I I I~IIIIIIIMCEER

EARTHQUAKE ENGINEERING TO EXTREME EVENTS

ISSN 1520-295X

Performance-Based Assessment and

Design of Squat Reinforced Concrete
Shear Walls

by
Cevdet Kerem Gulec and Andrew S. Whittaker

"
,.
;
s "

- 1 11)

>

,
'.

F_

F!t,,,...I..

: ..~.!. :
.:
:!.I... . :
s , ~ ; 1.1 l..:
. .
f

.. I

.t

i,

:".d
,\ " I tl
' I ' .. 1'"

-, :t"!-"
i
,' . '0 '. : "
I

II) ~

..

fl , rbo;lI

. : ;i 0.

,
." M ' l\'I.
'

I ~~

.,,
..

,~

""I- -.,_~.----!~~J
1J 15J1 lacan~n1lmml

Technical Report MCEER-09-001 0

September 15, 2009

This research was co nd ucted at th e Unive rsit y a t Buffalo, Sta te Univer sity of New York a nd was suppor ted prim aril y by th e
Ea rthquake Engineering Research Centers Program of the Nati on al Science Founda tion und er awa rd number HC 970147 1.

3 SQUAT WALL DATABASE

3.1 Introduction
Gulec (2005) review ed and catalogue d the results af tests of 352 reinforced concre te squat wa lls
with three different cros s-sec tion types, namely, rectangular, barbell, and flanged. Thi s database
was expanded to 434 walls. Of the 434 walls in the database, 150 have a rectangular cross
section and 284 have boundary elements [barbells (19 1), flanges (93)]. The assembled data can
be used to assess the performance of existing models that are used to predict squat wall response
and to develop new models. Detail ed inform ation for each wa ll in the database is tabul ated in
Appendix A. In thi s section, a brief summary of the assembled data is presented. Th e test
specimens in the database are selected usin g the followin g cr iteria : I) a minimum web thickness
of 5 ern. (1.97 in.); 2) symmetric reinforcement layout ; 3) no diagonal reinforcement or
additional wa ll-to-foun dat ion reinforcement to control sliding shear; and 4) aspect ratios (hw / I,..)
less than or equal to 2.0. Wa lls that do not comply with these criteria are not includ ed in the 434wa ll database and information on these wa lls is not present ed.
Figure 3-1 presents the variation of the expe rimentally measur ed pe ak shear strength (Vp eak ) with
moment-to-shear ratio' ( M I VI,..) for the wa lls in the database. In the figur e, the experimental
peak shear strength is norm alized by the product of A.. andR , where

A..

is the web area

I:

calculated as the pro duct of the wa ll length ( lw) and the web thickness ( 'w), and
is the
concrete compressive strength. The data presented in Figur e 3-1 shows that the experimental
peak shear strength for wa lls w ith boundary elements is generally much higher than those of the
wa lls w ith rectangular cross-sections. Figure 3-1 indicates that the ranges of measured pe ak shear
strengths for barbell and flanged wa lls are compa rable . The norm alized peak shear strengths for
rectangular wa lls are generally smaller than lO R
wherea s the majority of the norm ali zed peak
shear strengths for wa lls with boundary elements exceed 1O J1l , which is the upp er limit on the
peak shear strength equation of Section 21.9 of AC I 318-08 . As shown in Sections 1 and 2, the
peak shear strength calculation procedures for reinforced concrete walls generally do not
recognize the effects of boundary elements. For two wa lls, one with a rectangular cross-section
and the other with boundary elements, the current procedures predict identical peak shear
strengths provided that the wa lls have the same web area, reinforcement rati o, aspec t ratio, axial
force and Ie'. Neither wall-cro ss section type nor boundary element reinforcement are
considered as variables for calculating peak shear strength. These issues are addressed in the
follow ing sections .
Sections 3.2, 3.3 and 3.4 present summary information on rectangular, barbell and flanged wa lls
in the database, resp ectively. In these sections , mon otonic loading refers to incrementally
increasing the load in one direction until failure; repea ted loading refers to loading in one
directi on, unl oading to the origin, and reloading to a similar or larger displacement in the same
directi on; cyclic loading refers to applica tion of lateral force alternatively in both horizontal

Mome nt-to-she ar ratio is no rmalized by the wall length in this study .

35

25

,
,

20

"

~
~

-"

.0;

<

15

.
,

<,

>

[ 10

, ,,
,

, ,
,,

-..
It.'~ . ~ ,
"" ,t 1'
t" .:
. :. I ,.
,
! 4.

0.25

t" .

t , . - . I-. ..

," ' IV
, , ,
0.5

,,
,

Flanged

,"

,
,

,,,

, J

, ,, ""
,I,

,I

0.75

Rect angula r
Ba rbell

"
.
,,
,I ,' ,.,-t I'l , lo,

i t

,
,

,,

"

.,

1.25

L5

,,

,I

1.75

1\1 1(VIw )
Figure 3-1 Va riation of shear stress [norm aliz ed by the product of we b area ( A.,.) and J l ]
obtained usin g expe rime ntally dete r mined peak shear st r ength (Vp""k) with m oment-toshear ratio (M / Vlw )

directions with incre mented force or displacement following a conventional quasi-static cyclic
testing protocol; dynami c loading refers to the use of earthquake simulators ; and blast loading
refe rs to application of large amp litude dynamic pulses.

3.2 ' Valls with R ectangular Cross-Section

Th e rectangular-wa ll database includ ed experiments of 150 specimens at various scales . The data
for the 150 rectangular wall tests were obtained from Alexander et al. (1973), Hirosawa (19 75),
Cardenas et al. (198 0), Synge (19 80), Ma ier and Thnrlimann (19 85), Wiradinata (1985), Pilette
(1987), Huang and Sheu (198 8, 1994), Lefas and Kotsovos (1990), Lefas et a1. (1990),
Pilakoutas ( 199 1), Ro the (1992), Cheng (199 2), Cheng et a1. (1994), Mohammadi-Doostdar
(1994), Cheng and Yan g (19 96), Hidalgo et al. (1998), Salonikios et al. (1999), Xie and Xiao
(2000), Greifenhagen et al. (2005), Massone (2006), Sheu (2007), and Kuang and Ho (2008).
Figure 3-2 presents summary information on the 150 rectangular walls included in this database.
Wall aspect ratios (wall height divided by wall length) ranged between 0.25 and 2.0; moment-toshear ratios ranged betwee n 0.33 and 2.13; web thi ckn ess ranged fro1ll 2.36 to 6.30 in. (60 to 160
mm); wall length varied between 23.6 and 11 8.1 in. (600 and 3000 nun}; wall height varied
between 19.7 and 78 .7 in. (500 and 2000 nun}; 45 walls were tested with coexisting axial load 2

Self-weight ofthe walls are not included in the axial load calculations pre sented in this chapter.

36

100 r--

- - - -- - - - -----,

SO r - -- - - - -- - - -----,

Vl

GO

'-

o
~

~o

-'"
E
~

3
~
5
G
a) Web thickness (m)

M / (Vl)

11"/ 1"

20

10

15

20

25

30

b) Wall length I web th ickness

50
~

~ ~o

~ ~o

'-

'1.3

-'"

[) 30
-'"

E
20
~

E
20
~

10

0.25 0.5 0.75 1 1.25 1.5 1. 75 2 2.25

o
o

c) Aspect ratio

10

12

1~

d) Web reinforcement spacing (in)

horizontal

vert ical

100 I~------;::;~~===il
_
horizontal
_

yert ical

'- 30
o
~

.D

E
~
Z

20

to
()

0.50

i.~L
1.50

2.50

12

_.3

t) Number of web reinforcement layers

e) W eb remforcement ra110 (%)

Figure 3-2 Histograms of geom etr ic, m aterial, and loading p r op er ties of the 150 sq u at
rectangular walls

37

SO
~

Oil

100

60

~
~

horizonta l

,"ert ical

Oil

SO

40

..0

..0

40

Z 20

60

200n

-loon
g) (, (psi)

6000

20
0

sooo

40

~j~

50

60

r..
70

SO

Ion

90

110

h) Web reinforcement yield stress (ksi}

120

120

-;;

Oil

..0

..0

cyclic

dyna mic monotonic repeated

i) Loading type

10

j) P I(A

15

20

r') (%)

W<

Figu re 3-2 H istogra ms of geo me tric, m at erial, an d loading prope rties of th e 150 squat
rectangula r walls (co nt' d)

that ranged between 0.022 ~fc' an d 0 .182 ~f: ; reported concrete compressive strength] varied
bet ween 1991 and 7395 psi (13.7 and 5 1 MPa); horizontal web reinforcement ratios ranged
between 0.00 and 0.0161; and vertica l web reinforcement rati os ranged between 0.00 and 0.0287.
Boundary element reinforcement with rati os up to 0.128 wa s used in 110 of the 150 walls in
addition to the unifonnly distributed vertical web reinforcement. Fifteen of the 150 wa lls in the
dataset did not have horizontal we b reinforcement; 12 did not have vertical web reinforcement; 7
had neither horizont al nor vertical web reinforcement and included only boundary element
reinforcement at wa ll end s. The reported yield stress of the wall vertical web reinforcement
range d between 43.5 and 88.5 ksi (300 and 610 MPa), and that of the horizontal web
reinforcement ranged between 47.3 and 108.1 ksi (326 and 745 MPa). The rep orted yield stress
for the bound ary element reinforcement ranged between 43.5 and 84.8 ksi (300 and 585 MPa).
3 Some authors used cube strength rath er than cylinder strength to rep ort the compressive strength of concrete; cube
strengths were converted to cylinder strengths per M indess et al. (2003).

38

Vertical web reinforcement was provid ed in a single layer for 71 of the 150 walls, in two layers
for 65 walls and in thr ee layers for 2 walls. Horizontal web reinforcement was provid ed in a
single layer for 71 of the 150 walls, in two layers for 62 walls and in thr ee layers for 2 walls. The
maximum spacing for horizontal and vertical web reinforcement was 13.5 in. Note that Section
21.9 of ACI 3 18-08 [ACI (2008)] requires that web reinforcement in special structural walls be
provided in two layers and with a spac ing of less than 18 in.
Th e selected walls were tested using one of four types of loading: monotonic (quasi-static),
repea ted (quasi-static), cyclic (quasi-static), and dynam ic. Thirty of the 150 walls were tested
using monotoni c loading, 2 walls were tested using repeated loading, 110 walls were tested usin g
cyclic loadin g, and 3 walls were tested using dynamic loadin g.

3.3 Walls with Barbells

Th e data for the 191 squat barbell walls were obtained from (in chronological order), Benjamin
and Williams (1953 , 1954, 1956), Antebi et al. (196 0), Shiga et al. (19 73, 1975), Hirosawa
(1975), Ogata and Kabeyasawa (1984), Kabeyasawa and Somaki (1985), AIl ( 1985a, b, 1986a,
b, c), Ro the (1992 ), XiangDong (1999), Bouchon et al. (2004), and Sheu (2007). Figure 3-3
presents summary information associate d with other experimental parameters, in the form of
histograms, on the 191 squat barbell walls included in the database. Wall aspect ratios (wall
height divided by wall length) ranged between 0.28 and 1.60 an d moment-to-shear ratios ranged
between 0.06 and 1.90. The web thicknesses of the walls ranged from 1.97 to 6 .30 in. (50 to 160
nun}. The wall length varied between 20.0 and 155.9 in. (507 and 3960 mm) and the wall height
varied between 19.7 and 70 in. (500 and 1778 nun}. The ratios of the boun dary element area
(total barbell area, Abe ) to the total area of the wall ( A,) var ied between 0.22 and 0.60. Seventyfive walls were tested with coexisting axial load that ranged between 0.013 .'1, fc' and 0.321 Arle' ;
the axial forces on the remainin g walls were limited to the self-weight of the wall and upper
loading beam (or slab) . Reported concrete compressive strength varied from 1451 to 8463 psi
(10 to 58.3 MPa ); 59% of the walls had compressive strengths between 3000 an d 5000 psi (20.7
and 34.5 MPa ). Both horizontal and vertical web reinforcements ranged between 0.00 an d 0.028.
Boundary element reinforcement (reinforcement restricted to the barbells) was provid ed in all
191 walls; reinforcement ratios ranged between 0.71 and 8.27% of each boundary element area .
Ten of the 191 walls in the dataset had neither horizontal nor vertical web reinforcement and
included only boun dary element reinforcement. Th e reported yield stress of the wall vertical and
horizontal we b reinforcement range d between 39.3 and 90.5 ksi (271 and 624 MPa). The
reported yield stress for the boundary element reinforcement ranged between 37.8 and 81.9 ksi
(261 and 565 M Pa).
Th e selected walls were tested using one of five types of loadin g: monotonic (quasi-static),
repea ted (quasi-static), cyclic (quasi-static), blast, and dynamic. Fifty-seven of the 191 walls
were tested using monotonic loadin g, 8 walls were tested using repeated loading, 94 walls were
tested using cyclic loading, 2 walls were tested using dynamic loading and 30 walls were tested
using blast loadin g.

