design does concentrate the deformations in the column above. With the

additional fin plates around the splice section, region around the splice is much stiffer. Thus, the

deformation demands are now moved outside of the splice region and into the column above. This

may result in a level of yielding and plasticity that decreases the stresses at the weld.

This shear lag issue is exacerbated by the eccentricity of the load application. Because the PJP

weld only attaches a portion of the column flanges, there is an eccentricity between the tension

from the column above and the resisting force in the weld. This eccentricity, shown in Figure 14,

creates a larger bending moment in the same direction as the moment from the issue discussed

previously.

Figure 11: Von Mises stresses on the fin plates retrofit


17

Figure 12: Von Mises stresses on the fin plates retrofit at splice section, only stresses above 12

ksi are shown in color

Figure 13: Vertical tensile stresses, S33, from isolated strong-way bending only, displacement is

magnified by 100
18

Figure 14: Pre-Northridge PJP weld eccentricity

5.3 Box Plate Retrofit

The FEA results of this retrofit were closer to the expected results. A smaller portion of the existing

PJP reaches the critical fracture stress of 12 ksi and the maximum stress at the splice is 31 ksi

(compared to 49 ksi in the original fin plate retrofit design) as seen in Figure 15. Although the

critical fracture stress is still exceeded, the box plates did help reduce the problem with the shear

lag. The box plate on the interior edge of the flange holds the PJP crack together, thus reducing

the tensile stresses felt by the weld. By constraining the free end of the PJP weld, this box plate

also mitigates the effects of the weld eccentricity. However, there are still some shear lag issues.

The free edge of the fin plates is in tension, but the tension on the fixed edge of the fin plate is still

larger.

Once again, there is a concentration of the deformations in the column above due to stiffening of

the splice region. This will likely result in yielding and plasticity in the column above prior to the
19

premature fracture of the PJP weld. Furthermore, because the fracture stresses are localized as seen

in Figure 16, and there is a large increase in the net section area, it is possible that a fracture in the

PJP will not result in complete loss of strength. However, the retrofit was not designed to resist

the demand without the PJP and sudden loss of strength may amplify the tension, thus, this will

require further research.

Figure 15: Von Mises stresses on box plate retrofit


20

Figure 16: Von Mises stresses on box plate retrofit at splice section

5.4 Extension Plate Retrofit

Figure 17 shows the FEA results of the extension plate retrofit with the same limits where only the

stresses exceeding the critical fracture capacity are colored. The peak stress in the column splice

weld is reduced to 43 ksi from the original 52 ksi. Though it is an improvement to the original fin

plate retrofit, there is a larger portion of the weld that exceeds the critical fracture stress as seen in

Figure 18. This retrofit is still problematic because of the weld eccentricity and remaining shear

lag issue. As opposed to the box plate retrofit, the extension plate retrofit does not constrain the

unwelded edge of the column flange. This results in larger tensile stresses over a larger portion of

the PJP weld. Similar to the original fin plate retrofit, the deformations are concentrated in the

column above due to the stiffened splice region. However, the box plate retrofit seems to be a

better option in reducing the stresses felt at the PJP weld.


21

Figure 17: Von Mises stresses on extension plate retrofit

Figure 18: Von Mises stresses on extension plate retrofit at splice section
22

6. Conclusion

Pre-Northridge steel moment-frame column splice connections are a deficiency that has not been

studied as extensively as beam-column connections. The most widely accepted details for general

retrofit of these column splices include costly procedures to replace the deficient PJP weld with a

CJP weld. This study examined alternative fin plate retrofit designs for an example project.

Although the stiffened fin plates retrofit concentrate the deformations away from the column

splice, as it is currently designed, it does not sufficiently reduce the stress at the weld to an

acceptable level. The FEA has made apparent the shear flow through the fin plates did not follow

the initially expected path with increasing tensile stresses further away from the neutral axis. Two

modifications of fin plate retrofit were further examined. With the box plate and extension plate

retrofits, the behavior is improved; however, they still do not follow the linear stress distribution.

