Direct Displacement-Based Design of Frame-Wall Structures-Sullivan, Priestley, Calvi-2006

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 34

April 28, 2006 13:47 WSPC/124-JEE 00274

Journal of Earthquake Engineering


Vol. 10, Special Issue 1 (2006) 91124
c Imperial College Press
DIRECT DISPLACEMENT-BASED DESIGN OF
FRAME-WALL STRUCTURES
T. J. SULLIVAN, M. J. N. PRIESTLEY and G. M. CALVI
ROSE School, University of Pavia, Via Ferrata 1, 27100 Pavia, Italy
A direct displacement-based design (DBD) procedure for structures that comprise both
frames and walls is presented in this paper. Within the new procedure, strength pro-
portions between walls and frames are assigned and are used to establish the design
displacement prole before any analysis has taken place. Knowledge of the displacement
prole and recommendations for the combination of frame and wall damping compo-
nents enables representation of the structure as an equivalent single-degree of freedom
system. The Direct DBD process is then utilised to set the required strength level which
is proportioned to the structure in line with the initial strength assignments. To test the
design methodology, two sets of 4-, 8,- 12-, 16- and 20-storey reinforced concrete struc-
tures are designed. The rst set considers frame-wall structures in which the frames are
parallel to the walls and the second considers structures in which link-beams connect
from the frames directly onto the ends of the walls. A suite of time-history analyses are
conducted to validate the methodology, which is seen to perform excellently.
Keywords: Displacement-based design; frame wall; dual system; seismic design.
1. Introduction
This paper presents guidelines for the direct displacement-based design (DBD) of
structures that utilise both frames and walls to resist earthquake actions in parallel.
There is a need for a design methodology that is applicable to this particular form
of structure, commonly known as a frame-wall structure or dual system structure,
because the dynamic behaviour of dual systems is considerably dierent from pure
frame or wall structures for which many design recommendations already exist.
Such dierences in dynamic behaviour are attributed principally to the interaction
that takes place between the frames and walls, which is not well accounted for in
current design practice.
A further motivation for this work stems from the consideration that the com-
bined structural form is a very ecient and convenient way to resist earthquake
actions that is not currently being widely exploited. It can be argued from both
structural and aesthetic points of view that the combination of frames and walls
presents considerable advantage over structures formed purely out of frames or
walls. The sti nature of cantilever reinforced concrete (RC) walls means that they
are naturally suited to control storey-drifts in the lower levels of buildings. In con-
trast, frames typically restrain deformation in upper storeys but moreover, they
91
April 28, 2006 13:47 WSPC/124-JEE 00274
92 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
oer signicant energy dissipation up the height of the building which reduces
the total displacements that a building experiences. From aesthetic and functional
points of view, frames enable large open spaces within minimum constraints on
usage. On the other hand, walls are an attractive means of forming stair wells and
lift shafts in a building, while at boundary lines they are commonly used to provide
re resistance between buildings.
2. Challenges for the Direct DBD of Frame-Walls
An ideal seismic design procedure will establish the minimum basic strength of a
structure sucient to ensure pre-dened performance criteria for the building are
satised at the design ground motion intensity, with a minimum of eort. Previous
work by Sullivan et al. [2005] investigated a trial methodology which provided
encouraging results when applied to regular frame-wall structures in which the
frames were parallel to the walls. The research identied that the following two
tasks were required to improve the accuracy of the methodology and thereby enable
the Direct DBD [Priestley, 2003] approach to be used for frame-wall structures:
Development of an expression for the displaced shape of frame-wall structures at
maximum response, to enable equivalent SDOF characteristics to be established.
Development of an expression for the equivalent SDOF system ductility or equiv-
alent viscous damping that takes into account the frame-wall interaction.
Sullivan et al. [2005] proposed that the design displacement prole be set as a func-
tion of the moment prole in the walls, using proportions of strength assigned at
the start of the design procedure. There is experimental evidence that supports the
validity of this approach as reported by Sullivan et al. [2004]. Another recommen-
dation made by Sullivan et al. [2005] was that the equivalent SDOF system viscous
damping could be obtained by factoring the individual frame and wall components
by the proportions of overturning they resist.
The challenge in this paper is therefore to nalise the design procedure proposed
by Sullivan et al. [2005] and to verify its accuracy through examination of a range
of case study structures.
3. Description of the New Design Procedure
The various steps of the seismic design procedure for frame-wall structures are
shown as a owchart in Fig. 1. The rst set of steps aims to develop an equiva-
lent SDOF representation of the MDOF structure. This is achieved by assigning
strength proportions and subsequently using the moment prole in the walls to
set a design displaced shape. With knowledge of the displacement prole, various
equivalent SDOF properties of the structure are obtained. The second important
set of steps in the design aim to determine the required eective period and then
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 93
Assign strength proportions to frames and walls.
Determine wall inflection height, h
inf
.
Determine yield displacements of
walls and yield drift of frames.
Calculate design displacement profile.
Determine effective height, h
e
, effective mass, m
e
,
and design displacement,
d
.
Plot displacement spectra at system damping level and use design
displacement to obtain required effective period, T
e
. Check, T
e
= T
e,trial
?
Determine equivalent viscous damping
values for frames and walls.
Determine effective stiffness and design base shear, V
b
= K
e d
.
Obtain beam & column
strengths by factoring strength
proportions by base shear.
Reduce
drift
limit.
Calculate the ductility demands on the frames and
walls. Are ductility demands excessive?
NO
YES
Use proportions of overturning moment resisted by the frames and walls to
factor damping values & obtain an equivalent system damping value
sys
.
Choose a trial
effective period,
T
e,trial
.
Reset
T
e,trial
= T
e
YES
Distribute base shear up height in proportion to
displacements of masses. Subtract frame shears from
total shears to obtain wall shears & thereby moments.
Perform capacity design with allowance for higher mode effects, to obtain design
strengths in non-yielding elements and design shears in frames and walls.
NO
Fig. 1. Flowchart of recommended design procedure for frames-wall structures.
stiness using the substitute structure approach [Gulkan and Sozen, 1974; Shibata
and Sozen, 1976]. The design base shear is obtained through multiplication of the
necessary eective stiness by the design displacement and the strength of individ-
ual structural elements is set taking care to ensure that initial strength assignments
are maintained. Each of these phases is described in more detail in the sub-sections
that follow.
April 28, 2006 13:47 WSPC/124-JEE 00274
94 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
3.1.1. Assignment of strength proportions to establish the wall
inection height
In order to develop a suciently accurate SDOF representation of the frame-wall
structure, strength proportions are assigned at the very start of the design proce-
dure. This involves setting the proportion of base shear or overturning resistance
oered by the frames and walls, in addition to the relative strength distribution
of yielding elements (beams and ground storey columns) within the frames. As
mentioned above, by assigning these strength proportions the shear and moment
prole in the walls can be established and this then enables determination of the
inection height. Figure 2 locates the inection height for a frame-wall structure in
which the frames and walls resist the total base shear in equal proportions and the
frames provide a constant shear resistance over their height. The inection height
is of particular interest as it will be used to form the design displacement prole.
Note that the proportions of strength assigned at this stage of the design process
are related to the forces expected at formation of a 1st mode plastic mechanism.
They should not be confused with the proportions of force that are expected to
develop at maximum response. The maximum forces are aected by overstrength
and higher mode eects and are established following DBD as part of a capacity
design procedure.
The storey shear above the base of the walls cannot be obtained directly from the
design base shear since the walls remain elastic above the ground storey and upper
storey shears will depend on the proportion of shear carried by the frames. As such,
wall shears are obtained as the dierence between the total shear and the frame
shear as shown in Eq. (1). Recall that the frame storey shear can be determined
since it is dependent only on the strength of the beams up the building height.
V
i,wall
V
b
=
V
i,total
V
b