39

100 r-- -- --

- - - - - ----,

80

J
~
5
6
a) Web thickness (m)

10

20

30

-10

50

60

70

b) Wall length I web th ickness

M / tV1)

120

80

50

h" /I"

horizontal

vert ical

~ -10

'1.3

[) 30

""E 20
~

to
0.25 0.5 0.75

1.25 1.5 1.75

o
o

c) Aspect ratio

50

-I

10

12

1-1

16

d) Web reinforcement spa cing (in)

horizontal

vert ical

t 20 1_"_-~~;====il
100

horizontal

yert ical

~ ~o
'o
[) 30

""E
~

20

20

to

0.50

1.50

2.50

t) Number of web reinforcement layers

e) Web remforcement ra110 (%)

Figure 3-3 Histograms of geom etr ic, m aterial, and loading p r op er ties of the 191 sq u at
barbell walls

40

so
~

Oil

100 I-----;::;::;~~==ll
_
horizonta l

60

so

,"ert ical

.0

'"

E
~
Z 20

()

lOOO

1000

6000

800n

1nooo

40

50

60

70

SO

90

100

h) Web reinforcement yield stress (ksi}

g) (, (psi)
1 20 .-~--------------,

1"0

120

100
~

Oil 100
~

so

.0

60

E
~

"0
20

blast

cyclic dyna mic mono repealed

10

15

20

25

30

35

j) P I(A r') (%)

i) Loading type

"

Figu re 3-3 H istograms of geom etr ic, m at erial, an d loading prope rties of th e 19 1 squat
barbell walls (cont' d)

3.4 Walls with F langes

Th e data for the 93 squat flanged wa lls in the database were obtained from Barda (1972 ), Synge
(1980), Ma ier and Thn rlimann (1985 ), Saito et al. (1989), Sato et a1. (1989), Seki et al. (1995),
Mo and Chan (1996), Kitada et a1. (1997), Naze and Sidaner (200 1), Palermo and Vecchio
(2002a), Farvashany et al. (2008). Figure 3-4 presents summary information associated with
other experime ntal parameters, in the form of histograms, on these 93 squat flanged walls. For
these wa lls, aspec t ratios (wall height divided by wa ll length) ranged between 0.2 1 and 1.25 and
moment-to-shear ratios ranged between 0.20 and 1.36. The web thickness of the walls ranged
from 2.76 to 7.87 in. (70 to 200 mm). The wa ll length varied between 33 .9 and 122 in. (860 and
3100 nun) and the wall height varied between 15.8 and 103.1 in. (400 an d 2620 mm ). The ratios
of the boundary element area (total barbell area , Abe) to the total area of the wa ll ( .4,) varied
between 0.26 and 0.74. Fifty-six wa lls we re tested with coexisting axial load that ranged between

41

0.0 15 AJ ; and 0.273 ~fc' ; the axial forces on the remaining walls were limited to the selfweight of the wall and upper loading beam (or slab). Reported concrete compressive strength
varied from 2170 to 15084 psi (16.3 to 104 MPa); 69% of the walls had compre ssive strengths
between 400 0 and 6000 psi (27 .6 and 41.4 MPa). Horizontal web reinforcement ratios ranged
between 0 .0 and 1.69, and vertical web reinforce ment ratios ranged between 0.0 and 2.54.
Boundary element reinforcement (rein forcement restricted to the flanges) was provid ed in all 93
walls; reinforcement ratios ranged between 0.35 and 6.39% of each boundary element area . One
wall was tested without horizontal web reinforcement and one wall tested without vertical web
reinforcemen t. The reported yield stress of the vertical and horizontal web, and vertical bound ary
element reinforcement ranged between 42.9 and 87.7 ksi (296 and 605 MPa).
Th e selected walls were tested using one of three type s of loading: monotonic (quasi-static),
cyclic (quasi-static), and dynamic . Thirteen of the 93 walls were tested using monotonic loading,
78 walls were tested using cyclic loading, and 2 walls were tested usin g dynamic loading.

42

60 r-- -- -- - - - - - ----,

!OO r - -- - - - -- - - - ----,

50

~ ~O

SO

(, ()

'-+-

4-

~ JO

.0

.0

E
~
Z

E 20
~

10

-to
20
OL--

M / (Vl)

11"/ 1"

10

20

~O

JO

50

b) Wall length I web th ickness

a) Web thickness (m)

5()
~

~ ~o

~ -10

4-

'1.3

[) J()

.0

.0

E
20
~

E
20
~

10
O L-~

0.25 0.5 0.75

1.25 1.5 1. 75

c) Aspect ratio

.j

(,

10

12

d) Web reinforcement spacing (in )

horizontal

vcrt ical

SO ~~i==::::;----1
70

horizontal

vcn ical

'- J()

.D

E
~
Z

20

11-

to
OL

0.50

1.50

2.50

t) Number of web reinforcement layers

e) Web remforcement ra110 (%)

Figure 3-4 Hi stograms of geom etr ic, m aterial, and loading properties of the 93 squat
flanged walls

43

'"
~

Oil

30

Oil

~
~

horizonta l

,"ert ical

30

40

20

..0

20

..0

E
~
Z 10

E
~
Z 10

()

2(XlO

GOOO

1noon

uooo

40

~I
m

00

h) W eb reinforcement yield stress (ksi}

g) (, (psi)

80

-;

Oil

60

4-

40

..0

..0

E
~
Z

E
~
Z

cyclic

dynamic

15

monotonic

20

30

j) P I(A r') (%)

i) Loading type

"

Figur e 3-4 Histograms of geo me tric, material, and loading prope rties of the 93 squat
flan ged wall s (co nt'd)

44

8 DAMAGE STATES AND FRAGILITY CURVES

FOR SQUAT REINFORCED CONCRETE WALLS
8.1 Introduction
A genera l frame work for seism ic performance assessment and loss computations is provided in
th e 50% draft of the A Te-58 Guidelines f or Seismic P erf ormance Assessment ofBUildings [ATe
(2008)]. A key compo nent of th e fram ework is fragility fun ctions for compo nents and elements
of seismic framing sys tems. Fragility fun ction s relate the prob ability of exceeding one or more
damage thr esholds (described usin g damage states and repair mea sur es) to a demand p aramet er
such as story drift or comp onent plastic deformation. Thi s section summarizes the development
of fragility curve s for squat reinforce d concrete wa lls.
Dam age states are characterized typica lly using descriptors such as concrete crack width, ext ent
of concrete crushing , sliding shear displacement, and reinforcement yielding and bu cklin g.
However, to compute dollars losses and estimates of bui lding down time , damage states must be
charac terized by me asur es (or scopes) of repair and such descriptions are present ed herein , and
Appe ndices C, D an d E.

Information is used from experiments on squat walls rep orte d in the literature to generate
damage data as a functi on of an effic ient deman d param eter. Alternate prob abilit y distributions
are used to present th e damage data. Goo dne ss-of-fit tests are performed to evaluate the utility of
th ese distribution s.

8.2 Summary of Experimental Data

As presente d in Section 3, a significant number of squat rein forced concrete wa lls have been
tested in the past 60 years . Mo st of the experime ntal programs focused on the maximum strength
and initial (elastic) stiffness of such wa lls. The progression of damage with increasing lateral
displacement (drift) wa s of secondary interest. Informati on th at is key to the development of
damage states an d frag ility curve s such as m aximum crac k width, rein forcement yielding and
bucklin g, extent of concrete crushing, and ph otograph s illustratin g the condition of specim ens at
various stages of the expe riments wa s not always reported. According ly, only a fraction of th e
expe rimental data presented in Section 3 could be used to develop fragility function s.
Th e properties of the test specim ens used to develop fragility functi ons are summ arized in Table
8-1 through Table 8-6 . Specimens are grouped with respec t to wa ll geome try : rect angular,
barbell and flanged. In these tables, P is axial force; ~ is total wall area ; f e' is concrete
compressive strength ; h.... is wa ll height ; lw is wa ll length; t w is web thickn ess; h, and h, are
width an d depth of th e boundary element, respectively ; M / V is moment-to-shear ratio at th e
base of th e wall; Pix! is boundary element re inforce ment rati o, P v is vertical web reinforcement
ratio , and Ph is th e horizontal we b rein forcement rati o. The column labeled scale in Table 8-2,
Table 8-4, and Table 8-6 identifi es th e size of th e test specimen with respect to typi cal bui ldin g
construction by presenting the ratio of web panel thickn ess to 8 inches. The acro nym MoR in
th ese tables denote s Method of Repa ir, which is defined in Section 8.4 of thi s section .

199

Table 8-1 S u m mary of rec ta ngula r w all d at a used to create fragility functio ns

Xo

Reference

W an ID

Loading

PI A,!;

N u mber of damage
d at a u er l\loR

(%)

SW4

Cyclic

0 .0

SW5

Cyclic

0 .0

SW6

Cyclic

0 .0

SW7

Cyclic

0 .0

SW8

Cyclic

0 .0

SW9

Cyclic

0 .0

M1

Cycli c

2.2

M2

Cycli c

2.2

M3

Cycli c

9.5

10

M4

Monotonic

5.0

11

SW 11

Monotonic

0.0

12

SW 12

Monotonic

8.7

13

SW 13

Monotonic

18.1

14

SW 14

Monotonic

0 .0

15

SW 15

Monotonic

8.8

SW 16

Monotonic

18.0

SW 17

Monotonic

0 .0

18

SW2 1

Monotonic

0.0

19

SW22

Monotonic

9.1

20

SW23

Monotonic

18.2

21

SW24

Monotonic

0 .0

22

SW26

Monotonic

0 .0

S4

Monotonic

6.7

S9

Cyclic

7.5

Wall-I

Cycli c

0 .0

Wall-2

Cycli c

0.0

Wall-4

Cycli c

0 .0

Wall-5

Cycli c

0.0

Wall-7

Cyclic

0 .0

Wall-8

Cycli c

0.0

3
4

8
9

16
17

23
24
25
26
27
28
29
30

Pilakouras (199 1)

Greifenhagen et al. (2005)

Lefas et al. (1990)

Maier and Thurlimann ( 1985)

Wiradinata (1985)

Pilette (1987)

Mohammadi-Doosrdar (1994)

200

Table 8-1 Su mma ry of recta ngular w all data used to crea te fragility fun cti ons (cont 'd )

Xo

Reference

W an ID

Loading

PI A,!;

N umber o f damage
data u er l\loR

(%)

MSWI

Cyclic

0 .0

MSW3

Cyclic

7.0

MSW6

Cyclic

0 .0

LSW3

Cyclic

7.0

Wall-I

Cyclic

0 .0

36

wp l ll -9

Cyclic

10.0

37

wpll l-IO

Cyclic

10.0

wpll05-8

Cyclic

5.0

wpll05-7

Cyclic

5.0

40

wp llO-5

Cyclic

0 .0

41

wp llO-6

Monotonic

0 .0

W-IA

Cyclic

9.4

23

Cyclic

0 .0

27

Cyclic

0 .0

45

SW I I

Cyclic

0 .0

46

SW I2

Cyclic

0.0

47

SW I3

Cyclic

0 .0

SW I4

Cyclic

0 .0

49

SW I5

Cyclic

0 .0

50

SW I6

Cyclic

0 .0

51

SW I7

Monotonic

0.0

31
32
33

Salonikios et al. (1999)

34
35

38
39

42
43
44

48

Synge (1980)

Massone (2006)

Xie and Xiao (2000)

Hidalgo et al. (2002)

Lopes and Elnashai (199 1)

201

Table 8-2 Geo metric and material prope rties of the rectangu lar wall data
No

hw /

/ ....

M i V/.

h.
(i n)

I.
(in)

'.

(in)

t;

Sc ale

bi
(in)

P,
hi
Po. P.
(in) (%) (%) (%) (p si)

2.00

2.13

47.2 23 .6 2.36

0. 30

2.4

4.3

6.9

0.50 0. 39

5352

2.00

2.13

47.2 23 .6 2.36

0. 30

2.4

2.4

12.7 0.59 0. 31

4612

2.00

2.13

47.2 23 .6 2.36

0. 30

2.4

4.3

6.9

0.50 0. 31

5599

2.00

2.13

47.2 23 .6 2.36

0. 30

2.4

2.4

12.7 0.59 0. 39

4641

2.00

2.13

47.2 23 .6 2.36

0. 30

2.4

4.3

7.1

0.50 0 .28

6643

2.00

2.13

47.2 23 .6 2.36

0. 30

2.4

4.3

7.1

0.50 0.5 6

5642

0.61

0.69

24.0 39.4 3.94

0.49

0 .0

0.0

0.0

0.34 0. 37

7352

0.61

0.69

24.0 39.4 3.94

0.49

0 .0

0.0

0.0

0.34 0.00

7395

0.68

0.77

24.0 35.4 3.15

0. 39

0 .0

0.0

0.0

0.39 0 .26

29 15

10

0.68

0.77

24.0 35.4 3.15

0. 39

0 .0

0.0

0.0

0.39 0 .26

3538

11

1.00

LlO

29.5 29 .5 2.76

0. 34

3.9

5.1

8.5

2.14

Ll 7

7140

12

1.00

LlO

29.5 29 .5 2.76

0. 34

3.9

5.5

6.5

2.14

Ll 7

7332

13

1.00

LlO

29.5 29 .5 2.76

0. 34

2.8

5.5

3.1

2.14

Ll 7

5423

14

1.00

LlO

29.5 29 .5 2.76

0. 34

2.8

5.5

3.1

2.14

Ll 7

5642

15

1.00

LlO

29.5 29 .5 2.76

0. 34

2.8

5.5

3.1

2.14

Ll 7

58 19

16

1.00

LlO

29.5 29 .5 2.76

0. 34

2.8

5.5

3.1

2.14

Ll 7

7052

17

1.00

LlO

29.5 29 .5 2.76

0. 34

2.8

5.5

3.1

2.14 0. 37

6553

18

2.00

2.12

51.2 25 .6 2.56

0. 32

2.8

5.5

3.1

2.09 0.87

5745

19

2.00

2.12

51.2 25 .6 2.56

0. 32

2.8

5.5

3.1

2.09 0.87

689 1

20

2.00

2.12

5 1.2 25 .6 2.56

0. 32

2.6

5.5

3.3

2.09 0.87

6480

21

2.00

2.12

5 1.2 25 .6 2.56

0. 32

2.6

5.5

3.3

2.09 0.87

6553

22

2.00

2.12

51.2 25 .6 2.56

0. 32

2.6

5.5

3.3

2.09 0.40

388 1

23

1.02

Ll2

47.2 46.5 3.94

0.49

2.6

5.5

3.3

1.02

1.01

4772

24

1.02

Ll2

47.2 46.5 3.94

0.49

2.6

5.5

3.3

1.02 0.00

4235

25

0.50

0.58

39.4 78.7 3.94

0.49

0 .0

0.0

0.0

0.59 0 .26

3626

26

0.25

0.33

19.7 78.7 3.94

0.49

0 .0

0.0

0.0

0.59 0 .26

3191

27

0.50

0.58

39.4 78.7 3.94

0.49

3.9

12.6

1.3

0.59 0.80

4786

28

0.50

0.58

39.4 78.7 3.94

0.49

3.9

12.6

1.3

1.07 1.20

39 16

29

0.75

0.82

59.1 78.7 3.94

0.49

3.9

12.6

1.3

0.59 0.80

6527

30

1.00

1.09

59.1 59.1 3.94

0.49

3.9

9.8

1.6

0.51 0.80

6527

31

1.50

1.60

70.9 47.2 3.94

0.49

3.9

9.4

1.7

0.57 0.5 7

3785

32

1.50

1.60

70.9 47.2 3.94

0.49

3.9

9.4

1.3

0.28 0 .28

3495

202

Table 8-2 Geo metric and m aterial prope rties ofthe rectangu lar wall data (co nt'd)
No

hw /

/ ....