Both of these retrofits reduced the stress at the existing weld with stress distributions that were

more intuitive. All three plate retrofits displayed larger stress demands at the weld compared to

the critical fracture stress.

7. Future Work

This topic still needs significant research as this study assumed that the displacements and rotations

from Perform3D remained the same throughout the retrofits, and only adjusted to similar section

forces at the splice location. Though the effect is minor, in future research with more time, the

Perform3D model should be updated to account for the retrofit’s stiffness, such that the

deformations correspond to a NLRHA of the retrofit.

More importantly, this study was purely a finite element analysis of a few retrofit options. The

findings from this study can be useful for design of future models or physical specimen to be
23

tested. Further research could include physical testing of all three plate retrofits and the CJP

retrofit, examination of different plate configurations, or design different retrofits entirely.


24

8. References
AISC. (2020). Seismic Provisions for Evaluation and Retrofit of Existing Structural Steel
Buildings, AISC 342 Draft for public review. Chicago, IL: American Institute of Steel
Construction.
ASCE. (2006). Seismic evaluation and retrofit of existing structures, ASCE/SEI 41-06. Reston,
VA: American Society of Civil Engineers. doi:https://doi.org/10.1061/9780784408841
ASCE. (2016). Minimum design loads for buildings and other structures, ASCE/SEI 7-16. Reston,
VA: American Society of Civil Engineers. doi:https://doi.org/10.1061/9780784414248

ASCE. (2017). Seismic evaluation and retrofit of existing structures, ASCE/SEI 41-17. Reston,
VA: American Society of Civil Engineers. doi:https://doi.org/10.1061/9780784414859

ASTM A6 / A6M-10a, Standard Specification for General Requirements for Rolled Structural
Steel Bars, Plates, Shapes, and Sheet Piling, ASTM International, West Conshohocken, PA,
2010, www.astm.org

Barsom, J., & Rolfe, S. (2006). Chapter 1 overview of the problem of fracture and fatigue in
structures. Fracture and Fatigue Control in Structures: Applications of Fracture Mechanics,
Third Edition. doi:10.1520/mnl10323m

Barsom, J., & Rolfe, S. (2006). Chapter 2 stress analysis for members with cracks—. Fracture and
Fatigue Control in Structures: Applications of Fracture Mechanics, Third Edition.
https://doi.org/10.1520/mnl10324m

Chisholm, M. (2021, March 23). Personal communication.


Roufegarinejad, A., Nudel, A., Marusich, S., & Dana, M. (2015, December 3). Evaluation and
remediation of pre-northridge steel moment frame column splices.
https://ascelibrary.org/doi/10.1061/9780784479728.024.
Roufegarinejad, A. (2021, February 18). Personal communication.

SAC Joint Venture (2000). State of the art report on welding and inspection, FEMA 355b.
Washington, D.C.: Federal Emergency Management Agency (FEMA).

Shaw, S., Stillmaker, K., & Kanvinde, A. (2015). "Seismic Response of Partial-Joint-Penetration
Welded Column Splices in Moment-Resisting Frames," Engineering Journal, American
Institute of Steel Construction, Vol. 52, pp. 87-108.
Shen, J., Sabol, T., Akbas, B., & Sutchiewcharn, N. (2010). “Seismic Demand on Column Splices
in Steel Moment Frames,” Engineering Journal, American Institute of Steel Construction,
Vol. 47, pp. 223-240.
25

Appendix A
University of California, Berkeley

Davis Hall
Berkeley, CA 94720
Phone: 916.752.4311

Project: Column Splice Project Job Number: Fin Plates Retrofit Date: 5/28/2021
Section: Appendix A By: Hagen Tam
Checked By:

Existing Column and Splice Properties


Wide-Flange Properties:
Column Below: W14x500 Column Above: W14x342
depth, dcb = 19.60 in depth, dca = 17.50 in
width, b = 17.00 in width, b = 16.40 in
web thickness, tw = 2.19 in web thickness, tw = 1.54 in
flange thickness, tf = 3.50 in flange thickness, tf = 2.47 in