V
i,frame
V
b
, (1)
where V
b
is the total base shear, V
i,wall
is the wall shear at level i, V
i,total
is the
total shear at level i, and V
i,frame
is the frame shear at level i.
Total shear
(solid line)
Frame shear
(dashed line)
Wall shear
(shaded area)
MWall
Wall BMD
Frame
overturning
Wall
inf lection
height, h
i nf
MOMENTS SHEARS BEAM-SWAY MECHANISM
1.0Vb 0.5Vb
Fig. 2. Use of frame-wall strength proportions to locate inection height in walls.
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 95
For the purpose of establishing the inection height, a triangular distribution
of the fundamental mode inertia forces up the height of the structure is assumed.
This approximation enables the total storey shear to be obtained as a function of
the base shear as shown in Eq. (2).
V
i,total
V
b
= 1
i
n
(i 1)
(n + 1)
, (2)
where V
i,total
is the total shear at level i, V
b
is the total base shear, and n is the
total number of storeys in the building.
As Eq. (2) provides the distribution of total storey shear up the building height,
the only unknown of Eq. (1) is the frame storey shear distribution. To obtain this
shear proportion, the relative strength distribution of yielding elements within the
frames is used.
Although the designer is free to choose any strength distribution they prefer, it
is proposed that the use of beams of equal strength up the height of the structure
is advantageous for design and construction. Assuming that beam moments are
carried equally by columns above and below a beam-column joint, the frame storey
shear is obtained as a function of the beam strength using Eq. (3).
V
i,frame
=
(

M
b,i
+

M
b,i1
)
2(h
i
h
i1
)
=

M
b,i
h
col
, (3)
where V
i,frame
is the frame shear at level i, M
b,i
are the beam strengths at level i,
and h
col
, is the inter-storey height. Although the beam strengths are not actually
known to begin with, Eq. (3) is useful as it indicates that provided beams of equal
strength are to be used then the frame storey shear is constant up the building
height. Consequently, if 40% of the base shear is being carried by the frames, this
40%V
b
will be carried up the entire height of the frame. As such, the shear propor-
tion carried by the frame can be substituted into Eq. (1) and the wall shears and
bending calculated, all as a function of the design base shear.
A perfectly constant shear up the height of the frame requires that the sums
of the base column strengths and roof beam strengths are both equal to half the
sum of the intermediate level beam strengths. If roof level beams are assigned
strength equal to those on other stories, then the frame shear at roof level should
be considered to be 50% greater than that at other levels. Larger base column
strengths will also imply larger ground storey shears, with the column inection
height shifting above 0.5h
col
.
The storey shear and consequently the moment in the walls are used to establish
the inection height in the walls, h
inf
, where the moment and curvature is zero. This
inection height will be used to nd the displacements of the structure at yield of
the walls and to develop the design displacement prole, as detailed in the next
subsections.
Other important design quantities that should be obtained from the strength
assignments are the proportion of overturning resisted by the frames and walls
respectively. The proportions of overturning can be obtained directly from the shear
April 28, 2006 13:47 WSPC/124-JEE 00274
96 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
prole up the height of the structures. These overturning proportions are used later
in the design procedure for denition of the system damping and for adjustment of
the design drift to allow for higher modes.
3.1.2. Yield deformations of the walls and frames
As the walls tend to control the response of frame-wall structures, the wall yield
curvature and displacements at yield are important for the development of the
design displacement prole. The frame yield displacement, or yield storey drift, is
also important to the design process as it is used to provide an indication of the
energy absorbed through hysteretic response of the frame.
The yield curvature of the walls,
yWall
, is rstly obtained using Eq. (4)
[Priestley, 2003].

yWall
=
2
y
L
w
, (4)
where
y
is the yield strain of the longitudinal reinforcement in the wall and L
w
is
the wall length.
The displacement prole of the structure at yield of the wall,
i,y
, can then
be established using the wall yield curvature, inection height and storey height in
accordance with Eqs. (5a) and (5b).

iy
=

yWall
h
inf
h
i
2


yWall
h
2
inf
6
for h
i
h
inf
, (5a)

iy
=

yWall
h
2
i
2


yWall
h
3
i
6h
inf
for h
i
< h
inf
. (5b)
The frame yield drift,
yframe
, used to estimate the ductility and equivalent viscous
damping of the frames, is obtained in accordance with Eq. (6) [Priestley, 2003]:

y
frame
=
0.5l
b

y
h
b
, (6)
where l
b
is the average beam length,
y
is the yield strain of beam longitudinal
reinforcement and h
b
is the average depth of the beams at the level of interest.
3.1.3. Design displacement prole and equivalent SDOF characteristics
The design displacement prole is developed using the various values obtained in
the preceding subsections, together with the design storey drift, as shown in Eq. (7).

i
=
iy
+
_


yWall
h
inf
2
_
h
i
, (7)
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 97
where
i
is the design displacement for level i,
i,y
is the displacement of level
i at yield of the walls,
d
is the design storey drift,
yWall
is the yield curvature
of the walls, h
inf
is the inection height, and h
i
is the height at level i. Note that
the design storey drift can be initially taken as the code limit for non-structural
damage, reduced to allow for higher mode eects in accordance with Eq. (8).