M i V/.

h.
(i n)

I.
(in)

'.

(in)

t;

Sc ale

bi
(in)

P,
hi
Po. P.
(in) (%) (%) (%) (p si)

33

1.50

1.60

70.9 47.2 3.94

0.49

3.9

9.4

1.7

0.57 0.5 7

3988

34

1.00

1.10

47.2 47.2 3.94

0.49

3.9

9.4

1.3

0.57 0.5 7

32 19

35

0.50

0.57

59.1 118.1 3.94

0.49

3.9

9.4

1.9

0.81

1.61

3945

36

0.89

0.44

48.0 54.0

6.00

0.75

6.0

7.5

0.9

0.25 0.27

4100

37

0.89

0.44

48.0 54.0

6.00

0.75

6.0

7.5

0.9

0.25 0.27

4550

38

0.89

0.44

48.0 54.0

6.00

0.75

6.0

7.5

0.9

0.25 0.27

4630

39

0.89

0.44

48.0 54.0

6.00

0.75

6.0

7.5

0.9

0.25 0.27

4640

40

0.89

0.44

48.0 54.0

6.00

0.75

6.0

7.5

0.9

0.25 0.27

4340

41

0.89

0.44

48.0 54.0

6.00

0.75

6.0

7.5

0.9

0.25 0.27

4500

42

0.50

0.59

48.0 96 .0 6.00

0.75

0 .0

0.0

0.0

0.37 0.31

4250

43

1.38

0.69

70.9 51.2 3.94

0.49

1.8

3.1

5.9

0.00 0.25

3655

44

1.00

0.50

55.1 55.1 3.94

0.49

1.8

3.1

5.9

0.00 0.25

3597

45

1.90

1.10

33.7

17.7 1.77

0.22

1.8

3.1

5.9

0.39 0.92

6230

46

1.90

1.10

33.7

17.7 1.77

0.22

1.8

3.1

5.9

0.39 0.92

642 1

47

1.90

1.10

33.7

17.7 1.77

0.22

1.8

3.1

5.9

0.39 0.92

7537

48

1.90

1.10

33.7

17.7 1.77

0.22

1.8

3.1

5.9

0.39 0.92

6289

49

1.90

1.10

33.7

17.7 1.77

0.22

1.8

3.1

5.9

0.39 0.51

6436

50

1.90

1.10

33.7 17.7 1.77

0.22

3.9

9.4

1.7

0.00 0.62

5995

51

1.90

1.10

33.7 17.7 1.77

0.22

3.9

9.4

1.3

0.00 0.80

6083

Min

0.25

0.33

19.7

17.7 1.77

0.22

0.0

0.0

0.0

0.00 0 .00

29 15

Max

2.00

2.13

70 .9 118.1 6.00

0 .75

6.0

14.2

12.7 2.14

Mean

1.27

1.15

43.3 4 1.7 3.41

0 .43

2.9

5.6

3.3

203

1.61

7537

0.83 0 .61

5249

Table 8-3 Summary of b arbell wall data used to create fragilit y functions
Number of damage
data per .M oR
1
2
3
4

'Vall ID

CW-0.6 -1.2-20

Cyclic

5.8

CW-0.6 -0.6-20

Cyclic

6.6

CW-0.6 -0.8-20

Cyclic

4 .9

CW-0.6 -1.6-20

Cyclic

5.8

CW-0.6 -2.0-20

Cyclic

5.7

CW-0.6 -1.2-40

Cyclic

12.3

CW-OA-1.2-20

Cyclic

5.9

CW-0.8 -1.2-20

Cyclic

5.9

CW-O.6-0.6-20a

Cyclic

6.7

CW-O.6-0.8-20a

Cyclic

6.6

11

CW-0.6- 1.2-0

Cyclic

0.0

12

CW-0.6-0-20

Cyclic

5.6

13

CW-0.6-0.3-20

Cyclic

5.6

14

CW-0.6-2 A-20

Cyclic

5.8

15

CW-0.6-2.8-20

Cyclic

6.2

16

CW-0.6-0-0

Cyclic

0 .0

CW-0.6-0-40

Cyclic

12A

18

CW-0.6-0.6-0

Cyclic

0.0

19

CW-0.6-0.6-40

Cyclic

11.5

20

CW-OA-0.6-20

Cyclic

5.8

21

CW-0 .8-0.6-20

Cyclic

5.8

22

CW-OA-2.0-20

Cyclic

5.8

23

CW-0.8-2.0-20

Cyclic

5.7

24

CW-0.6-2-0

Cyclic

0 .0

25

CW-0.6-2-40

Cyclic

11.6

CW-0.6-2-20B

Cyclic

5.5

27

CW-0.6-0.6-20L

Cyclic

7.8

28

CW-0.6-1.2-20L

Cyclic

7.6

29

CW-0.6-2-20L

Cyclic

7.8

30

No l

Cyclic

7.2

N02

Cyclic

5.1

N03

Cyclic

3.4

10

17

26

31
32

Referen ce

PIA,!;

Loading

No

AU (1985, )

AU (I 986c)

AU (1985b)

AU (1986b)

AU (1986, )

204

(% )

Table 8-4 Geome tric and m aterial prope rties of the barbell wall data
No

hw /

/ ....

M i vt;

h.
(in)

I.
(in)

'.
(in)

Sc ale

bi
(in)

t:

hi
P,
P..
P.
(in) (%) (%) (%) (psi)

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 1l .8

1.0

1.20

1.20

4935

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 1l .8

1.0

0.60 0.60

4281

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 1l .8

1.0

0.80 0.80

5760

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 11.8

1.4

1.60

1.60

4878

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 11.8

1.8

2.00 2.00

5021

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 11.8

1.0

1.20

1.20

4608

0.28

0.35

25.6 90.6 3.15

0.39

11.8 11.8

1.0

1.20

1.20

4793

0.63

0.70

57.1 90.6 3.15

0.39

11.8 11.8

1.0

1.20

1.20

4850

0.46

0.52

41.3 90.6 3.15

0.39

11.8 11.8

1.0

0.60 0.60

4224

10

0.46

0.52

41.3 90.6 3.15

0.39

11.8 11.8

1.0

0.80 0.80

4295

II

0.46

0.52

41.3 90.6 3.15

0.39

11.8 11.8

1.0

1.20

1.20

4 167

12

0.46

0.52

41.3 90.6 3.15

0.39

11.8 11.8

1.0

0.00 0.00

5106

13

0.46

0.52

41.3 90.6 3.15

0.39

11.8 11.8

1.0

0.30 0.30

5106

14

0.46

0.52

41.3 90.6 3.15

0.39

11.8 11.8

1.8

2.40 2.40

4878

15

0.46

0.52

41.3 90.6 3.15

0.39

11.8 11.8

1.8

2.80 2.80

4594

16

0.46

0.52

41.3 90.6 3.15

0.39

11.8

11.8

1.0

0.00 0.00

4594

17

0.46

0.52

41.3 90.6 3.15

0.39

11.8

11.8

1.0

0.00 0.00

4594

18

0.46

0.52

41.3 90.6 3.15

0.39

11.8

11.8

1.0

0.60 0.60

5106

19

0.46

0.52

41.3 90.6 3.15

0.39

11.8

11.8

1.0

0.60 0.60

4935

20

0.28

0.35

25.6 90.6 3.15

0.39

11.8

11.8

1.0

0.60 0.60

4878

21

0.63

0.70

57.1 90.6 3.15

0.39

11.8

11.8

1.0

0.60 0.60

4864

22

0.28

0.35

25.6 90.6 3.15

0.39

11.8

11.8

1.8

2.00 2.00

4935

23

0.63

0.70

57.1 90.6 3.15

0.39

11.8

11.8

1.8

2.00 2.00

4978

24

0.46

0.52

41.3 90.6 3.15

0.39

11.8

11.8

1.8

2.00 2.00

4893

25

0.46

0.52

41.3 90.6 3.15

0.39

11.8

11.8

1.8

2.00 2.00

4921

26

0.46

0.52

41.3 90.6 3.15

0.39

11.8

11.8

1.0

2.00 2.00

5149

27

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 11.8

1.0

0.60 0.60

3655

28

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 11.8

1.0

1.20 1.20

3755

29

0.46

0.52

41.3 90 .6 3.15

0.39

11.8 11.8

1.8

2.00 2.00

3655

30

0.48

0.55

41.3

85.8 3.15

0.39

7.1

7.1

2.9

1.20 1.20

3940

31

0.48

0.55

41.3

85.8 3.15

0.39

7.1

7.1

2.9

1.20 1.20

5561

32

0.48

0.55

41.3

85.8 3.15

0.39

7.1

7.1

2.9

1.20 1.20

8463

Min

0.28

0.35

25.6

85.8 3.15

0.39

7.1

7. 1

1.0

0.00 0.00

3655

Max

0.63

0.70

57.1 90.6 3.15

0.39

11.8

11.8

2.9

2.80 2.80

8463

Mean 0.46

0.52

41.3

0.39

11.4

11.4

1.4

1.18

4824

90.1 3.15

205

1.18

T able 8- 5 S u m mary of flanged w all data used to create fr agility fun ct ions
~o

Reference

WalllD Loading

P I ,1,/;

Nu mber of damage
d at a ue r l\I oR

(%)

Bl -l

Monotonic

0.0

B2- 1

Monotonic

0.0

B3-2

Cyclic

0.0

B4-3

Cyclic

0.0

B5-4

Cyclic

0.0

B6-4

Cyclic

0.0

B7-5

Cyclic

0.0

B8-5

Cyclic

0.0

SI

Monotonic

6.6

10

S2

Monotonic

24.2

S3

Monotonic

6.5

S5

Cyclic

6.3

13

S6

Monotonic

6.6

14

S7

Cyclic

27.3

4
5

11

12

Bania (1972)

Maier and Thiirlimann (1985)

15

Synge ( 1980)

Wall-3

Cyclic

0.0

16

Kitada et al. (1997)

V-I

ES

3.9

DPI

Cyclic

5.4

DP2

Cyclic

0.0

W 12- 1

Cyclic

5.6

20

24M 8-30

Cyclic

5.3

21

24M 8-40

Cyclic

5.6

22

24M 8-50

Cyclic

5.7

23

36M 8-30

Cyclic

5.1

36M 8-40

Cyclic

5.2

25

36M 8-50

Cyclic

5.3

26

48M 8-30

Cyclic

7.3

27

48M 8-40

Cyclic

7.3

28

48M 8-50

Cyclic

7.1

17
18
19

24

Palermo and Vecc hio (2002a)

Saito et al. (1989 )

Sato et al. (1989)

20 6

Table 8-6 Geometric and material properties of the flanged wall dat a
No

hw /

/ ....

M i vt;

h.
(in)

I.
(in)

'.
(in)

t:

Sc ale

bi
(in)

hi
P,
P..
P.
(in) (%) (%) (%) (psi)

0.46

0.50

34.5 75.0 4.00

0.50

24.0

4.0

1.8

0.50 0.4 8

4200

0.46

0.50

34.5 75.0 4.00

0.50

24.0

4.0

6.4

0.50 0.4 8

2370

0.46

0.50

34.5 75.0 4.00

0.50

24.0

4.0

4.1

0.50 0.4 8

3920

0.46

0.50

34.5 75.0 4.00

0.50

24.0

4.0

4.1

0.50 0.00

2760

0.46

0.50

34.5 75.0 4.00

0.50

24.0

4.0

4.1

0.00 0.4 8

4 190

0.46

0.50

34.5 75.0 4.00

0.50

24.0

4.0

4.1

0.26 0.4 8

3080

0.21

0.25

15.8 75.0 4.00

0.50

24.0

4.0

4.1

0.50 0.49

3730

0.96

1.00

72.0 75.0 4.00

0.50

24.0

4.0

4.1

0.50 0.50

3400

1.02

1.12

47.2 46.5 3.94

0.49

15.7

3.9

1.1

1.13

1.01

5352

10

1.02

1.12

47.2 46.5 3.94

0.49

15.7

3.9

1.1

1.13

1.01

5568

11

1.02

1.12

47.2 46.5 3.94

0.49

15.7

3.9

2.5

2.54 1.01

5323

12

1.02

1.12

47.2 46.5 3.94

0.49

15.7

3.9

1.1

1.13 1.01

54 10

13

1.02

1.12

47.2 46.5 3.94

0.49

15.7

3.9

1.1

1.13 0.5 7

5163

14

1.02

1.12

47.2 46.5 3.94

0.49

15.7

3.9

1.1

1.13

1.01

4946

15

0.50

0.57

59.1 118.1 3.94

0.49

19.7

3.9

1.8

0.37 1.61

377 1

16

0.65

0.77

79.5 122.0 2.95

0.37

117.3 3.9

0.5

1.20

1.20

4153

17

0.66

0.76

79.5 121.1 2.95

0.37

119.9 3.7

0.4

0.82 0.76

3147

18

0.65

0.76

79.5 121.5 2.95

0.37

119.9 3.9

0.4

0.82 0.76

2727

19

0.35

0.4 7

29.5 83.5 4.72

0.59

19.7

4.7

4.8

1.32

1.32

5106

20

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

0.8

0.80 0.80

55 12

21

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

0.6

0.60 0.60

5192

22

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

0.5

0.4 8 0.4 8

5076

23

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

1.2

1.16

1.16

5700

24

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

0.9

0.90 0.90

5628

25

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

0.7

0.72 0.72

5439

26

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

1.6

1.60

1.60

3974

27

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

1.2

1.16

1.16

3989

28

0.65

0.74

55.1 84.6 5.91

0.74

39.4

5.9

1.0

0.96 0.9 6

4061

Min

0.21

0.25

15.8 46.5 2.95

0.37

15.7

3.7

0.4

0.00 0.00

2370

Max

1.02

1.12

79.5 122.0 5.91

0.74

119.9 5.9

6.4

2.54

1.61

5700

Mean 0.67

0.75

50. 1 78.8 4.51

0 .56

37 .0

2.0

0.87 0 .82

4389

207

4.6

Experimental data from 111 squat walls that met the above criteria provided the damage data
presented below. Twenty-eight walls had flanges, 32 walls had barbells, and 51 walls had
rectang ular cross-sections. Th e web thickn ess range d between 1.77 in. and 6 in. The aspec t ratios
(hw / I,..) range d between 0.21 and 2.0; moment-to-shear ratios (M / Vlw ) ranged between 0.25
and 2.13. The maximum aspec t ratio in the barbell and flanged wall datasets are 0.63 and 1.02,
respectively. Sixty-five walls were tested with applied axial load ranging between 0.022 A,/:

f:

and 0.273 ~ fc' , where AI is total wall area and

is concre te compressive strength. Concrete
compressive strength , based on standard cylindrical tests, varied between 237 0 and 8463 psi.
Eight walls did not have vertical web reinforcement , 6 walls did not have horizontal web
reinforcement and 3 walls did not have any web reinforcement. Eleven of the III test specimens
did not comply with the minimum reinforcement requirements ( Ph = P v =0. 25%) of ACI 318-08
[ACI (2008)] for Special Structural Walls and Concrete Beams (Chapter 21.9). The maximum
observed horizontal and vertical reinforcement ratios were 2.80%. Eighty-nine of the III test
specimens were tested using cyclic loading, 21 were tested using monotonic loadin g and 1 was
tested using an earthquake simulator. The scales in the dataset ranged between 0 .22 and 0.75.

8.3 Demand Parameter Selection

Th e experimental data describ ed in Section 8.2 was mined to prepa re fragility curve s. The
fragility curves are presented as a function of racking drift, which is the best single story-level
demand parameter for most structural elements. Imp ortantly, drift is the demand parameter most
reported in the literatur e.
Two other demand parameter s were considered for this study but were set aside. Dissipated
hysteretic energy [Park and Ang (1985 )] has been used to assess damage to concrete
components. However, the literature rarely includes hysteretic loops that are amenable to
digiti zation and calculation of dissipated energy and so this demand parameter cannot be
computed with high confidence . Pagni and Lowes (20 06) and Brown and Lowes (2007)
proposed a variant on dissipated hysteretic energy for assessment of reinforced concrete beamcolumn j oints, namely, a demand paramet er whose functional form included maximum story
drift and number ofload (or displacement) cycles as variables. Unfortunately, many sources did
not report the relationship between damage an d number of loading cycles and so the Lowes
demand paramet er was not pursued.

8.4 Damage States and Methods of Repair

Damage states (DS) define threshold levels of damage sustained by structural components und er
earthquake loadin g. A famil y of damage states, which are listed in Table 8-7, was assembled
following analysis of test data and review of the literatur e. Damage states are characterized
generally by direct indicators of damage such as initiation of cracking, maximum concrete crack
width, extent of concrete crushing, sliding shear displacement, and reinforcement yielding,
bucklin g, and fractur e. Each of these damage states is linked with one of four methods of repair
in the table.

208

Table 8-7 Damage sta tes and correspo nding methods of repairs
ID

Damage States

Method of Repair (\loR)

OS1.1

Initi ati on of cracking

OS1.2

Initi ati on of flexural crac king

OS1.3

Initi ati on of shear crac king

OS1.4

Maximum m easured crac k widths less than

0.02 in. (0 .5 mm)

OS2.1

Initi ati on of yielding in horizontal we b

re inforce me nt

OS2.2

Initi ati on of yielding in vertical web

re inforce ment

OS2.3

Initi ati on of yielding in vertical boundary

element reinforcement

Cosme tic repair (Mo R- l )

Maximum m easur ed shear crac k widths larger

OS2.4a than 0.02 in (0 .5 1mn) but less than 0 .12 in. (3
mm)

Epoxy injecti on (MoR-2)

Maximum m easured flexural crack widths

OS2.5a larger than 0 .02 in (0.5 mm) but less than 0.12
in. (3 mm)
Maximum me asur ed shear crac k widths larger
O S2.4b th an 0.04 in (1.0 mm) but less than 0 .12 in. (3
mm)
Maximum m easured flexural crack widths
O S2.5b larger than 0 .04 in (1.0 mm) but less than 0.12
in. (3 mm)
OS3.1

Concrete crushing at the compression toes I

initiation of crushing in th e wa ll web

OS3.2

Vertical crac king in th e toe regions of the we b

Buckling of boundary element vertical

OS3.3
re inforce ment
Flexural crac k widths exceeding 0.12 in. (3
OS3.4
mm)
OS4.1

Initi ati on of sliding

OS4.2

Wide diagon al cracks

OS4.3

Widespread crushing of concrete

OS4.4

Reinforcement fracture

OS4.5

Shea r crack widths exceeding 0 .12 in (3 mm)

Partial wa ll replacement
(MoR-3 )

Wa ll replacement (M oR-4)

209

Docum ents that provid e guidelines for repair of reinforce d concre te walls (e.g., FEMA 306
[ATC (l998b)] and FEMA 308 [ATC (l998a)]), repair of concre te (e.g. , ACI 546R-04 [ACI
(2004)]), observations from experimental programs, previous research on retrofit of squat walls
and expert opinion [Hooper (200 8)] were used to identify the most appropriate damage states and
their correspo nding methods of repair. The followin g subsections pre sent information on each
Me thod of Repa ir (MoR) and the corre sponding damage states. The number of damage data
obtained for each wall and method of repair is presented in Table 8-1, Table 8-3, and Table 8-5,
for rectangul ar, barbell and flanged walls, respectively. Appendix C presents all of the damage
data ana lyzed in the body of the report.

8.4.1 MoR-I, Cosmetic Repair

Damage states OS1.1 through OS1.4 of Table 8-7 are associated with cosmetic repair. These
damage states represent the initiation of cracking in concrete and propaga tion of these cracks
und er earthquake loading. For cosmetic repair, crack widths are small (less than 0.02 in. as
defin ed in OS1.4) and structural repair to restore strength and stiffn ess is unn ecessary. Repa ir of
the surface finishes may be required to restore the aesthetic appearance, maintain fire resistance
and prevent water infiltration into the wall [ATC (1998a)].
Cosmetic repair has no impact on structural performan ce. Method of Repair MoR-1 is similar to
Cosmetic Repair I (CR I) of FEMA 308. Examples of damage states 1.2 and 1.3, each of which
is linked to MoR- I, are presented in Figur e 8-1 and Figure 8-2.

8.4.2 MoR-2, Epoxy Injection

8.4.2.1 Introduction
Damage states OS2.1, OS2.2, OS2.3, OS2.4a, OS2.5a, OS2.4b , and OS2.5b of Table 8-7 are
associated with epoxy-injection repair. Epoxy injection is widely used to restore the stiffness and
strength of cracked concrete components. Crack width (OS2 .4a, OS2.5a, OS2.4b , and OS2.5b)
and reinforcement yielding (OS2.1, OS2.2, and OS2.3) are considered two indi cators of the need
for epoxy injection repair.
Method of Repa ir MoR-2 is sim ilar to Structura l Repa ir I of FEMA 308. Exa mples of damage
states OS2.1, OS2.3, and OS2.4a, each of which is link ed to MoR-2, are presented in Figur e 8-3,
Figure 8-4 and Figur e 8-5, respec tively.
Th ere is no consensus in the design professional community on the minimum residual crack
width for epoxy injection. Herein, two options are presente d, MoR-2a and MoR-2 b.

8.4.2.2 :\IoR-2a
Lowes and Li (2008) note that residual crack widths in excess of 0 .01 to 0.02 in. require epox y
injection to restore component strength and stiffn ess . However, residual crack widths are not
reported in the literature and could therefore not be used to identi fy the need for injection
grouting . Rather, max imum crack width, which is presented in the literature, is used as a
surrogate for residual crack width: if the reported maximum crack width under loadin g exceeds
0.02 in (0.5 nun}, (assum ed equivalent to OS2.4a and OS2.5a in Table 8-7) epoxy injection is

210

Figure 8-1 Initial flexural cracks on wall M4 tested by Greifenhagen et al. (2005)

......

'~

". ....

...

..
~ .. .

-,

.'.:

Figure 8-2 Initial shear cracks on wall 52 tested by Maier and Thiirlimann (1985)

2 11

Figure 8-3 Cracking pattern for wall 53 tested by Maier and Thiirlimann (1985) at first
yielding of horizontal web reinforcement

Fi gure 8-4 Cr acking pattern for wall 52 tested by Maier and Thiirlimann (1985) at first
yielding of flange vertical rein forcement

212

Figure 8-5 Cracking pattern for wa ll DP I tested by Palermo and Vecchio (2002a) at a
maximum shear crack width of 0.5 mm

required. Reinforcement yielding (DS2.l , DS2.2 , and DS2.3 ) was used as another indicator for
epoxy injection, on the basis that reinforcement yielding will result in residual cracks of
substantial width. This relationship is studied in Section 8.6.

8.4.2.3 NI oR -2b
Method of Repair MoR-2b involves injection grouting of cracks for which the maximum crack
width under loading exceeds 0.04 in (1.0 mm), that is DS2.4b and DS2.5b in Table 8-7.

8.4.3 M oR-3, P arti al Wall R eplacement

Damage states DS3 .1 through DS3.4 of Table 8-7 are associated with removal and replacement
of damaged concrete. Localized crushing at boundary elements and in the wall web will require
replacement of damaged concrete. Vertical splitting cracks in boundary elements (DS3.2) and
buckling of boundary element vertic al reinforcement (DS3.3) are also linked to this MoR.
Photographs of test specimens and observations made by the cited authors were used to identify
data for damage states DS3.l , DS3.2 , and DS3 .3. Damage state DS3.4 assume s that a wall with
flexural crack widths greater than 0.12 in. (3 mm) cannot be repaired to its pre-earthquake
condition using epoxy injection and that damaged concrete and rebar must be replaced.
Examples of damage states DS3.1 and DS3.4 are presented in Figure 8-6 and Figure 8-7,
respecti vely.

213

Figure 8-6 Cracking pattern for wall SWll tested by Lefas et al. (1990) at compression
zone failure

Crack width

= 3.0 mm

Figure 8-7 Cracking pattern for wall M2 tested by Greifenhagen et al. (2005) at a
maximum flexural crack width of 3.0 mm

2 14

8.4.4 MoR-4, Wall Replacement

Damage states DS4.1 through D54 .5 of Table 8-7 are associated with complete replacement of
the wall panel.
Sliding at the interface between the wall web and the foundation (D5 4.1) will requir e wall panel
rep lacement, especially if the residual displacement is significant. Squat walls with low vertical
reinforcement ratio an d low axial load are susceptible to sliding shear. A sample cracking pattern
for damage state DS4. 1 is presented in Figure 8-8. Substantial com er-to-comer diagonal cracks
(OS 4.2) which are indicators of diago nal tension (DT) or flexur e-diagonal tension (F-DT)
failures [see Section 1.2], trigger a rapid loss of strength and stiffn ess . Such walls cann ot be
repa ired using epoxy injection because the crack widths are too great and parti al wall
rep lacement is not feasible because of the orientation and length of the cracks. Squat walls with
light horizontal reinforcement are susceptible to this type of damage. Figure 8-9 pre sents a
cracking pattern for damag e state DS4.2.
Walls with heavy reinforcement and/or high axial loads usually fail in diagonal compre ssion
(crushing of compression struts) as shown in Figure 8-10 . The damage is so widespread that
partial wall replacement is not feasible and the entire wall panel mu st be replaced (DS4.3).
Re inforcement fractur e (OS4.4) is rarely seen in tests of squat walls. Reinforce ment fractur ed in
only 3 of the 111 walls. Damage state DS4.4 is linked to wall rep lacement beca use reinforcement
fracture followed other significant damage in all three cases. Figure 8-11 shows the cracking
pattern in a wall at the fracture of boundary element vertical reinforcement. Widespread damage
at the interface betwee n the wall web and foundation is evident in the figur e. Damage State
DS4.5 assumes that wall panels with maximum shear crack widths larger than 0.12 in (3 mm)
must be replaced. FEMA 306 [ATC (l998b)] notes that when the width of shear cracks in a wall
exceed 1/8 in. (3.2 nun}, the damage is considered heavy and restoration of the pre-earthquake
strength and stiffn ess requires rep lacement of the wall. Thi s damage state is similar to damage
state DS4.2 in the sense that it is also closely related to diago nal tension (DT) and flexur ediagonal tension (F-DT) failures observed in squat walls.
Th e available images of damage together with the author' s descr iptions of damage were used
whereve r possible to identify the drift associated with wall rep lacement. A major challenge was
the identifi cation of drift corresponding to sliding shear failure (DS4. 1). The author-reported
drifts for slidin g shear failures exhibited significant variance, for exa mple, in some cases the
authors reported the drifts at initiation of sliding whereas in others the author-reported drift s
corre sponding to significant slidin g displacements. Two supplemental criteria are therefore used
herein to aid in the identifi cation of drift for DS4.1: 1) residual drift exceeding 0.5% (SCI), and
2) drift associated with wall strength less than 50% of the peak strength (SC z). Each criterion is
describ ed in more detail below. For a wall reported to have failed by sliding shear, the drift is
calculated using SC t. If a drift cannot be computed using SCI, SC z is used. If a drift cannot be
computed using SC z, no data is reported for DS4.1.