Wide-Flange Material:
ASTM Designation: A36
Building Date: 1961-1990
Group: 4 ASTM A6 Table A
yield strength, fy = 37 ksi ASCE41-17 Tbl 9-1
ultimate strength, fu = 62 ksi ASCE41-17 Tbl 9-1
expected strength factor, exp = 1.1 ASCE41-17 Tbl 9-3
expected yield strength, fye = 40.7 ksi
expected ultimate strength, fue = 68.2 ksi

Weld Properties:
weld leg size, w = 1.25 in per original drawings
Charpy V-Notch value, CVN = 8 ft-lbs. 16th percentile (Kaufmann and Fisher 1997)

Fin Plate Retrofit Properties


Fin Plate Reinforcement Properties:
ASTM Designation: A572, Gr 50
yield strength, fyR = 50 ksi
fin plate thickness, tpR = 2 in
fin plate length, dpR = 7.00 in
number of fin plates, NpR = 4 plates
dist. of fin plate to col above, l = 1.5 in
University of California, Berkeley

Davis Hall
Berkeley, CA 94720
Phone: 916.752.4311

Project: Column Splice Project Job Number: Fin Plates Retrofit Date: 5/28/2021
Section: Appendix A By: Hagen Tam
Checked By:

Existing Capacity Checks


Existing Column Capacity:
gross section area, Ag = 101
in2
tension capacity, Tu = 4110.7 kips 𝑇 𝐴 𝑓

Existing Load in Column:


gravity axial compression, PG = -170.5 kips Perform3D Gravity Load Case
compressive stress, σG = -1.6881 ksi

Existing Splice Capacity:


critical fracture toughness, KIC = 80 ksi-√in AISC 342 Tbl C5.3
PJP crack length, a0 = 1.22 in
base material yield strength, σye = 40.7 ksi AISC 342, Eqn. C5-14
base materail ult. strength, σue = 34.5142 ksi AISC 342, Eqn. C5-14
weld fracture strength, σwe = 11.9129 ksi AISC 342, Eqn. C5-14 & C5-13
critical stress, σcr = 11.9129 ksi 𝜎 min 𝜎 , 𝜎 , 𝜎

PJP Weld Demands with Retrofit


Gravity Axial Stress Demands
gravity axial compression, PG = -170.5 kips Perform3D Gravity Load Case
compressive stress, σG = -1.6881 ksi

Seismic Axial Stress Demands:


tension demand, Tu = -353.86 kips Perform 3D Response History
gross section area, Ag = 101 in2

fin plate thickness, tpR = 2 in


fin plate width, bpR = 7 in
number of fin plates, NpR = 4 plates
fin plate area, ApR,tot = 56 in2 𝐴 , 𝑁 𝑏 𝑡
total area with retrofit, Atot = 157 in2
𝐴 𝐴 𝐴 ,
axial stress, σP = -2.25 ksi
University of California, Berkeley

Davis Hall
Berkeley, CA 94720
Phone: 916.752.4311

Project: Column Splice Project Job Number: Fin Plates Retrofit Date: 5/28/2021
Section: Appendix A By: Hagen Tam
Checked By:

Bending Stress Demands:


strong-way bending demand, Mux = 9035 kip-in Perform 3D Response History
weak-way bending demand, Muy = -5101 kip-in Perform 3D Response History

b h Ix,self Iy,self A dx dy Ix Iy
W14x342 - - 4900 1810 101 0 0 4900 1810
vert. fin plate 1 2 7 57.1667 4.66667 14 6.2 13.75 2704.04 542.827
vert. fin plate 2 2 7 57.1667 4.66667 14 6.2 13.75 2704.04 542.827
vert. fin plate 3 2 7 57.1667 4.66667 14 6.2 13.75 2704.04 542.827
vert. fin plate 4 2 7 57.1667 4.66667 14 6.2 13.75 2704.04 542.827
total: 15716.2 3981.31
cy, cx = 17.25 8.2
Sx, Sy = 911.082 485.525
section modulus about x, Sx = 911.08 in 3