d
=
d,limit
_
1
(N 5)
100
_
M
OT,frame
M
OT,total
+ 0.25
__

d,limit
, (8)
where N is the number of stories, M
OT,frame
is the overturning resistance of the
frame and M
OT,total
is the total overturning resistance of the structure. This approx-
imate equation was proposed after reviewing the results of initial trial case studies
[refer to Sullivan, 2005]. As mentioned earlier, the ratio of frame to total overturning
resistance can be obtained in terms of the base shear using the strength assignments
made at the start of the design procedure. The design drift given by Eq. (8) may
be reduced further if it is found that inelastic demands on the structure are likely
to be excessive. Alternatively, the critical value of storey drift can be determined
before the design displacement prole is developed.
With knowledge of the displacement prole at maximum response;
i
, the seis-
mic masses; m
i
, and storey heights; h
i
, the equivalent SDOF design displacement;

d
, eective mass; m
e
, and eective height; h
e
, can be calculated as shown in
Eqs. (9) to (11) [Priestley, 2003] respectively.

d
=

n
i=1
(m
i

2
i
)

n
i=1
(m
i

i
)
, (9)
m
e
=

n
i=1
(m
i

i
)

d
, (10)
h
e
=

n
i=1
(m
i

i
h
i
)

n
i=1
(m
i

i
)
. (11)
3.1.4. Design ductility values, eective period and equivalent viscous damping
The only other substitute structure characteristic required for Direct DBD
[Priestley and Kowalsky, 2000] is the equivalent viscous damping. This is a function
of ductility and according to recent recommendations by [Blandon and Priestley,
2005] and [Grant et al., 2005], the eective period.
The ductility demands on the walls for use within this equivalent viscous damp-
ing approach should be calculated using displacement at the eective height. The
wall ductility demand,
wall
, is therefore simply the design displacement divided
by the yield displacement of the walls at the eective height, as shown in Eq. (12).

wall
=

d

he,y
, (12)
April 28, 2006 13:47 WSPC/124-JEE 00274
98 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
where
d
is the design displacement (from Eq. (9)) and
he,y
is the yield displace-
ment of the wall at the eective height (obtained substituting the eective height
into the appropriate version of Eq. (5)).
The displacement ductility demands on the frames at each level up the height
of the structure can be obtained using the storey drifts as shown in Eq. (13).

frame,i
=
_

i1
h
i
h
i1
_
1

y
frame
, (13)
where
i
,
i1
, h
i
, and h
i1
, are the displacements and heights at level i and level
i 1 respectively,
frame,i
is the frame ductility at level i, and
yframe
is the yield
drift of the frame (from Eq. (6)). When beams of equal strength are used up the
height of the structure, the ductility obtained from Eq. (13) for each storey can be
averaged to give the frame displacement ductility demand.
Before proceeding with calculations of the equivalent viscous damping, it is
necessary to check that the ductility demands are sustainable. Ductility demands
on frames are typically not critical as the walls tend to have smaller yield curvatures
and yield displacements. For frame-wall structures in which frames are parallel to
walls, ductility demands will be fairly low and can typically be detailed for relatively
easily. However, when link-beams connect between frames and walls then these link-
beams are likely to be subject to higher curvatures than other beams and should be
checked separately, as is discussed Sec. 4 where the procedure is applied to various
case study structures.
Although the wall displacement ductility demand indicated by Eq. (12) is appro-
priate for estimation of the equivalent viscous damping, it is not a good representa-
tion of the inelastic deformation that the walls must undergo. A more appropriate
parameter is the wall curvature ductility,
wall
, which can be obtained in accor-
dance with Eq. (14).

wall
= 1 +
1
L
p

yWall
_


yWall
h
inf
2
_
, (14)
where L
p
is the wall plastic hinge length,
d
is the design storey drift,
yWall
is
the yield curvature of the walls and h
inf
is the inection height. Note that because
the curvature ductility demand is a function of the inection height and not the
total height, inelastic deformation demands in walls of frame-wall structures will
typically be larger than those in plane wall structures.
The wall plastic hinge lengths to be used within Eq. (14) are taken as the
minimum of Eqs. (15a) and (15b).
L
p
= 0.022f
y
d
b
+ 0.054h
inf
, (15a)
L
p
= 0.2L
w
+ 0.03h
inf
, (15b)
where f
y
is the yield stress and d
b
the diameter of the longitudinal reinforcement
in the wall, L
w
is the wall length and h
inf
is the inection height. These equations
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 99
have been adapted from [Priestley, 2003] with the inection height substituting the
total height.
The curvature ductility capacity of a RC wall will depend on the strain limits
selected for the concrete in compression (
c
) and longitudinal reinforcement in ten-
sion (
s
). For reasonably conservative values of
c
= 0.018 and
s
= 0.06, Priestley
and Kowalsky [1998] found that the ultimate curvature of reinforced concrete walls
is well represented by Eq. (16).

u
=
0.072
L
w
, (16)
where
u
is the ultimate curvature and L
w
is the wall length. This equation was
shown to be representative of ultimate curvature over a range of axial load ratios
and longitudinal reinforcement contents. Combining Eqs. (16) and (4), it is found
that the curvature ductility capacity is approximately equal to 0.036/
y
.
If the checks on ductility indicate that the inelastic deformations associated with
the design drift will be excessive then the design drift must be reduced and the
design displacement prole re-computed as discussed in the previous sub-section.
If the ductility demands are sustainable then the next step in the design procedure
is to compute equivalent viscous damping values.
Recent work by Blandon and Priestley [2005] (developed further by Grant et al.
[2005]), recommends that the equivalent viscous damping be computed as a function
of the eective period. As this is unknown at the start of the design process, a trial
value can be used and an iterative design process adopted. A reasonable estimate
for the trial value of the eective period can be obtained from Eq. (17).
T
e,trial
=
N
6

sys
, (17)
where N is the total number of stories and
sys
is the system ductility. Equation (17)
is similar in form to a code based equation that uses the height or number of storeys
to estimate the initial period. The ductility term accounts for the dierence between
the initial and eective periods, neglecting the eect of strain hardening. Given the
approximate nature of Eq. (17) [refer Sullivan, 2005] trial eective period values
may be some 30% dierent than the nal eective period, however by using such
a trial value, it will be found that convergence is attained within one or, at most,
two iterations.
Having set the trial eective period and established expected ductility values,
the frame and wall equivalent viscous damping components are calculated using
Eqs. (18) and (19) respectively [Grant et al., 2005].

hyst,wall
=
95
1.3
_
1
1

0.5
wall
0.1r
wall
__
1 +
1
(T
e,trial
+ 0.85)
4
_
, (18)

hyst,frame
=
120
1.3
_
1
1

0.5
frame
0.1r
frame
_
_
1 +
1
(T
e,trial
+ 0.85)
4
_
, (19)
April 28, 2006 13:47 WSPC/124-JEE 00274
100 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
where r is the post-elastic stiness coecient, typically taken as 0.05 for new RC
structures. Note that by considering the inuence of the period on the damping
values, it could be argued that the period-dependence of the damping values can
be neglected when eective periods are greater than 1.0 s, which is usually the case
for frame-wall structures. The equivalent viscous damping for the frames and walls
is obtained adding the elastic and hysteretic components together and then a value
of damping for the equivalent SDOF system is determined using Eq. (20).