Sup pleme ntal Criterion SCf If a cyclic force-d isplacement relationship was available, the peak
transient drift associated with a residual drift angle of 0.5% was estimated. The residual drift
angle of 0.5% was associated with a zero-force displacement intercept of 1% story drift angle.
Given that the coda portion of earth quake groun d motions will generally serve to partially recenter displaced compo nents, elements and building frames, we assumed that a residual drift

215

/
I

/~

~e AR WALL DP2
FINAL
DISP ;~ 1 0

Figure 8-8 Cr acking pattern for wall DP2 tested by Palermo and Vecchio (20 02a) at a
slidin g failure between the wall web and top slab

..

~. '
~ .

.....
. . ..
.

'-

r
.,.

"
,

'

.
.

., ..
- . -,...
.'. .
,.;

... . .,.
. ....
... .-. -....

,.

-...

....

-,

...

"

,-

.'

.. .
,

1 .. ,

"

, - - -

' , '

. ,

"-

.. . .

Figure 8-9 C r acking pattern for wall 59 te sted by Maier and Thiirlimann (1985) at
diagonal tension failure

216

Figure 8-10 Cracking pattern for wall 57 tested by Maier and Thiirlimann (1985) at
crushing of diagonal compression struts

Figure 8-11 Cracking pattern for wall M1 tested by Greifenhagen et al. (2005) at
reinforcement fracture

2 17

angle of 1% from a displacement-controlled cyclic test would be reduced to 0.5% in an

earthquake . Only first cycle curves at each displacement level were considered for computation
of the drifts at which the drifts at zero-load exceed 1.0%.To illustrate this computation , con sider
Figure 8-12 that pre sents the load-di splacement data for wall Ml reported in Greifenhagen et a1.
(2005). In the figure, the peak transient drifts in the first (1.47%) and third quadrants (2.70%) for
which the residual drifts exceed 1.0% are identified using red circles. The smaller of the two
drifts is assumed to mark the onset of sliding shear failure and used for further analysis.
2:50
- 0.0

20 0

l SS4

1:50

Drift = -1.0%

I
l S58

100

V>

50

" , =25

", =20

<:I

- 50

- 100

Drift = + 1.0%

- 150

- 0.8 F
51

- 200

_
lS~

- 25
North

- 20
~

Sout h

- 15

-1 0

-f

l S 37

-5

Specimen M1

l ateral Displacement [ml\l]

10

15

20
25
North _ South

Figure 8-12 Residual drift computation on the force-drift relationship for M1 tested by
Greifenhagen et al. (2005)

Supplemental Criterion SC:f A backb one load-displacement rela tionship for the first and third
quadr ants is prepared using the first cycle shear strengths at each displacement increment. The
peak resistance of the wall is compu ted as the maximum force in the first and third quadrants.
The story drifts in the first and third quadrants at which the resistance dropped below 50% of the
peak resistance are calculated and the smaller of the two story drifts is chosen for further
analysis. The computation is illustrated in Figure 8-13 in which backb one force-di splacement
curve s are presented in the first and third quadrants. The peak shear strengths of the wall (in the
first quadrant in this case) and the drift s at a post-capping shear strength equal to 50% of the
peak resistance in the first quadrant and third quadrants, equal to 3.84% and -3.4 7%,
respectively, are identified in the figure .
The supplemental criteri a are used primarily to identify transient drifts associated with sliding
shear failure s but are also used if a failure by either diagonal tension or compression is reported
but the drift at failure is not, the drift associated with SC2 is reported,

218

zso

-o.nz,

- 0 03h,

-o.on-,

o 02h,

00311,

200

0.8 F
150

100

50

co
~

~1~=2 5

!,l

11, =25

p, =20

- 50

-1 0.<1-:--

----'

- 0.8 F
-f

l S46

- 25

- 20

- 15

North c- South

-1 0

-5

Specimen M1

Lateral DospIa ce", ent [mOl]

10

15

2D
North

--4

25
Scu m

Figure 8-13 Det ermination of drifts at which the for ce-peak shear strength drops to 50% of
th e peak for M1 tes ted by G reifenhagen et al. (2005)

8.5 Fragility Analysis

8.5.1 Intr oduction
Herein, fragility functions are developed to characterize the probability that a specific MoR will
be required as a function of racking (shear) drift. The data that links drift to damage states, and
thus repair, is pre sented using continuous prob ability distributions. The method of maximum
likelihood is used to fit the data . Four continuous probability distributions are con sidered,
namely: lognormal , gamma, Weibull, and beta. Empirical cumulative distribution functions are
pre sented below together with the fitted distributions.

8.5.2 Characterization of Dam age Data

Careful mining of damage data is essential because data for more than one damage state was
identified for a given method of repair in many of the walls.
Two methods were considered to capture data for the purpose of developing fragility curves:
Meth od 1: Data for all damage states is used , resulting in multiple data points per
specimen for a given MoR.
Meth od 2: Each MoR is represented by a single data point that corre sponds to the
damage state with the smallest drift value for a given specimen.

2 19

Figure 8-14 illustrates the application of the two methods. In this figur e, MoR-N is any method
of repair, and DSN.l thr ough DSN.5 are the available damage data for MoR-N.

MoR-N

DSN .1

DSN.2

DSN.3

DSN .5

DSNA

Method (1)

Data from
all damage
states.

Method (2)

Min (DSN.1, DSN .2, DSN .3, DSN A , DSN.5)

Figure 8-14 Sch ematic representation of data characterization

Using all available demand para meter-damage state pairs to create fragility curve s will yield the
greatest numb er of data points for each MoR (a maximum of 5 pairs pe r Figure 8-14). However ,
in most cases, the use of Me thod I will result in a higher mean value for a fragility fun ction than
that of Me thod 2 (one value per Figur e 8-14). Fragility curves were developed using both
methods to investigate the impact of the choice of data representation. Mea ns and standard
deviations for the drift data obtained usin g the two methods are pr esented in
Table 8-8 for each MoR; data are presented for both MoR-2a and MoR-2b. As seen in the table,
the means calculated usin g Method I are higher than those calculated usin g Method 2.
Variability, as measured by coefficient of variation, is higher for Method 1 excep t for MoR-3.
Th e impact of the choice of method is more significant for MoR- l and MoR-2a because, in many
cases, two or more values of drift could be associated with these methods of repair for a given
wall. Inmost instances, a single drift could be identified for each wall for MoR-2b, MoR-3, and
MoR-4.

8.5.3 Probability Distributions

Four probability distributions that are widely used for engineering applications, namely,
lognormal, gamma, Weibull, and beta, were used to fit the compiled damage data. All four
distributions requir e positive values of the random variable (demand para meter). Figure 8-15
presents sample probab ility density func tions for the four distributions. The followin g
subsections present summary information on these probabili ty distributions. Th e variable x is
used to represent the random variable (= story drift).

220

Table 8-8 Statistical summary of the drifts (a ll cross-sections) obtained using two methods
for data mining for each ~loR
Method 2

Method 1

MoR
Mean

Standard Coeff. of
deviation Variation

Mean

Standard Coeff. of
deviation Variation

0.12

0 .14

1.20

0.07

0.09

1.20

2a

0.50

0.28

0.55

0.42

0.18

0.44

2b

0.62

0.22

0.36

0.62

0.23

0.37

0.84

0.39

0.46

0.83

0.41

0.5 0

1.23

0 .51

0.42

1.19

0.47

0.39

8.5.3.1 Lognormal Distribution

Th e lognormal distribution is a one-sided probability distribution of a random variable whose
logarithm is normally distributed. This distribution is widely used for fragility studies because
the demand parameter (drift or acce leration) must be positive and its relationship with the normal
or Gaussian distribution. Equation 8-1 presents the probabili ty density fun ction (pdf) for the
lognormal distribution.

1
tx (x) =

exp[_(lll(X)-,Po,)' ]

for r z O
(8-1)

20-0 ,

X<Yo,.;z;;

0, elsewhere

In Equation 8-1,

P ln x

and

O"ln x

are the mean and standard deviation of the natur al logs of the

demand parameter. The standard deviation of the natural log of the data, O"ln x, is term ed
dispersion in the ATC-5 8 project [ATC (2008)] and is denoted as P. The median ( 8), mean ( p)
and standard deviation (0" ) for a lognormally distributed demand parameter (x) are presented in
Equations 8-2 through 8-4.
(8-2)

(8-3)

(8-4)

22 1

41

211

- - - P = 0.125
- - - P = O.25

3 &... . . . .... . . . . ...:

, 1 - - -

p = 0.5
p=1

1.51-1:..... .:

- - - P = 1.5
--- p = 3

:s

,-""

2 H :

I I

- - - k = 3 1, = 5
- - - k = 3 1, = 3
k = 3 1,= 1
- - - k = l l, = l
- - - k =0.5 1, = 1

"'"'
,.::.,
,-""

hnx=1
0 .5

0.5

1.5

2.5

a) lognormal distribution
N
N
N

21

b) gamma distribution
I

--- k=
--- k=
k=
:.: 1 - - - k =

41

0.5 1, = 2
1 1, = 2
2 1,= 2
2 1, = 3

3.5 ~

J I

I;,

u = 0.5 p = 2
--- u = l p = l
- - - ex= 6 1l= 2
- - - ex = 2 1l= 6

- - - k=2 1,= 4

2.5

6"'"'
,-""

- - - ex = 0.5 p = 0.5
I - - - ex =2 P=0.5

"'"'
,.::.,
,-""

0'
4

2
X

c) Weibull distribution

====

---

0.25

0.5

===0.75

d) beta distribution

Figu re 8-15 Fa milies of probability den sity functions for th e four distributions utili zed herein

8.5.3.2 Gamma Distribu tion

Sim ilar to the lognormal distribution, the gamma distribution is also one-sided. Equ ation 8-5
presents the probability density fun ction for the gamma distribution. The gamma distribution
uses two parameters, k and A . The parameter k defin es the shape of the distribution and A is a
scale paramet er. The probability density functi on for the gannn a distribution is unimodal with its
peak at x = Ofor k :<: :; l ,andat x = (k - l) /A for k > l [Soong (2004)].

Ix (x)

~xk-le-).x:

for x;:::': 0

(8-5)

, (k )

0, elsewhere

where r (k ) is the gamma fun ction :

r (k) = u k - 1e- lldu

(8-6)

If k is a p ositive integer , the gannna function takes the fonn:

,(k) ~ ( k-l )!

(8-7)

where ! denotes factorial. Th e mean (II) and standard deviation ( 0") for a gannna -distributed
variable are :

11 = -

and

Jk

(8-8)

(Y ~ -

8.5.3.3 \Veibull Dis tribution

Th e Weibull distribution is an extreme value distribution of type III [Soong (2004)]. Equation 89 presents the probability density fun ction for the Weibull distribution. Similar to the gamma
distribution, k defines the shape of the distribution and A is a scale parameter.
(8-9)

Th e mean ( II) and the standard deviation ( 0") for a Weibull-distributed variable are:

(8- 10)

223

Brown (2008) notes that the Weibull distribution is appropriate for developin g fragility functions
because it provid es accurate results with small data sets and represents a broad range of
distribution shapes, which enables easier fittin g.
8.5.3.4 Beta Distribution
Th e beta distribution is a versatile distribution defined on the interval [0, I]. The probability
density function for the beta distribution is:
I

xa-l{l _ X)P-l

for 0 :<: :; x:<:::; 1

Ix (x) ~ B(a , fJ)

{

(8- 11)

0, elsewhere

where B is the beta function as calculated using Equ ation 8-12.

B(a ,fJ) ~ r(a )r(fJ)
r(a + fJ)

(8- 12)

Th e parameters a and fJ are both shape parameters that take on positive values only. When
a,fJ > I, the density function is unim odal with the peak at (a - I) /(a+fJ - 2) . The density
func tion becomes U-shaped when a,fJ < I ; J-shaped when a ;: :': I, fJ < I ; reverse J-shaped when
a < I , fJ ;:::': I ; and uniform when a = fJ = I [Soong (2004)]. The mean (J1 ) and standard
deviation ( a ) of a beta-distributed variable are:

a
J1 = a+fJ

and

a=

a+fJ V~

(8-13)

Sin ce the beta distribution is define d on the interval [0, I] , expe rimental data mu st be
transformed to the interval [0, I] before fittin g the distribution.

8.5.4 Method of Maximum Likelihood

Th e method of maximum likelih ood is the most widely used rule for findin g point estimations of
distribution pa rameters for sample data and is used below to estimate the parameters for the
afore mentioned distributions. This method makes use of the sample likelihood fun ction that is
presented below for completeness.
First, let f(x; {)) be the probabili ty density function of the pop ulation where () is(are) the
par ameter(s) of the distribution X . The joint density function of a sample X l ' X 2 ,
form:

Ix,x,.x. (xpx" ., x. 18 ) ~ I x, (x,)l x, (x, )

>

where II denotes product.

224

>

I x. (x. ) ~ ITl x (x, 18)

.. ,

X n has the

(8-14)

Th e likelihood func tion L (O) of a set of

11 sample

values is

"

L (Olx" x, ..x") ~ ITl x (x,lo )

(8- 15)

i_I

which gives the relative likelih ood of havin g observed this particular sample ( X l =
X 2 = x2 ,

.. ,

Xn =

Xn )

Xl'

as a function of 0 [Benjamin and Comell (19 70)].

Th e maximum likelihood estimate (MLE) of 0 is the value that maximi zes the likelih ood
function L(O) . Per Benjamin and Cornell, maximum likelihood estimators possess the followin g
desirable properties

Asymptotical unbtasedness : means are asymptotically (1I ----t 00 ) equal to the tru e
para meter value(s), 0

Efficiency : minimum expec ted squared error among all possible unbi ased estimators,

Cons istency : be close to the true parameter values as the sample size increases with
increasingly high probability, and

Sufficiency : make maximum use of the information contained in the data.