section modulus about y, Sy = 485.53 in3


max. strong-way stress, σMx = 9.92 ksi
max. weak-way stress, σMy = 10.51 ksi

The stress at the weld is not the same as the


stress at the extreme tension fiber of the
retrofit. The weld will experience a lower stress
level that is has a linear relationship with the
distance to the centroid. This stress level is
depicted in the image on the right. σMx
σMy

extreme fiber of weld, cweld,y = 8.75 in 𝑐 , 𝑑 /2


𝑑
extreme fiber of section, cy = 17.25 in 𝑐 𝑙 𝑡
2
𝑐 ,
strong-way stress on weld, σw,Mx = 5.03048 ksi 𝜎 , 𝜎
𝑐

extreme fiber of weld, cweld,x = 8.2 in 𝑐 , 𝑑 /2


𝑑
extreme fiber of section, cx = 8.2 in 𝑐 𝑙 𝑡
2
𝑐 ,
weak-way stress on weld, σw,My = 10.5066 ksi 𝜎 , 𝜎
𝑐

Total Stress on the PJP Weld:


ultimate stress, σU = 11.60 ksi
weld stress DCR = 0.97332 OK
University of California, Berkeley

Davis Hall
Berkeley, CA 94720
Phone: 916.752.4311

Project: Column Splice Project Job Number: Fin Plates Retrofit Date: 5/28/2021
Section: Appendix A By: Hagen Tam
Checked By:

Plate Local Buckling Checks:


Slenderness Limits:
elastic modulus, E = 29000 ksi assumed
yield strength, Fy = 50 ksi
limiting width-to-thickness ratio, λr = 8 AISC 341-16 Tbl. D1.1

Plate Slenderness:
full plate width, bpR = 8.50 in 𝑏 𝑑 𝑙
plate thickness, tpR = 2 in
width-to-thickness ratio, b/t = 4.25 The section is compact; no local buckling issue

Size Fin Plate Fillet Weld
stress where fin plate meets col, σ1 = 13.28 ksi 𝜎 𝜎 𝜎 , 𝜎 ,
stress at extreme fiber, σ2 = 18.17 ksi 𝜎 𝜎 𝜎 𝜎
average stress on fin plate, σfin = 15.73 ksi 𝜎 0.5 ∗ 𝜎 𝜎
fin plate area, Afin = 14 in2
equiv. fin axial force, Pfin = 220.17 kips 𝑃 𝜎 𝐴

→ Try two 1/2" thick, 20" long weld per fin plate
fillet weld thickness, ww = 0.5 in
fillet weld throat, tw = 0.3536 in.
fillet weld length, lw = 40 in
shear per lineal inch from P, rpu = 5.5043 kip/in AISC Manual v15 Eqn. 8-5a

strong-way moment of inertia, Ix = 1899.8 in4


weak-way moment of inertia, Iy = 0.1473 in4
polar moment of inertia, Ip = 1899.9 in4
load eccentricity, ew = 4.25 in
distance to extreme fiber, c = 4.25 in
shear per lineal inch from M, rmu = 2.09319 kip/in AISC Manual v15 Eqn. 8-8a

max resultant shear, ru = 7.5975 kip/in AISC Manual v15 Eqn. 8-11a

strength reduction factor, φ = 0.75


weld strength, FEXX = 70
weld capacity, φrn = 11.1
DCR:  0.6822 OK
University of California, Berkeley

Davis Hall
Berkeley, CA 94720
Phone: 916.752.4311

Project: Column Splice Project Job Number: Box Plate Retrofit Date: 5/28/2021
Section: Appendix A By: Hagen Tam
Checked By:

Existing Capacity Checks


Existing Column Capacity:
gross section area, Ag = 101
in2
tension capacity, Tu = 4110.7 kips 𝑇 𝐴 𝑓

Existing Load in Column:


gravity axial compression, PG = -170.5 kips Perform3D Gravity Load Case
compressive stress, σG = -1.6881 ksi