SDOF
=
M
wall

wall
+M
OT,frame

frame
M
wall
+M
OT,frame
, (20)
whereM
OT,frame
is the overturning resistance of the frames and M
Wall
is the over-
turning resistance (exural strength) of the walls. At this point of the design process,
all of the substitute structure characteristics have been established and as such, the
displacement spectrum is developed at the design level of damping. This can be
done using a damping-dependent scaling factor appropriate for the seismological
characteristics of the design region. The Eurocode 8 [CEN, 1998] recommends that
the value obtained from Eq. (21) be used to scale the elastic spectrum to the
damping level of interest.
=
_
10/(5 +
SDOF
) 0.55, (21)
where
SDOF
is the equivalent viscous damping of the system as given by Eq. (20).
The design displacement is then used to read o (or interpolate between known
points) the required eective period, T
e
, as shown in Fig. 3.
The eective period obtained from the Direct DBD process illustrated in Fig. 3 is
then compared to the trial eective period value. If the period values do not match,
then the period obtained from Fig. 3 replaces the trial period and the design step
0.00
0.20
0.40

0.60
0.80
1.00
0.0 1.0 2.0 3.0 4.0 5.0
Period (s)
S
p
e
c
t
r
a
l

D
i
s
p
l
a
c
e
m
e
n
t
a
t

S
D
O
F

d
a
m
p
i
n
g

(
m
)
T
e
d
Displacement spectrum at
system damping level.
Fig. 3. Direct displacement based design to obtain the required eective period.
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 101
is repeated. When eective periods nally match, the designer is in a position to
determine the eective stiness and design base shear as outlined next.
3.1.5. Determining the design base shear and member strengths
With the eective period established, the eective stiness, K
e
, is determined in
accordance with Eq. (22).
K
e
= 4
2
m
e
T
2
e
, (22)
where m
e
is the eective mass (from Eq. (10)) and T
e
is the eective period. This
eective stiness is then multiplied by the design displacement,
d
, to obtain the
base shear, V
b
, as shown by Eq. (23).
V
b
= K
e

d
. (23)
Individual member strengths are then determined maintaining the strength propor-
tions assigned at the start of the design process. Note however, that rather than
use a triangular lateral force distribution, better results are obtained distributing
the base shear up the height of the structure according to Eq. (24).
F
i
=
m
i

N
i=1
m
i

i
V
b
, (24)
where F
i
is the portion of base shear applied at level i, m
i
is the mass at level i,
and
i
the displacement at level i.
This then completes the DBD process. It is evident that there are several steps
to the design procedure, however, the process is simple and does provide excel-
lent control of displacements and storey drifts as is demonstrated in the following
section.
4. Verication of the Design Method
The design method is veried through examination of several case studies. A range
of frame-wall structures are designed using the new procedure with the aim of
maintaining storey drift and curvature ductility limits typical of a life-safety per-
formance level. The design strengths obtained for each case study are then used
to set the strength of accurate non-linear analytical models which are subject to a
series of time-history analyses using earthquake records compatible with the design
spectrum. The success of the design process is gauged by comparing the target
deformations anticipated during the design phase, with the actual deformations as
predicted by the time-history analyses.
4.1. Description of case study structures
The frame-wall structures shown in plan in Fig. 4 and elevation in Fig. 5 are
designed using the new procedure. Two types of structure are considered; (i) struc-
tures with walls and frames connected only by oor slabs, and (ii) structures with
April 28, 2006 13:47 WSPC/124-JEE 00274
102 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
RC Walls
8m long w. L-Beams
10m long w/o L-Beams
8m 8m
20m
Structures with & without Link-Beams
56m
8m 8m
EQ
Excitation
Direction
EQ
Excitation
Direction
Fig. 4. Plan view of frame-wall structures being examined in the verication of the proposed
DBD method.
(i) Eight-storey structure without link-beams (ii) Eight-storey structure with link-beams
Fig. 5. Elevation of frame-wall structures; (i) without link-beams and (ii) with link beams, to be
examined as part of nal verication of the DBD method.
link-beams extending from frames directly to the ends of the walls. In order to
comprehensively test the approach, buildings of 4, 8, 12, 16 and 20 storeys are
examined.
These case study structures are regular in layout with a RC frame-wall system
being used to resist lateral loads acting along the longitudinal axis of the building.
In the transverse direction it is assumed that a regular arrangement of RC walls
would be used to resist lateral loads, however, this does not aect the design pro-
cedure proposed here for the frame-wall system. The structures are considered as
having rigid foundations with oor slabs that act as rigid-diaphragms in plane, fully
exible out of plane. It has been shown elsewhere [Sullivan, 2005] that for frame-
wall structures of typical layout diaphragm exibility does not require consideration
within the design procedure.
The concrete and reinforcement material properties assumed for the struc-
tures are values that could typically be found in building practice. Values for
the concrete include: (i) f