8,5.5 Goodness-or-Fit Testing

To quanti fy the utili ty of the distributions, goodness-of-fit tests were performed on each fit. The
Kolmogorov-Srnimov test (denoted hereafter as the K-S test) is a general test that is applicable to
any distribution. The Lilli efors test is a special case of Kolmogorov-Smim ov test that can assess
normality only.
8.5.5.1 Kolmogorov-Smirnov test (K -S test)
Given a set of samp le values Xl ' X 2 ,
the K-S test ( D ) is calculated as:

... ,

x n observe d from a population X , the test parameter for

(8- 16)
In Equation 8-16, Fx (x j ) and Sx(:\";) are the theoretical and empirical CDFs, respectively,
calculated at the ill observation. Thus, the test para meter ( D ) of the K-S test corresponds to the
maximum of the absolute values of 11 (sample size) differences between the emp irical CDF and
the hypothesized CDF evaluated for the observed samples. The distribution of D is independent
of the hypothesized distribution and is a fun ction of 11 only [Benjamin and Comell (19 70)].
Th e null hypoth esis, H o ' for the K-S test is that the popul ation X comes from the hypothesized
proba bility distribution. At a specified significance level ( a) , if D is less than or equal to D crit ,
the null hypothesis is accep ted.

(8- 17)

225

Values for D crit have been tabul ated [e.g., Soong (2004)] as a fun ction of the samp le size ( 11 )
and significance level ( a ).
Th e advantage of the K-S test is that it is applicable to all sample sizes and uses data in unaltered
form (does not requir e arbitrary grouping of the sample data) unli ke the chi-square goodness-offit test. However, the K-S test is valid strictly for continuous distributions and values for D cr!1 are
based on a complete ly specified hypothesized distribution (parameters known). Thi s is not the
case herein since the distribution parameters are unknown and mu st be estimated from the data.
Using the K-S test with estimated distribution parameters may result in an unconservative
acceptance of the null hypothesis [Benjamin and Cornell (1970)]. Herein, the K-S test is used to
evaluate the quality of the fit of the distributions relative to one another.

8.5.5.2 Lilliefors test

Th e Lilli efors test evaluates the normality of a given data. Herein, the test is used to assess the
accep tability of the lognormal distribution by testing the logarithm of the available data. This test
is a variant of the K-S test to account for computation of the distribution parameters from the
sample data [Lilliefors (1967 )]. The test statistic for the Lilli efors test is calculated in a similar
manner to that for the K-S test; the difference is in the calculation of D crff The Lilli efors test
accounts for the unknown sample mean and variance and uses Monte Carlo simulation to
determine approximate values of D crit . The Lilli efors test is valid for small sample sizes but is
also appropriate when the distribution p arameters are unknown [Lilli efors (196 7)], which is the
case in this study.

8.5.6 Parameters That Effect the Earthquake Performance of Squat Walls

A detailed discussion of the failure modes for squat reinforced concrete walls was presented in
Section 1. Based on that discussion, wall geometry, aspec t ratio, horizontal web reinforcement
ratio, vertical web reinforcement ratio, and axial load are identifi ed as the five key parameters
that affect the earthquake response of squat reinforced concrete walls. Walls with bound ary
elements are more susceptible to diagonal compression failure than rectangular walls. As aspect
ratio increases, the likelihood of developin g wall flexur al strength increases and the likelih ood of
shear failure decreases. Flexural and shear failures exhibit different damage patterns. Horizontal
reinforcement is effective in resisting diagonal tension failure, and shear crack width is affected
by horizontal web reinforcement ratio. Vertical web reinforcement is effective in resisting the
opening of diagonal cracks and helps to anchor compression struts that transfer lateral force s to
the wall foundation. Higher axial force delays diagonal tension cracking but can trigger diagonal
compression failure . The effect of these five parameters on the damage data is investigated
below.
Table 8-9 and Table 8-10 present a statistical summary of drifts for each MoR obtained using
Me thods I and 2, respectively, with respect to wall geometry . Re sults are presented to two
decimal digits. In these tables, means ( x) , stan dard deviations (sd) , and coefficient of variation
( cv) are calculated as follows

226

T ab le 8-9 Statistical summary of drifts for each l\loR and Method 1

'Vall geometry
Rectangu lar

MoR

Barbell

Flanged

sd

cv

sd

cv

sd

cv

0.17

0. 16

0.95

0.04

0 .02

0.57

0.13

0.16

1.24

2a

0.47

0 .19

0.40

0.43

0 .20

0.46

0.59

0.38

0.64

N /A

N /A

0.75

0 .27

0.37

2b

0.58

0.20

0.34

N /A

1.07

0.29

0.27

0.35

0 .16

0.46

0.80

0.33

0.42

1.39

0.51

0.36

0.88

0 .14

0.16

1.45

0.63

0.4 3

1. There

IS

one data pomt for MoR-2b and barbell walls.

T ab le 8-10 St atistical summary of drifts for each MoR and Method 2

' Vall geometry
Rectangular

MoR

Barbell

Flanged

sd

cv

sd

cv

sd

cv

0.11

0.11

1.06

0.03

0 .01

0.33

0.06

0.05

0.86

2a

0.43

0 .15

0.36

0.38

0 .19

0.50

0.42

0.22

0.53

2b

0.58

0.20

0.34

N /A

N /A

N /A

0.77

0 .30

0.39

1.10

0.31

0.29

0.35

0 .16

0.46

0.80

0.34

0.4 3

1.32

0.42

0.32

0.88

0 .14

0.16

1.45

0.63

0.4 3

1. Th ere

IS

one data po mt for MoR-2b and barbell walls.

1 "

X ~ - L x,
11 i_ I

,]1"

1 "
_
, sd ~ L( x,- x )
[ 11 - 1 i_I

, cv

r-

sd Ti

(8-18)

where n is th e number of samples of Xi . The dri ft va lues obtain ed for wa lls with barbell cross
sections are generally much lower th an those for wa lls with rec tangular and flanged cross
sections. No te th at the m ean drifts for Mo R-3 are smaller than Mo R-2 for wa lls with barb ell
cross sections, which is attributed to the dataset used to generate dam age data for barbell wa lls
th at includes many wa lls with low-aspect ratios and high vertical and horizontal web
reinforce ment rati os (see Tabl e 8-3 and Tabl e 8-4). Given th e results of Table 8-10 and th e
range s of data present ed in Table 8-1 throu gh Tabl e 8-6, families of frag ility fun ctions are
presented for eac h wa ll geometry .
Figure 8-16 presents the variation of drift with aspec t rati o for Mo R- I, MoR-2a, MoR-3, and
MoR-4 . All ava ilable damage data are included in the figure (also in Figure 8-17, Figure 8- 18,
and Figure 8- 19) . For barbell an d flanged wa lls, no trends are evident for any of the MaRs
considered. For wa lls with rec tangular cross-sec tions, the re is a wea k trend of increasing drift

227

ba rbo ll
fianged

....ctangul.. r

0.75

0'

i:'

."

.~

0.5

'

Vi

1.5

'
.

..

' "..1

"[~

If

-0
if.

3~

'1

bar bel
flanged

,.. "tangutar

o5 ~
-

0'
()

'

t .iI
"

"

~
-0
'

1.5

0.5

',

bar bo l
flanged

,,,,,tang ular

'

V; 1

C
-

I.
'
,

'::::

"
...

2.5

I'
I .

Aspect ratio (II / I )

1.5

b) MoR-2a

~
-e
,
C'

'

"

':::: t .5

"

0.5

a) MoR-1
00

'

Aspect ratio (II / I )

If

' . 'I ' .

. I:I ll . .' . I

rec ta ngular

: .
'.

0.5

. .. . U

I ~ -.:::

()

".

" ' ,'i

025

c'

'C
-0

bar bo l
flanged

'

e:

' CC "

U5

1.5

"

". ,

. :.

. .
. '

. . ..'
.

'

,
!

0.5

1.5

Aspect ratio (II / I )

If

c) MoR-J

1\

0.5

1.5

Aspect rat io (II 1/ )

W

1\

d) MoR-4

Figure 8-16 Variation of drift wit h aspec t ratio for di ffe ren t met hods of repa ir

with increasing aspect ratio for MoR- I . However, scatter is significant and the size of the
rectangular wall dataset for MoR- 1 for aspec t ratios of between 1.0 and 2.0 is small, with most of
the data clustered aroun d 2.0. In summary , the available data do not reveal any strong correlation
between aspect ratio and damage for any wall geometry.
Figure 8- 17 presents the variation of drift with horizontal web reinforcement ratio for each
method of repair. No relationship between drift and horizontal web reinforcement ratio can be
identifi ed for any method of repa ir for either rectangul ar or flanged walls. Th e limiting drifts for
MoR-1, MoR-3, and MoR-4 for barbell walls are independent of horizontal web reinforcement
ratio. A weak trend of increasin g drift with increasin g horizontal web rein forcement ratio is seen
for M oR-2a but the scatter is quit e large and sample size is modest.
Figure 8-18 presents the variation of drift with vertical web reinforcement ratio for each method
of repair. No visible trends between drift and vertical web reinforcement ratio are evident for any
method of repair for rectangular and flanged walls. For the barbell walls, the limiting drift s for
MoR- l , MoR-3 and MoR-4 are independent of vertical web reinforcement ratio. For MoR-2a, a
trend of increasing drift with increasin g vertical web reinforcement ratio is evident. Note that all
barbell walls considere d herein had equal percentages of horizontal and vertical web
reinforcement an d that the trends obs erved with drift and vertical web reinforcement ratio for
barbell walls for all methods of repair are similar to those observed with drift and horizontal web
reinforcement ratio.
Figure 8-19 pre sents the variation of drift with normali zed axial load for each method of repair.
No significant trend s between drift an d normalized axial load are evident for any method of
repa ir for an y cross-section type. However, the numb er of data for normalized axial loads of
higher than 10% is very small as seen in Figur e 8-19.
In summary, the available data do not support the use of aspect ratio, horizontal and vertical web
reinforcement ratio, and axial load as variables in the presentation of fragility curves. Although
horizontal and vertical web reinforcement ratios have an impact on the limiting drifts for MoR2a, the dispersion is substantial. Only wall geometry substantially influ ences the relationships.
Th e barbell wall dataset yielded significantly different damage charac teristics than walls with
flanged an d rectangular walls as presented in Table 8-9 and Table 8-10 . Although the damage
data for rectangular and flanged walls are compa rable, the ranges on the two datasets are
significantly different.

8.5.7 Fragility Functions

Fragility functions that define the probability that a certain method of repair will be required
conditional on story drift developed in this section. Th e damage states, which are associated with
specific methods of repairs (see Section 8.4), are used as the basis for collecting the damage data.
Th e data characterization methods discussed in Section 8.5.2 are used to group the available
damage data.
This section utili zes the distributions summarized in Section 8.5.3 to fit the damage data usin g
the method of maximum likelihood (see Section 8.5.4). Fragility functions are generated for each
type of wall geometry .

229

0.75

barbel
flanged

rectangul ar

--::::~

1.5

.~

0.5

C
-0

i::'
o
U5

.. '

c'

Vi

l e o

0 25

"

..

o
()

"C

0.5

. .

1.5

2.5

'"

"

I ..

. . ..

0.5

25[

7;
'-'
c

.5

'.

-.:;l

C'
c

if.
0 5f
-

I.

'

7;
'-'
-0

'

2.5

0 :0
I

'

..

'"

I
t

, , '

.
.

,,,,,ja"9u1a,

0
'

barbol
fla nged

I .

1.5

I
I

I
0.5

0.5

ot
()

0
r7.

..'

c,

2.5

I'
l ,..-, ..
.
'

,,,,,,ta"9uta,

.
' , "

'

3~

barbel
flanged

b) MoR-2a

~ OO

a) MoR-1

.
,

1.5

~ OO

rectangular

'

O
0

I ' ~1 . 11"
f"'
o ., ~"' ,

0 .5

'

barbol
fla nged

e:c

1.5

~ OO

C) MoR-J

2.5

0
0

0.5

1.5

2.5

~ OO

d) MoR-4

Figure 8-17 Variation of drift with horizon tal web reinforcement ratio for differe nt methods of repair

0.75
--::::-

barbel
flanged

rectangul ar

.~

1.5

i::'

U5

. .
.

()

0.5

'
1.5

C
-0

I :

0 25

Vi

at

'.

. 1

0.5

~_ _

1.5

2.5

2~

'1

'-'1 .5

.2

"

v.

o5 ~

."j

'

barbel
flanged

,,,,,,ta"9uta,

"
' . .

'-'
C
C
-0
,

'

.2
V;

1.5

~~

C) MoR-J

2.5

.,

barbol
fla nged

,,,,,ja"9u1a,

I .

, I

. ,, ' ,
.
".
.'
.

, '

0.5

I
0.5

1.5

ot
()

. .

'l 2

I .

2.5

..

3~

-e
.

b) MoR-2a

"

~ (~

I.

' "

a) MoR-1

2.5 [

~~

-t::

....

tI "

'~o,-o~

' .

I' I' ~ _r" .

1
'"
.
,
.
:
..
,-__

00

2.5

0 .5

. '
I

. '.

' .'

C,
0

:
I

rectangular

e:c

0 .5

barbol
fla nged

"C

,
.

'

'"

.'
_

0
0

0.5

1.5

2.5

~ (~

d) MoR-4

Figure 8-18 Variation of drift wit h vertica l web reinforce ment ratio for diffe rent methods of repair

0.75

barbel
flanged
rectangular

"' OS

-,: ':..

"

'C

'C

-e

.; ~

~ -!

'" 025
0

C'

" 0'
C-

10

,
20

IS

2S

0
30

.. .
-...... ..

10

'1
'-'

l .5

"

'C
~

C'

"'"

O.5 r

, . -.'

.....,

2,5

,,,,,,ta"9 uta,

'C

1.5

-0

"'"

10

,
IS

PI A!c' (% )

C) MoR-J

20

2S

30

C'

'l

"

20

OS

~.