Existing Splice Capacity:


critical fracture toughness, KIC = 80 ksi-√in AISC 342 Tbl C5.3
PJP crack length, a0 = 1.22 in
base material yield strength, σye = 40.7 ksi AISC 342, Eqn. C5-14
base materail ult. strength, σue = 34.5142 ksi AISC 342, Eqn. C5-14
weld fracture strength, σwe = 11.9129 ksi AISC 342, Eqn. C5-14 & C5-13
critical stress, σcr = 11.9129 ksi 𝜎 min 𝜎 , 𝜎 , 𝜎

PJP Weld Demands with Retrofit


Gravity Axial Stress Demands
gravity axial compression, PG = -170.5 kips Perform3D Gravity Load Case
compressive stress, σG = -1.6881 ksi

Seismic Axial Stress Demands:


tension demand, Tu = -353.86 kips Perform 3D Response History
gross section area, Ag = 101 in2

fin plate area, ApR,tot = 106 in2 𝐴 , 𝑁 𝑏 𝑡


total area with retrofit, Atot = 207 in2
𝐴 𝐴 𝐴 ,
axial stress, σP = -1.71 ksi
University of California, Berkeley

Davis Hall
Berkeley, CA 94720
Phone: 916.752.4311

Project: Column Splice Project Job Number: Box Plate Retrofit Date: 5/28/2021
Section: Appendix A By: Hagen Tam
Checked By:

Bending Stress Demands:


strong-way bending demand, Mux = 9035 kip-in Perform 3D Response History
weak-way bending demand, Muy = -5101 kip-in Perform 3D Response History

b h Ix,self Iy,self A dx dy Ix Iy
W14x342 - - 4900 1810 101 0 0 4900 1810
Fin Plate 1 2 7 57.1667 4.66667 14 6.2 13.75 2704.04 542.827
Fin Plate 2 2 7 57.1667 4.66667 14 6.2 13.75 2704.04 542.827
Fin Plate 3 2 7 57.1667 4.66667 14 6.2 13.75 2704.04 542.827
Fin Plate 4 2 7 57.1667 4.66667 14 6.2 13.75 2704.04 542.827
Box Plate 1 2 12.56 330.231 8.37333 25.12 6.2 0 330.231 973.986
Box Plate 2 2 12.56 330.231 8.37333 25.12 6.2 0 330.231 973.986
total: 16376.6 5929.28
cy, cx = 17.25 7.2
Sx, Sy = 949.37 823.511
section modulus about x, Sx = 949.37 in3
section modulus about y, Sy = 823.51 in3
max. strong-way stress, σMx = 9.52 ksi
max. weak-way stress, σMy = 6.19 ksi

The stress at the weld is not the same as the


stress at the extreme tension fiber of the
retrofit. The weld will experience a lower stress
level that is has a linear relationship with the
distance to the centroid. This stress level is
depicted in the image on the right. σMx
σMy
extreme fiber of weld, cweld,y = 8.75 in 𝑐 , 𝑑 /2
𝑑
extreme fiber of section, cy = 17.25 in 𝑐 𝑙 𝑡
2
𝑐 ,
strong-way stress on weld, σw,Mx = 4.8276 ksi 𝜎 , 𝜎
𝑐

extreme fiber of weld, cweld,x = 8.2 in 𝑐 , 𝑑 /2


𝑑
extreme fiber of section, cx = 7.2 in 𝑐 𝑙 𝑡
2
𝑐 ,
weak-way stress on weld, σw,My = 7.05483 ksi 𝜎 , 𝜎
𝑐

Total Stress on the PJP Weld:


ultimate stress, σU = 8.49 ksi
weld stress DCR = 0.7124 OK
University of California, Berkeley

Davis Hall
Berkeley, CA 94720
Phone: 916.752.4311

Project: Column Splice Project Job Number: Extension Plate Retrofit Date: 5/28/2021
Section: Appendix A By: Hagen Tam
Checked By:

Existing Capacity Checks


Existing Column Capacity:
gross section area, Ag = 101
in2
tension capacity, Tu = 4110.7 kips 𝑇 𝐴 𝑓

Existing Load in Column:


gravity axial compression, PG = -170.5 kips Perform3D Gravity Load Case
compressive stress, σG = -1.6881 ksi

Existing Splice Capacity:


critical fracture toughness, KIC = 80 ksi-√in AISC 342 Tbl C5.3
PJP crack length, a0 = 1.22 in
base material yield strength, σye = 40.7 ksi AISC 342, Eqn. C5-14
base materail ult. strength, σue = 34.5142 ksi AISC 342, Eqn. C5-14
weld fracture strength, σwe = 11.9129 ksi AISC 342, Eqn. C5-14 & C5-13
critical stress, σcr = 11.9129 ksi 𝜎 min 𝜎 , 𝜎 , 𝜎

PJP Weld Demands with Retrofit


Gravity Axial Stress Demands
gravity axial compression, PG = -170.5 kips Perform3D Gravity Load Case
compressive stress, σG = -1.6881 ksi

Seismic Axial Stress Demands:


tension demand, Tu = -353.86 kips Perform 3D Response History
gross section area, Ag = 101 in2

fin plate area, ApR,tot = 60 in2 𝐴 , 𝑁 𝑏 𝑡


total area with retrofit, Atot = 161 in2
𝐴 𝐴 𝐴 ,
axial stress, σP = -2.20 ksi
University of California, Berkeley

Davis Hall
Berkeley, CA 94720
Phone: 916.752.4311

Project: Column Splice Project Job Number: Extension Plate Retrofit Date: 5/28/2021
Section: Appendix A By: Hagen Tam
Checked By:

Bending Stress Demands:


strong-way bending demand, Mux = 9035 kip-in Perform 3D Response History
weak-way bending demand, Muy = -5101 kip-in Perform 3D Response History

b h Ix,self Iy,self A dx dy Ix Iy
W14x342 - - 4900 1810 101 0 0 4900 1810
Top Web Ext. 2 7 57.1667 4.66667 14 0 13.75 2704.04 4.66667
Bot Web Ext. 2 7 57.1667 4.66667 14 0 13.75 2704.04 4.66667
Top Left Flange Ext. 4 2 2.66667 10.6667 8 -10.2 10.985 968.028 842.987
Top Right Flange Ext. 4 2 2.66667 10.6667 8 -10.2 10.985 968.028 842.987
Bot Left Flange Ext. 4 2 2.66667 10.6667 8 -10.2 10.985 968.028 842.987
Bot Right Flange Ext. 4 2 2.66667 10.6667 8 -10.2 10.985 968.028 842.987
total: 14180.2 5191.28
cy, cx = 17.25 13.2
Sx, Sy = 822.04 393.279
section modulus about x, Sx = 822.04 in3
section modulus about y, Sy = 393.28 in3
max. strong-way stress, σMx = 10.99 ksi
max. weak-way stress, σMy = 12.97 ksi

The stress at the weld is not the same as the


stress at the extreme tension fiber of the
retrofit. The weld will experience a lower stress
level that is has a linear relationship with the
distance to the centroid. This stress level is
depicted in the image on the right. σMx
σMy
extreme fiber of weld, cweld,y = 8.75 in 𝑐 , 𝑑 /2
𝑑
extreme fiber of section, cy = 17.25 in 𝑐 𝑙 𝑡
2
𝑐 ,
strong-way stress on weld, σw,Mx = 5.57537 ksi 𝜎 , 𝜎
𝑐

extreme fiber of weld, cweld,x = 8.2 in 𝑐 , 𝑑 /2


𝑑
extreme fiber of section, cx = 13.2 in 𝑐 𝑙 𝑡
2
𝑐 ,
weak-way stress on weld, σw,My = 8.05775 ksi 𝜎 , 𝜎
𝑐

Total Stress on the PJP Weld:


ultimate stress, σU = 9.75 ksi
weld stress DCR = 0.8182 OK

You might also like