c
= 30.0 MPa and (ii) E
c
= 25 740 MPa. The expected
strengths adopted for the reinforcing steel include: (i) f
y
= 400 MPa and (ii) E
s
=
200 000 MPa. For seismic design, material strengths are not factored to dependable
strength levels and instead these values have been taken as the expected strength
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 103
and stiness characteristics. The seismic weights of individual oors have been esti-
mated assuming a concrete density value of 24.5 kN/m
3
, a superimposed dead-load
of 1.0 kPa, a reduced live-load of 1.0 kPa and a loaded oor area of 1105 m
2
per
level. Axial load ratios have been computed using these oor weights factored by
the tributary area of oor supported by the individual elements. Floor weights, axial
load ratios and dimensions of individual elements are presented for the structures
without link-beams in Table 1 and for the structures with link-beams in Table 2.
Axial load ratios shown are for the elements at the ground storey of the buildings.
Case study structures with link-beams are being examined in this work to ensure
that the design procedure performs adequately for this peculiar form of frame-wall
structure. The interaction between the frames and walls of structures with link-
beams is more signicant than in the classical form of frame-wall structure in which
the frames are parallel to the walls. As the walls deform their ends either lift or
drop, depending on whether the bending in the wall puts that part of the wall in
compression or tension, as illustrated in Fig. 6. Additional curvatures are imposed
on the link beams due to the change in elevation of the wall ends. The magnitude
of these curvatures can be gauged taking the shift in elevation of the wall edge and
dividing by the beam length, which gives the equivalent chord rotation imposed on
the link beams.
Table 1. Characteristics of frame-wall structures without link-beams, examined in the veri-
cation of the proposed DBD method.
4 storey 8 storey 12 storey 16 storey 20 storey
Wall length (mm) 8000 10 000 10 000 10 000 10 000
Wall thickness (mm) 350 350 350 350 350
Beam depth width (mm) 750 450 750 450 750 450 750 550 750 550
Int. column depth width (mm) 750 600 750 600 750 600 750 600 800 650
Ext. column depth width (mm) 600 600 600 600 600 600 600 600 650 650
Inter-storey height (mm) 3600 3600 3600 3600 3600
Wall axial load ratio 0.021 0.040 0.060 0.080 0.101
Int. column axial load ratio 0.089 0.178 0.267 0.374 0.412
Ext. column axial load ratio 0.060 0.120 0.180 0.251 0.276
Floor seismic weight (kN) 11400 11800 11800 11900 12000
Table 2. Characteristics of frame-wall structures with link-beams, examined in the verication
of the proposed DBD method.
4 storey 8 storey 12 storey 16 storey 20 storey
Wall length (mm) 8000 8000 8000 8000 8000
Wall thickness (mm) 350 350 350 350 350
Beam depth width (mm) 750 450 750 450 750 450 750 450 750 450
Int. column depth width (mm) 600 600 600 600 600 600 650 600 750 750
Ext. column depth width (mm) 600 600 600 600 600 600 600 600 650 650
Inter-storey height (mm) 3600 3600 3600 3600 3600
Wall axial load ratio 0.021 0.042 0.063 0.084 0.105
Int. column axial load ratio 0.108 0.217 0.325 0.404 0.386
Ext. column axial load ratio 0.060 0.120 0.180 0.240 0.276
Floor seismic weight (kN) 11600 11600 11600 11600 11900
April 28, 2006 13:47 WSPC/124-JEE 00274
104 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
High curvature
ductility demands
expected at beam-
wall interface.
Assumed
NAD of wall
Wall edge lifts
Wall edge drops
Fig. 6. Illustration of high curvature ductility demands expected at ends of link-beams.
The walls are also aected by the link-beams since the moment and shear from
each beam must be carried by the walls. The link beam moments can change the wall
moment prole signicantly as will be seen in later sections, whereas the shears may
aect the axial load on the walls. For the frame-wall structures shown in Fig. 5 the
wall axial loads are not aected by the link-beam shears which apply equal shears
(owing to their equal strength) in opposing directions on either side of the wall and
therefore cancel each other out. The moments however will need to be accounted
for as these tend to sum together at the wall centreline and can reduce the wall
inection height, which in turn aects the design displacement prole. Specic
recommendations that account for the inuence of link beams will be presented in
a later section.
4.2. Design criteria
A design storey drift of 2.5% was selected for the design of the case studies. In
seismic design codes (e.g. NZS1170.5:2004 [2004]) this storey drift limit is commonly
associated with a life-safety performance level. A design spectrum was selected to
match a set of accelerograms available as shown in Fig. 7. The design spectrum can
be selected in this arbitrary manner for these case studies since the design method
should be applicable to any spectral shape and its applicability is not restricted to
a particular code.
The design displacement spectra at levels of damping greater than 5% were
observed to vary by a factor of , where is given by Eq. (25).
=
_
6/(1 +). (25)
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 105
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 1 2 3 4 5
Period (s)
5
%

D
a
m
p
e
d

A
c
c
e
l
e
r
a
t
i
o
n

(
g
)
Avg Sa 5%
Design Spectrum
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0 1 2 3 4 5 6 7
Period(s)
S
p
e
c
t
r
a
l

D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
Avg. 5%
Avg. 10%
Avg. 15%
Avg. 20%
Sd (5%)
Sd (10%)
Sd (15%)
Sd (20%)
Fig. 7. Case study design acceleration spectrum (left) and displacement spectra (right) at dif-
ferent levels of viscous damping, compared with the average spectra of selected accelerograms.
For a given increment of damping this equation reduces the displacement spec-
tra signicantly more than the more realistic Eurocode 8 [CEN, 1998] equation
presented in Eq. (21). It is not proposed that Eq. (25) should be used in place of
the Eurocode equation in normal circumstances. However, Eq. (25) does provide
the best representation of the accelerograms used in this study and therefore it is
used here only in order to obtain the most valid verication of the design procedure.
Fig. 7 shows that the factor from Eq. (25) provides good correlation between the
design displacement spectra and the accelerograms at damping levels of 10%, 15%
and 20%.
The design storey drift limit of 2.5% is intended to control damage of non-
structural items in the buildings. Damage to structural items was controlled by
imposing strain limits on the concrete and reinforcing. Ultimate compressive strains
of 0.018 for the concrete and 0.06 for the reinforcing steel were deemed appropriate
for these case studies. Priestley and Kowalsky [1998] have argued that these strain
limits are reasonably conservative estimates for well-conned concrete and well-
restrained reinforcement as results from detailing to the requirements of NZS3101
[1995]. Priestley and Kowalsky [1998] observed that the ultimate curvature ductility
of a wall,

, is well represented by Eq. (26), in which


y
is the yield strain of the
longitudinal reinforcement in the wall.