0'
0

barbel
flanged

IS

30

2S

b) MoR-2a
3

"I

PI A!c' (%)

a) MoR-1
N

PI A!c' (% )

rectangular

~~
5

barbol
fla nged

II

L5

-........
'.

barbol
fla nged
,,,,,ja"9 u1a,

10

IS

20

PI A!c' (%)

d) MoR-4

Figure 8-19 Variation of drift with no rmalized axia l load for different methods of repair

2S

30

8.5. 7.1 Fragility Functions Developed Using Method 1

Method 1 utili zes all ava ilable damage data for eac h method of repa ir. The distribution
parame ters (as defined in Section 8.5.3) estimated usin g th e method of maximum likelihood are
presented in Table 8-11 for rec tangular, barbell and flanged wa lls. Results for the K-S (all
distribution s) and Lilli efors (lognorma l distri buti on only) goo dne ss-of-fit tests are presented in
Tabl e 8-12 and Tabl e 8-13. The goodness-of-fit tests of Table 8-12 and Tabl e 8-13 were
conducted at the 5% significance level.
Table 8-11 Distribution parameters computed using Method 1
Distribution Parameters

w an
Geometry

Rectangular

l\loR

\Veibull

beta

f3

f3

0.11

0.92

1.40

8.37

0.18

1.16

1.20

5.54

2a
2b

0.43
0.54

0.43
0.36

6.06
8.49

12.98
14.76

0.53
0.64

2.61
3.26

2.63
2.16

3.38
2.79

1.03
1.30

0.28
0.37

13.43 12.59
7.69 5.52

1.18
1.56

4 .09
3.00

1.39
1.70

1.58
2.95

0.04
0.38

0.47
0.50

4.27 101. 72 0.05

4.61 10.76 0.4 8

1.92
2.3 7

0.87
2.26

2.47
3.20

0.32
0.87

0.45
0.17

5.29 15.05
25.20 28.82

0.40
0.94

2.3 7
6.24

1.97
3.95

2.17
4.20

0.07
0.48

1.03
0.68

1.02
2.54

8.08
4.32

0.12
0.66

0.94
1.66

0.68
2.15

3.85
4.94

0.7 1

0.34

8.78

11.69

0.84

3.28

0.36

0.40

0.75
1.32

0.32
0.44

8.79
5.54

11.00
3.82

0.90
1.64

2.42
2.56

0.69
1.01

2.66
1.52

2a
3
4
1

Flanged

gamma

4
1
Barbell

lognormal

2a
2b
3
4

Th e data presented in Tabl e 8- 12 include the K-S test pa rameter ( D), decision on the null
hyp othesis that the data comes from the hyp othesized probability distribution (A = Acc ept, R =
Reject), p va lue of th e test and th e critical test parame ter ( Dent). As noted previously, if

D :-:;; Dent ' the null hyp othesis ( H o) is accepted. The null hyp othesis for the K-S test is reject ed in
1 of the 14 cases (3 wa ll geome tries x 5 MoR except for MoR-2 b and barbell walls) for th e
gamma, We ibull and lognormal distributions (barbe ll walls, MoR4), and rej ected two times for
the beta distri buti on (rectangular wa lls, MoR-2; barbell wa lls, MoR- 4). The D statistic for the
K-S test repr esents th e maximum absolute difference between the emp irical CDF and th e
theo retical CDF . Therefore, it is a measur e of the devi ation between the repo rted data and th e
hyp othesized CDF . In Table 8-12, the yellow shaded cells rep resented th e sma llest D for the KS test cond ucted on the correspo nding wa ll geome try and m ethod of repair. For 6 of the 14 cases
investigated , the logn ormal distri buti on yields the smallest va lue s of D . In another thr ee
instances, th e va lue of D for the logn ormal distri buti on is only marginally 10%) larger th an

233

Table 8-12 K-S test results for each distribut ion obtained using Method I

K-S Test Results

wan
geo me try

Mo R

logno rmal

o.;
p

Rectangul ar

N
W

""

110

\ Vcibull

gamma

110

110

beta

110

0.166 0.9 11

0.068 0.492

0.102 0.498

0.10 1 0.237

0.126

2a

0.16 1 0.327

0.113 0.508

0.097 0.193

0.128 0.038

0.167

2b

0.294 0.288

0.213 0.399

0.194 0.720

0.150 0.49 1

0.180

0.215 0.627

0.119 0.556

0.126 0.535

0.128 0.656

0.1 16

0.203 0.678

0.108 0.892

0.086 0.92 1

0.082 0.8 15

0.095

0.20 1 0.770

0.098 0.558

0.117 0.273

0.147 0.348

0.138

2a

0.221 0.903

0.093 0.946

0.086 0.81

0.104 0.632

0.122

0.338 0.885

0. 145 0.857

0. 151 0.725

0.172 0.132

0.290

0.234 0.002

0.327 0.002

0.327 0.005

0.297 0.005

0.298

0.194 0.4 10

0.127 0.078

0.182 0.175

0.158 0.111

0.172

2a

0.183 0.853

0.082 0.66 1

0 .098 0.573

0.106 0.124

0.159

2b

0.483 0.7 14

0.249 0.737

0.244 0.792

0.232 0.97 1

0. 174

0.264 0.936

0. 104 0.730

0.134 0.273

0.194 0.24 1

0.200

0.301 0.999

0.08 1 1.000

0.078 0.976

0.106 0.983

0.103

Barh ell

Fla nged

Table 8-13 Lilliefors re sults for the lognormal distribution computed using Method 1

wan
Geometry

Lilliefors Test Results

MoR

1
Rectangular

2a
2b
3
4

1
Barbell

2a
3
4

1
F langed

2a
2b
3

o.;

H,

0.11 0
0.107
0.192
0.14 2
0.134
0.132
0.146
0.219
0.1 54
0.128
0.121
0.304
0.1 73
0.197

0.500
0.02 8
0.020
0.214
0.247
0.381
0.500
0.4 38
0.001
0.0 58
0.500
0.322
0.500
0.500

0.067
0.113
0.2 10
0.116
0.106
0.096
0.090
0.153
0.326
0.126
0.080
0.230
0.107
0.084

R
R
A
A
A
A
A
R
A
A
A
A
A

the smallest value of D . The logn orm al distribution is therefore judged to be the best of the
distributions considered here. Table 8-13 presents the Lilli efors goo dness-of-fit test results on the
lognormal distribution. The Lilli efors test yields smaller values of D cril than for the K-S test for a
given sample size and significance level (see Section 8.5.5). As seen in Table 8- 13, the
lognormal distri buti on fails the Lilli efors goo dne ss-of-fit test for MoR-2a and MoR-2b for the
rectangular wa lls and for M oR-4 for the barbell wa lls. Note that no outlier analysis was
und ertaken before processin g the data.
Th e functions developed using lognormal distribution for each wa ll geometry and method of
repa ir type can be compared using the medians ( 8 ) and logarithmic standard dev iations (fJ )
present ed in the gray shaded columns of Table 8- 11. The standard deviati ons for Mo R- l are
generally high, which is attributed to the use of Me thod 1 that includes all ava ilable damag e data.
Th e standard deviations for other methods of repa ir are genera lly reasonable. ATC -58 [AT C
(2008)] qualifi es the fragilit y fun ctions that p ass the Lilliefors goodness-of-fit test an d yield
logarithmic standard deviations ( fJ) ofless than 0.6 as high quality. One interestin g observation
is that for barbell wa lls, MoR-3 precedes MoR-2a, that is, the median drift associated for Mo R-3
is less than that of MoR-2a. This anomaly is attributed to the charac teristics of the barbell wa ll
data used herein, nam ely, low-aspect rati o and heavy web reinforcement in most cases. Figur e
8-20, Figur e 8-2 1, and Figure 8-22 present the empirical and theoretical fragilit y functions for
rectangular, barb ell and flanged wa lls, respectively.

235

0.8

.f'

:g

0.6

.D

....o

0-

<1)

....

..2

'@

""""'"
--""""'"
--"""""''''''''''''''
--""""'"
---

Empirical fragility function (MoR-1)

Theoret ical fragility function (MoR- 1)
Empirical fragility function (MoR-2a)
Theoret ical fragility funct ion (MoR-2a)
Empirical fragility function (MoR -2b)
Theoret ical fragility funct ion (MoR-2b)
Empirical fragility function (MoR-3)
Theoret ical fragility funct ion (MoR -3)
Empirical fragility function (MoR-4)
- - - Theoret ical fragility funct ion (MoR-4)

0.:1-

u,

0.2

1.5

0.5

2.5

Drift (%)
Figure 8-20 Me thod 1 fragility functions for rectangular walls

0.8

.f'

:g

0.6 1- 1..

,...-/"

.D

....o

0-

<1)

....::l 0.:1- 1- 1

+ /

'@
~

0.2

""..".........,,,.,,,,.

0,25

"""""'''''''''''''' Empirical fragility function (MoR -1)

- - - Theoret ical fragility function (MoR-1)
""""'" Empirical fragility function (MoR-2a)
,,,,, , ,. I - - - Th eoretical fragility funct ion (MoR- 2a)
"""""'''''''''''''' Empirical fragility funct ion (MoR -3)
- - - Theoret ical fragility function (MoR -3)
Empirical fragility function (MoR-4)
- - - Theoret ical fragility function (MoR-4)

0.5

0.75

1.25

Drift (%)
Figure 8-2 1 Met hod 1 frag ility functions for barbell walls

236

1.5

0 .8

" i

0.2

0.5

""""'" Empirical fragility fu nction (MoR -1)

Theoret ical fragility function (MoR-1)
""""'" Empirical fragility function (MoR-2 a)
- - Theoret ical fragility function (MoR-2a)
""""'" Empirical fragility function (MoR-2 b)
- - Theoret ical fragility function (MoR -2b)
"""""'''''''''''''' Empirical fragility funct ion (MoR-3)
- - Theoret ical fragility funct ion (MoR-3)
Empirical fragility function (MoR-4)
- - Theoret ical fragility function (MoR-4)

;......... .A - - -

1.5

2.5

Drift (%)
Figur e 8-22 Meth od 1 fra gility functions for flanged walls
8.5 .7.2 Fragility Functions Develop ed Using M et hod 2
Method 2 utilizes the damage data with the smallest demand parameter (drift) for each method of
repair of each specimen. The distribution parameters are pre sented in Table 8-14 for rectangular,
barbell and flanged wall s. Re sults for the K-S (all distributions) and Lilliefors (lognorm al
distribution only) goodness-of-fi t tests are pre sented in Table 8-15 and
Tab le 8-16 at the 5% significance level. As seen in Table 8-15, the null hypothesi s for the K-S
test is rejec ted once (rectangular walls, MoR-4) for each distribution type for Method 2.
In Table 8-15, the yellow shaded cells identify the smallest value s of D for the K-S test for each

wall geometry and method of repair. For 5 of the 14 cases inve stigated, the lognormal
distribution yields the smallest D . Accordingly, the lognorm al distribution is used to develop
fragility curves for squat reinforced concrete walls.
Table 8-16 presents the Lilliefors goodness-of-fit test results. The lognorm al distribution fails the
goodness-of-fit test for MoR-2a, MoR-2b and MoR-4 for rectangular walls and MoR-4 for
barbell walls.
The logarithmic standard devia tions corre sponding to MoR-l are generally high but less than
those observed for Method 1. The logarithmic standard devia tion s for MoR-2a, MoR-2b, MoR-3
and MoR-4 are less than 0.60 except for the flanged walls and MoR-2a ( /3 = 0.62). Similar to
that observed for Method 1, MoR-3 precedes MoR-2a for barbell walls and MOR-2a. Figure
8-23 , Figure 8-24, and Figure 8-25 pre sent the empirical and theoretical fragility functions
developed using data characterization Method 2.

237

Table 8- 14 Distribution parameters computed usin g Method 2

Distribution Parameters
Wan
Geometry

Re ctangular

l\loR

\Veibull

beta

f3

f3

0.07

0.81

1.58

15.04

0.1 1

1.18

2.13

15.34

2a
2b

0.40
0.54

0.42
0.36

6.81
8.49

15.88
14 .76

0.48
0.64

3.01
3.26

5.78
2.16

6 .34
2 .79

1.05
1.25

0.30
0.35

11.79 10.75
9.10 6.90

1.21
1.47

4.03
3.55

1.27
2.15

1.32
2 .96

0.03
0.33

0.31
0.49

10.11 316.82 0.04

4.4 8 11.93 0.43

3.19
2.21

1.32
0.90

2 .48
1.42

0.32
0.87

0.45
0.17

5.29 15.05
35.22 40.12

0.40
0.94

2.37
6.24

1.97
3.95

2 .17
4 .20

0.04

0.79

1.71

29.17

0.06

1.29

0.65

1.48

2a
2b

0.36
0.72

0.62
0.37

3.27
7.79

7.71
10.14

0.4 8
0.86

2.03
3.22

3.81
0.33

5.45
0 .35

0.75
1.32

0.32
0.44

8.45
5.54

10.58
3.82

0.90
1.64

2.39
2.56

0.65
1.01

2 .51
1.52

2
3

4
1
Flan ged

ga mma

Barbell

logn ormal

238

Tab le 8-15 K-S test results for eac h distribution computed using Me tho d 2

K-S Test Results

wan
geo metry

MoR

lognormal

Rectangula r

N
W

-o

\ Vcibull

gamma

beta

Dcrit

110

110

110

110

0.205 0.902

0.086 0.304

0. 147 0.365

0.139 0.234

0.156

2a

0.203 0.266

0.150 0.526

0. 12 1 0.369

0.137 0.555

0.119

2b

0.294 0.288

0.213 0.399

0. 194 0.720

0.150 0.49 1

0.180

0.242 0.752

0.120 0.766

0. 119 0.78 1

0.117 0.852

0.108

0.215 0.3 18

0.152 0.440

0. 138 0.7 17

0.111 0.088

0.198

0.246 0.662

0.132 0.529

0. 146 0.398

0.162 0.607

0.138

2a

0.309 0.869

0.136 0.893

0.131 0.785

0.149 0.386

0.206

0.338 0.885

0.145 0.857

0. 151 0.725

0.172 0.132

0.290

0.234 0.002

0.327 0.002

0.3 17 0.005

0.297 0.005

0.298

0.30 1 0.568

0.174 0.357

0.206 0.4 19

0.196 0.290

0.2 18

2a

0.254 0.88 1

0.110 0.993

0.081 0.945

0.099 0.299

0.182

2b

0.519 0.609

0.29 1 0.637

0.285 0.584

0.297 0.980

0.180

0.269 0.9 10

0.111 0.835

0. 123 0.282

0.196 0.226

0.207

0.301 0.999

0.08 1 1.000

0.078 0.976

0.106 0.983

0.103

Barhell

Flanged

Table 8-16 Lilliefor s results for the lognormal distribution computed using Method 2
Wall
Geometry