=
0.072
2
y
. (26)
For the material properties being used in these case studies, Eq. (26) suggested that
the curvature ductility of the walls should be limited to a value of 18.0. Therefore,
if structural deformations associated with the storey-drift limit impose curvature
ductility demands greater than 18.0, the design storey-drift should be reduced until
the ductility limit is satised. It will be shown that for these case studies the design
drift had to be reduced for both of the eight-storey structures in order to satisfy
the curvature ductility limit.
April 28, 2006 13:47 WSPC/124-JEE 00274
106 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
For the taller case studies the design drift was also reduced from the limit of
2.5% in order to control the deformations caused by higher modes. This was done
because results from the initial set of case studies reported by Sullivan et al. [2006],
indicated that despite the fact that the 1st mode controlled the displacements of
the structures, higher modes could increase storey drifts signicantly. This was
especially evident for taller structures. The reduction in storey drift was made
using Eq. (8).
Another control on the design of these case studies has been imposed to main-
tain realistic reinforcement contents and axial load ratios. Column dimensions were
set initially to be 600 600 mm square. These dimensions were then increased if
necessary, to limit axial load ratios (N/f

c
A
g
) to a maximum of 0.40. However, col-
umn dimensions are also inuenced by the necessary strength. In these case studies
the building layout was such that the limit on axial load ratio only aected the
dimensions of the interior columns of the taller structures with link beams. Axial
load ratios on the walls were not of concern in this set of case studies owing to
the large area of the walls. Longitudinal reinforcement ratios in the walls were of
more importance, and maximum and minimum longitudinal reinforcement ratios
were set at 1.6% and 0.3% respectively. For the columns, maximum and minimum
longitudinal reinforcement ratios were set at 3.0% and 0.5% respectively, while for
the beams, tension reinforcement limits of 1.5% and 0.35% were maintained.
4.3. Details of the design
The procedure for the design of frame-wall structures summarised in the owchart
presented in Fig. 1 has been used to design the case study structures. Rather than
describe the design steps therefore, this section identies the strength assignments
that were made for these case studies, provides recommendations specic to the
design of frame-wall structures with link-beams, and presents intermediate and
nal design results.
4.3.1. Strength assignments
The proportions of total shear resisted by the frames and the walls were assigned
arbitrarily to begin with, however, it was observed that by altering the shear pro-
portions the design could be improved. For example, an initial strength assignment
that assumes the frames will carry a large fraction of the lateral load is likely to
result in a low inection height. Having a low inection height implies that the
design displacement and damping is maximised and the minimum possible base
shear is obtained. However, a low inection height will impose large curvature
ductility demands on the walls, and if these are excessive then the design storey
drift should be reduced or a larger strength proportion assigned to the walls. By
assuming a large proportion of the shear is given to the walls then the opposite
will occur, the consequences being that the design may not be very ecient, with
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 107
heavily reinforced walls and poorly utilised beams and columns possessing only
minimum reinforcement contents. The proportions of base shear assigned to the
frames and walls are presented with the intermediate design results later in this
section.
Equal strengths are assigned to the beams up the height of the frames as this
is an attractive solution for construction. However, to avoid spikes in frame storey
shear at the top level of the structures, these case studies set the strength of the
top storey (roof) beams equal to half that of the other beams. The base column
strengths were set to be fractionally larger than the beam strengths to provide an
inection height of 0.6 times the storey height. This design choice was made to
provide some protection against column hinging at the top of the rst storey. Pro-
tection against column hinging is necessary in frame structures to avoid formation
of a soft-storey mechanism. However, in frame-wall structures this provision is not
necessary because the cantilevering walls will protect against soft-storey mecha-
nisms [refer to Paulay and Goodsir, 1986]. Nevertheless, the large column strength
is not unrealistic and therefore this strength assignment was maintained.
4.3.2. Design recommendations for frame-wall structures with link-beams
As stated earlier, frame-wall structures with link-beams possess peculiar character-
istics that must be allowed for in design. One of the rst adjustments that must be
made when link-beams exist, is to alter the wall moment prole associated with the
1st mode wall shears to account for the moments transferred from the link-beams.
Having decided on the strength assignments for the frame-wall system, the beam
strengths can be established as a fraction of the total design base shear. For the
strength assignments used for these case studies, the sum of the beam strengths,

M
b
, at a given level, i, is given by Eq. (27).

M
b
i
=
V
i,frame
h
col
_
1 +
d
col
L
b
_ , (27)
where V
i,frame
is the frame shear (known as a fraction of the total design base
shear), h
col
is the storey height at level i, d
col
is the depth of the columns and L
b
is
the beam length (between column faces). The beam strengths in this equation refer
to the strength at the face of the columns which have been projected to the column
centrelines using the d
col
on L
b
ratio. For simplicity, these case studies neglect the
eects of beam-column joints and assume that the beam strengths develop at the
column centrelines. This simplication implies that the d
col
on L
b
term drops out
of Eq. (27).
The strength of a single beam is obtained using Eq. (28), in which the sum of
the beam moments on the oor are divided by the number of beam ends, n
bj
, that
connect to beam-column joints. As the frame shear used in Eq. (27) is equal to
April 28, 2006 13:47 WSPC/124-JEE 00274
108 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
the sum of the column shears, the number of beam ends that connect to the walls
should not be included within n
bj
.
M
b
=

M
b
i
n
bj
. (28)
Since the link-beams will develop the same strength as given by Eq. (27) at the
edge of the wall, the moment transferred to the centre of the walls can be obtained
from the beam moments and geometry as shown in Fig. 8. Substituting Eq. (27)
into Eq. (28) and using the geometry and beam bending moment diagram presented
in Fig. 8, Eq. (29) is obtained for the moment transferred from a link-beam to the
wall centreline.
M
bWall
= V
i,frame
_
1 +
L
w
L
b
_
h
col
n
bj
_
1 +
d
col
L
b
_, (29)
where L
w
is the wall length and n
bj
is the number of beam ends connecting to
beam-column joints per link-beam. For these case studies the d
col
on L
b
term was
neglected for simplicity.
The moments transferred to the wall from the link-beams are used to adjust
the moment prole as shown for an eight-storey structure in Fig. 9. Using this
approach, the moment prole in the walls is known as a proportion of the design
base shear. This then allows the inection height to be determined and the design
can proceed as normal.
Another stage in the design process in which the inclusion of link-beams needs
to be accounted for is in determination of the frame displacement ductility. As
mentioned earlier, link-beams undergo larger plastic rotations than other beams at
the same level. In order to estimate the ductility demands on the link beams it is
worth reviewing how the ductility demands on a standard RC frame are established.
For a standard beam-column sub-assemblage, the yield drift for design is
obtained using Eq. (30) as recommended by Priestley [2003]. This is an approximate
expression developed by assuming that the columns and joints add respectively an
additional 40% and 25% of the displacement associated with the beams yielding in
L
w
/2 L
b
M
b,wall
M
b
d
col
Fig. 8. Illustration of bending moment transferred from link-beams to wall centrelines.
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 109
0
1
2
3
4
5
6
7
8
-5.00 0.00 5.00 10.00 15.00
Wall moments
for unit base shear
L
e
v
e
l
Moments from
shears only
Adjusted for
L-Beam
moments
Fig. 9. Wall moment prole of eight-storey structure, adjusted to allow for moments transferred
from link-beams.
exure, to the storey deformation. It also assumes that member shear deformations
add a further 10% to the yield drift.