Rectangular

Barbell

Flanged

Lilliefors Test Results

MoR
1
2a
2b
3
4
1
2a
3
4
1
2a
2b
3
4

o.;

u,

0.135
0.134
0.192
0.159
0.142
0.16 1
0.20 1
0.2 19
0.154
0.197
0.167
0.324
0.176
0.197

0.500
0.0 15
0.020
0.364
0.029
0.194
0.417
0.438
0.00 1
0.148
0.500
0.194
0.500
0.500

A
R
R
A
R
A
A
A
R
A
A
A
A
A

D
0.087
0.15 1
0.210
0.116
0.150
0.134
0.143
0.153
0.326
0.17 1
0.110
0.270
0.109
0.084

0.8

.f'

:.g

0.6

..D

...o0-

...::l

""""'"
-""""'"
-"""""''''''''''''''
-""""'"
--

Empirical fragility function (MoR -1)

Theoretical frag ility function (MoR -1)
Empirical fragility function (MoR -2a )
Theor etical frag ility fun ction (Mo R-2a)
Empirical fragility fun ction (Mo R-2b)
Theoretical frag ility function (MoR-2b)
Empirical fragility function (MoR -3)
Theoretical frag ility function (Mo R-3)
Empirical fragility fun ction (Mo R-4)
- - Theoretical frag ility function (MoR -4)

<1)

0.4-

-@
~

0.2

0.5

1.5

2.5

Drift (%)
Figure 8-23 Method 2 fragility functions for rectangular walls

240

0.8

.f'

:g

0.61-1 +

,....J-iI........

,..............

, oi

.D

....o

0-

<1)

....

..2 0.:1- 1

1 .. . ........................

'@

Id~,~===::::==============::; 1

"""""'''''''''''''' Empirical fragility function (MoR -1)

- - - Theoret ical fragility function (MoR-1)
""""'" Empirical fragility function (MoR-2a)

u,
, .;. """".."""""""

0.2

:...

I - - - Theoret ical fragility funct ion (MoR-2a)

"""""'''''''''''''' Empirical fragility function (MoR -3)

- - - Theoret ical fragility funct ion (MoR -3)
Empirical fragility function (MoR-4)
- - - Theoret ical fragility funct ion (MoR-4)

0,25

0.5

0.75
Drift (%)

1.25

1.5

Figure 8-2 4 Met hod 2 frag ility functions for barbell walls

0.81-+ """"",, ,, ",,

;" / +,,

,,"' ,,'1"""",,

,,"""",,,,""".,

"""""".""

,;>,,,.,.,,,,,, ,,,,,,,,,

,,.,..,,,,,,,,

, oi

.f'

:g

0.6

.D

....o

0-

""""'" Empirical fragility fu nction (MoR -1)

<1)

....

::l

0.:1-

I f

'f ;

'@

11 ................ '"!

.A - - - Theoret ical fragility funct ion (MoR-1)

""""'"
--"""""''''''''''''''
--..,.- "",""""..... ,
""""'"
---

Empirical fragility function (MoR- 2a)

Theoret ical fragility funct ion (MoR-2a)
Empirical fragility funct ion (MoR -2b)
Th eoretical fragility funct ion (MoR- 2b)
Empirical fragility function (MoR-3)
Theoret ical fragility function (MoR -3)
Empirical fragility function (MoR-4)
- - - Theoret ical fragility function (MoR-4)

0.2

0.5

1.5

2.5

Drift (%)
Figure 8-25 Method 2 fragilit y functions for flanged walls

24 1

8.6 Fragility Function Recommendations

8.6.1 Probability Distribution
Th e fragility data presented in Section 8.5 .7 reveal that the lognormal distribution is the best of
the four distributions cons idered for repre senting fragility functions for squat reinforced concrete
walls for both Me thod I and Me thod 2.

8.6.2 Data Char acter ization

Th e fragility curves obtained usin g the Iognonnal distrib ution differ only modestly for Me thod I
and Me thod 2 (see th e gray shaded cells in Table 8-11 and Table 8-14) . The dispers ions for
Me thod 1 are generally greater than those for Me thod 2 for the four methods of repa ir.
Accordingly, fragility functions are developed using Me thod 2.
Table 8- 17, which is based principally on Table 8-14, lists the values of median and dispersion
obtained using the lognormal distribution and Me thod 2 for the three wall geometries and four
methods of repair. Two significant figures are used to report median ( 8) drift. Two decimal
digits are used for dispersion (p) .

8.6.3 Evaluation of Damage States

8.6.3.1 Damages States As sociated with Reinforcement Yielding for :\IoR-2a
MoR-2a uses reinforcemen t yielding data (OS2. 1, OS2.2, and OS2.3) together with OS2.4a and
OS2.5a that are associated with crack widths of 0.5+ mm . This assumption is evaluated herein by
fitting lognormal distribution usin g Me thod 2 to a) reinforcement yielding data (OS2.1, OS2.2
and OS2.3), and crack width data (OS2 .4a and OS2.5a). Table 8- 18 pre sents the para meters of
the distribution. The pa rameters for barbell walls and OS2.4a and OS2.5a cannot be obtained due
to a lack of data. For rectangular and flanged walls, the medians calculated using yielding data
(OS2 .1, OS2.2, and OS2.3) and crack width data (OS2.4a and OS2.5a) are in close agreement.
Th e data of Table 8-18 support the use of reinforcement yielding data and crack wid th data for
MoR-2a.

8.6.3.2 Damage States Associated with Sup plemen tal C r iter ia for l\I oR -4
Two supplemental criteria are sugg ested for MoR-4 to aid in identification of damage for cases
where the rep orted data is insufficient to make an assessme nt. Th e first, SCI, serves to identify
drift at sliding shear failure. The second, SC 2 , is intended to the identify thr eshold drift for wall
rep lacement for walls that fail under diagonal tension or compression, for cases where the drift at
failure is unclear. The drifts identified using the supplemental criteria are presented in Appendix
C.

Table 8-19 pre sents the distribution pa rameters for MoR-4 calculated with and without SCI and
SC 2 . As seen in the table, the use of the supplemental criter ia does not significan tly alter the
distribution paramet ers. Criterion SCI identifies 13 data points for rectangular walls, 5 data

242

Table 8-17 Logn orm al distr ibut ion parameters calculated using
'Vall geometry

Rect angular

Ba rbell l

Flanged

l\loR

f3

0.07

0.81

2.

0.40

0.42

2b

0.54

0.36

1.05

0.30

1.25

0.35

0.03

0.31

2.

0.33

0.49

0.32

0.45

0.87

0.17

0.04

0.79

2.

0.36

0.62

2b

0.72

0.37

0.75

0.32

1.32

0.44

1. Data not presented for MoR -2b , see footnote

Table 8-10 .

~Ieth od

III

Table 8-18 Lognorm al distribution parameters calcul ated usin g Method 2 for
dam age states
'V all geometry

Rect angular

Barbell

Flanged

Damage states

f3

DS2.1, DS2.2, DS2.3

0 .41

0.3 1

DS2.4 a, D S2.5a
DS2.1, DS2.2, DS2 .3,
DS2.4 a, D S2.5a
DS2.1, DS2.2, DS2.3

0.42

0.54

0.40

0.42

0 .31

0.49

N /A '
DS2.4a, DS2.5a
DS2.1, DS2.2, DS2 .3,
0.33
DS2.4 a, D S2.5a
DS2.1, DS2.2, DS2.3 0 .41

N /A

DS2.4 a, D S2.5a
DS2.1, DS2.2, DS2 .3,
DS2.4 a, D S2.5a

0.49
0.5 1

0.40

0.84

0.36

0.62

1. Fitting IS no t poss ible since there are only 3 data po ints

associated with these damage states.

243

~l oR-2 a

points for flanged wa lls and 0 data points for barbell walls. Criterion SC z identifies 6 data points
for flanged walls, 2 data p oints for barbell wa lls an d 0 data points for rectangular wa lls.
The data of Table 8-19 shows that the use of the two supplemental criteria does not substantially
modify the values of the distribution p arameters. Accordingly, the reco mmended medians and
dispersions presented below mak e use of supplemental criteria.
Table 8-19 Effect of SCI an d SC 2 on the l\loR-4 lognormal d ist ribution p aramete rs for
~Iethod 2
SC I

SC,

f3

Considered

Considered

1.25

0.35

Considered

Not considered

1.25

0.35

Not considere d

N ot considered

1.15

0.40

Considered

Considered

0.87

0 .17

Considered

Not considered

0.86

0.18

Not considere d

N ot considered

0.86

0.1 8

Considered

Considered

1.32

0 .44

Considered

Not considered

1.31

0.51

Not considere d

N ot considered

1.27

0.61

'Vall geo metry

R ect an gular

Barbe ll

Fl anged

8.6.4 ACI 318 -08 Comp liance

Eleven of the III test specim ens (6 rectang ular, 2 flanged and 3 barbell) used to develop
fragility fun ctions do not comply with the minimum reinforcement requirements
( Ph = Pv = 0.25%) of AC I 318-08 [ACI (2008) ] for Sp ecial Struct ural Walls and COl/crete

Beams (Chapter 21.9). This section presents frag ility functions for wa lls that comply with the
minimum reinforcement requirements of AC I 318-08 [ACI (2008)]. The lognorm al distribution
and Method 2 are used to develop the fragility functions. Re sults are presented in Table 8-20 .
A compa rison of the distribution p arameters presented in Table 8-17 and Table 8-20 reveal that
limiting the min imum web reinforcement ratios p er AC I 318-08 [ACI (2008)] does not
significantly affect the distribution parameters. The most sign ificant effect is observed for MoR2a for flanged walls for which the median drift increased from 0.36% to 0.39% an d logarithmic
standard deviation decreased from 0.62 to 0 .51.
Given that most squat wa lls will likely comp ly with the AC I 318-08 [ACI (2008)] require ments
for minimum rebar ratio, the medians and dispe rsions recomm ended in Section 8.6.5 are based
on the data of Table 8-20 .

244

Table 8-20 Lognormal distribution parameters and the corresponding Lilliefors test results
for squ at walls that comply with the minimum rein forcement requirements of ACI 318-08
lAC) (2008)]
Lognormal
'Vall geometry

Rectangular

Barbell

Flan ged

l\loR

Lilliefors Test Results

o:

f3

0.07

0.79

2.

0.4 1

2b

H.

0.14 0 0.500

0.082

0.34

0.14 0 0.182

0.11 8

0.55

0.34

0.207 0.14 3

0.181

1.09

0.27

0.164 0.338

0.122

1.30

0.35

0.1 54 0.002

0.202

0.03

0.31

0.170 0.379

0.123

2.

0.34

0.53

0.219 0.500

0.14 0

0.33

0.33

0.242 0.500

0.159

0.87

0.18

0.161 0.001

0.327

0.05

0.76

0.207 0.190

0.173

2.

0.39

0.51

0.173 0.500

0.089

2b

0.72

0.37

0.324 0.194

0.270

0.76

0.33

0.1 84 0.500

0.10 0

1.34

0.4 5

0.201 0.500

0.11 0

8.6.5 Recommendations
8.6.5.1 Rectangular 'Valls
Table 8-2 1 presents the recommended medians and dispersions for rectangular walls . On the
basis of input from the ATC-58 project team, MoR-2 is represented by MoR-2b, namely, cracks
widths in excess of 1.0 mm. The methods of repair in COIUlIlll I of the table are mapped to
damage states per the ATC-58 notation in column 2.
Table 8-21 Distribution par ameters for rectangular walls
MoR

Damage
State

f3

DSO

0.07

0.79

DSI

0.55

0.34

DS2

1.09

0.27

DS3

1.30

0.35

245

8. 6.5.2 Barbell Walls

Table 8-22 pre sents the rec ommended medi ans and dispersions for barbell wa lls . Since the
median drifts for M oR-2a and MoR-3 are virtually identi cal , MoR-2 is set aside for barbell walls.
Th e meth ods of repair in column I of the table are mapped to dama ge states per the A Te-58
notation in column 2.
T able 8-22 Distribution parameters for barbell walls
l\loR

Damage
St ate

fJ

DSO

0.03

0.31

DS2

0.33

0.33

DS3

0.87

0.18

8.6 .5.3 Flanged Walls

Table 8-23 presents the recommended medians and dispe rsions for flanged walls. For the reason
given in Section 8.6.5. 1, MoR-2 is represented by MaR-2b. However, given that the median
drifts for MoR-2 (=0.72) is virtually iden tical to that for MoR-3 (=0.76), MoR-2 is set aside for
flanged walls . The meth ods of repair in column 1 of the tabl e are m apped to dama ge states per
the ATe-58 notation in colunm 2.
Table 8-23 Di stribution parameters for flan ged walls
l\loR

Damage
St ate

fJ

DSO

0.05

0.76

DS2

0.76

0.33

DS3

1.34

0.45

8.7 Scopes of Repair

To establish consequence functions and costs ofrepa ir, a list ofrepair activities must be assigne d
to each meth od of repair. Such lists are pre sen ted in Appendix E for me thods of repair MoR-1
through MoR- 4 .

246