y,beam
= (1.0 + 0.4 + 0.25 + 0.1) 0.283
y
_
l
b
h
b
_
= 0.5
y
_
l
b
h
b
_
, (30)
where
y
is the yield strain of the longitudinal reinforcement in the beams, h
b
is
the depth of the beams and l
b
is the beam length.
For a beam-wall assemblage it could be assumed that the column and joint
deformation contributions can be neglected. This would imply that the factor of
0.5 in the yield drift equation of (30) reduces to 0.31. As a link-beam is supported
at one end by a sti wall and at the other end by a column, it is apparent that
an average factor of 0.4 can be used to approximate the yield drift of a link-beam,

y,link
, as shown in Eq. (31).

y,link
= 0.4
y
_
l
b
h
b
_
. (31)
The displacement ductility demands on the link-beams and other bays of the frame
can be obtained using Eqs. (30) and (31) respectively, together with the storey drift
associated with the design displacement prole. A weighted average ductility value,

frame,i
, for each oor is then obtained in proportion to the number of link-beams,
as shown in Eq. (32).

frame,i
=

D,i

y,link
n
link
+

D,i

y,beam
(n
b
n
link
)
n
b
, (32)
April 28, 2006 13:47 WSPC/124-JEE 00274
110 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
where
D,i
is the storey drift associated with the design displacement prole at level
i, n
link
is the number of link-beams in the storey, and n
b
is the total number of
beams on the storey.
Equation (32) is valid when beams have equal length and strength. If this is
not the case, it would be more appropriate to factor the ductility demands by the
beam shears. For these case studies, beams have equal strength and length at each
storey and up the full height of the building. Therefore, the frame ductility has
been obtained as the average of the storey ductility values obtained using Eq. (32).
Having determined the frame ductility, the design proceeds as normal with the
equivalent viscous damping determined in the same manner as for the standard
frame-wall structures.
4.3.3. Design results
Design was only conducted to the point that would allow the strengths of plastic
hinge regions to be set. With knowledge of these strength values, accurate nonlinear
models of the structures could be developed for verication of the design solutions
through time-history analysis as explained later in Sec. 4.4.
Intermediate design values for the structures with and without link-beams are
presented in Tables 4 and 3 respectively. Note that in the design of these case
studies an elastic damping component was rst obtained in accordance with the
recommendations of Priestley and Grant [2005] and then added to a hysteretic
component determined using the recommendations of Blandon and Priestley [2005].
This was because the recommendations of Grant et al. [2005] were not available at
the time of this work.
For these case studies, it can be seen that desirable design solutions were
obtained when the walls were assigned around 60% of the total design base shear.
Wall curvature ductility demands were fairly large in general and for the eight-storey
structures the design storey drift had to be reduced to ensure material strain limits
were not exceeded. Note that both wall displacement ductility and curvature duc-
tility demands are reported since the former relates more to the equivalent SDOF
representation of the structure (being calculated at the eective height) whereas
Table 3. Intermediate design results for the frame-wall structures without link-beams.
4 storey 8 storey 12 storey 16 storey 20 storey
% base shear assigned to walls 60% 60% 60% 50% 45%
Inection height 14.4 21.6 30.3 33.5 36.6
Design storey drift 2.09% 2.23% 2.37% 2.27% 2.16%
Design displacement 0.211 0.430 0.664 0.840 0.995
Wall displacement ductility 9.52 7.43 5.39 4.31 3.59
Average frame ductility 1.84 1.99 2.09 2.01 1.91
System ductility 6.27 5.00 3.86 2.97 2.51
System damping 15.1 15.0 14.8 13.9 13.2
Eective mass 3806 7479 10 932 14 510 18 121
Eective period 1.47 2.70 4.17 5.39 6.95
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 111
Table 4. Intermediate design results for the frame-wall structures with link-beams.
4 storey 8 storey 12 storey 16 storey 20 storey
% base shear assigned to walls 70% 70% 60% 50% 55%
Inection height 13.3 20.7 23.0 17.4 29.5
Design storey drift 2.03% 2.41% 2.34% 2.21% 2.13%
Design displacement 0.207 0.463 0.669 0.855 0.993
Wall displacement ductility 9.62 6.50 5.12 5.78 3.35
Average frame ductility 1.81 2.14 2.10 2.03 1.90
System ductility 6.35 4.74 3.58 3.22 2.70
System damping 16.0 16.1 15.7 15.4 14.6
Eective mass 3877 7334 10 788 14 409 18 033
Eective period 1.48 2.99 4.34 5.95 7.58
the latter better reects the amount of nonlinear deformation the wall would have
to undergo.
Table 4 indicates that frame displacement ductility demands were fairly low,
however, even with low ductility demands it was anticipated that the frames would
provide a signicant amount of hysteretic energy dissipation. The eective period
values shown in Table 4 are long, however they lie within the spectrum compat-
ible range of the accelerograms, suggesting that the time-history analyses to be
presented in later sections will provide a good test of the design solutions.
Final design strengths and longitudinal reinforcement contents for the struc-
tures with and without link-beams are presented in Tables 6 and 5 respectively.
Reinforcement contents for the walls and columns were obtained using axial loads
associated with the gravity actions only. In reality exterior columns would be subject
to a signicant variation in axial load during seismic response, however, dierences
in compression from one side of the building to the other suggest that the actual
strength of the columns should be equivalent to the sum of the strengths considering
gravity loads only.
An interesting observation to be taken from these design results is that the base
shear for the buildings of considerably dierent height is relatively constant. This
is attributed to the fact that the seismic weight per oor, the length of the walls
and the depths of the beams did not change for the dierent height structures.
Constant section dimensions implied that the system damping values for the design
Table 5. Final design strengths for the frame-wall structures without link-beams.
4 storey 8 storey 12 storey 16 storey 20 storey
Base shear (kN) 14 691 17 429 16 475 16 565 14 735
Wall strength (kNm) 31 949 72 827 109 235 117 367 114 931
Wall long. reinforcement % 0.65% 0.95% 1.55% 1.61% 1.42%
Beam strength (kNm) 1014.93 1195.67 1124.75 1419.52 1395.31
Beam reinforcement % 1.23% 1.50% 1.42% 1.38% 1.36%
Ext. column strength (kNm) 767 907 859 1087 1072
Int. column strength (kNm) 1533 1814 1717 2175 2145
Ext. col. long. reinforcement % 1.84% 1.91% 1.43% 1.99% 0.91%
Int. col. long. reinforcement % 2.23% 2.44% 1.88% 2.87% 1.66%
April 28, 2006 13:47 WSPC/124-JEE 00274
112 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
Table 6. Final design strengths for the frame-wall structures with link-beams.
4 storey 8 storey 12 storey 16 storey 20 storey
Base shear (kN) 14 490 14 980 15 120 13 742 12 292
Wall strength (kNm) 31 018 63 528 66 696 39 261 72 981
Wall long. reinforcement % 0.63% 1.46% 1.42% 0.42% 1.38%
Beam strength (kNm) 530.51 529.56 715.63 915.28 742.48
Beam reinforcement % 0.62% 0.62% 0.89% 1.11% 0.71%
Ext. column strength (kNm) 454 466 630 716 586
Int. column strength (kNm) 909 931 1260 1431 1172
Ext. col. long. reinforcement % 0.79% 0.50% 0.53% 0.52% 0.50%
Int. col. long. reinforcement % 1.85% 1.36% 2.22% 2.14% 0.50%
drift were fairly constant. On the contrary, the target displacement (and therefore
eective period) as well as the eective mass were almost linearly dependent on
height. Since this implies that the eective stiness is inversely proportional to the
height, and the base shear is simply the product of the eective stiness and the
design displacement, it is clear why fairly constant base shears were obtained.
The fact that the base shear may depend only on the oor mass and section
dimensions suggests that the design procedure could be signicantly simplied with-
out the loss of signicant accuracy. This is an item for future research.
Another important observation to be gleamed from the results in Tables 5 and 6
is that the longitudinal reinforcement contents are all within the minimum and
maximum limits specied as part of the design criteria. The reinforcement contents
on individual elements were seen to be sensitive to the strength assignments and
axial loads, and this point was used to make the design solution for each of the
structures more ecient. The fact that reasonable reinforcement contents have been
stipulated indicates that the design solutions are all realistic.
4.4. Design verication procedure
Nonlinear time-history analyses have been performed using the program Ruaumoko
[Carr, 2004] to assess the performance of the proposed methodology. Models of the
case studies were constructed in which the strengths of the beams and walls were
selected to match the design values obtained using the design methodology. The
models were subjected to seven articial acceleration records which had been used
to construct the design spectrum. The displacement spectra of the seven records
are shown at viscous damping levels of 5% and 15% in Fig. 10. The records can be
expected to test the design solutions well as they possess relatively small scatter
over a large range of periods.
Of the seven accelerograms selected, six were articial records. Record A5 is the
North-South component of the 1978 Tabas earthquake, recorded at the Boshrooy
station. The record was scaled in both magnitude (by a factor of 3.5) and duration
(by a factor of two) in order to represent a large earthquake that causes a linear
displacement spectrum up to a period of 5 s. Given this modication, none of the
accelerograms can be considered as real earthquake records. Criticisms directed
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 113
0.0
0.4
0.8
1.2
1.6
2.0
0 1 3 2 4 5 6 7
Period (s)
5
%

D
a
m
p
e
d

D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
Record A1
Record A2
Record A3
Record A4
Record A5
Record A6
Record A7
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 1 2 3 4 5 6 7
Period (s)
1
5
%

D
a
m
p
e
d

D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
Fig. 10. Displacement spectra of the seven accelerograms at 5% (left) and 15% (right) viscous
damping.
towards the use of articial accelerograms in seismic analyses usually focus on the
dierent phase content and duration of actual earthquakes compared to articial
records. However, the frame-wall structures being designed in this study are not
aected by duration owing to the fact that their strength does not degrade provided
that the design displacement is not signicantly exceeded. Little is known about
the inuence of the phase content on structural response. However, no evidence
has been found to suggest that any dierences in phase content aect structural
response. Furthermore, time-history analysis of structures using both articial and
real records that possess similar demand spectra, have indicated that frame-wall
structures respond similarly using either articial or real records. For these reasons
it was considered that the use of articial accelerograms was acceptable for these
case studies.
The success of the new methodology can be measured by comparing the actual
displacement response for the design level earthquake with the target displacement
prole selected in the design. If the analysis and target displacements and storey
drifts match, then the intended level of damage occurred and it can be concluded
that the objective of the design approach has been met.
4.4.1. Modelling structures for time-history analysis
In modelling the structures elastic properties (with reduced stiness to account for
cracking) were assigned to elements that are not intended to yield. This implies
that appropriate capacity design would have ensured that inelasticity is concen-
trated only in regions associated with the collapse mechanism. An o-shoot of this
modelling technique is that the analyses can be used to test capacity design guide-
lines which are reported elsewhere [Sullivan, 2005]. These case studies model the
oors as rigid links fully exible out of plane and P-delta eects are not considered
since no attempt to account for these eects was made in the design.
Beams, columns and walls were modelled using 2-hinge Giberson beam ele-
ments [Carr, 2004]. It is evident that the beams and the walls of the structures
April 28, 2006 13:47 WSPC/124-JEE 00274
114 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
without link beams are likely to carry a constant axial load throughout the seis-
mic response. Therefore the beams and walls of these structures could be modelled
without the use of beam-column elements which account for changes in strength
due to variations in axial load. The interior columns and the walls of the struc-
tures with link-beams were also subject to fairly constant axial loads as they are
anked by beams of equal strength and length on either side. However, the exterior
columns are subject to varying axial loads during the seismic response and therefore
modelling these members as beam-elements is not accurate. Nevertheless provided
that the gravity load in these columns is well below balance-point axial load then
the strength discrepancy in the compressive column roughly balances that of the
tension column. In addition, the exterior column strengths form a very small por-
tion of the total overturning resistance of the frame-wall structures. As such, the
use of beam-elements for all members was deemed acceptable for the verication
studies.
Rigid elements were used to model the connections between the walls and link-
beams as shown in Fig. 11. When reinforced concrete walls are deformed in exure
to the extent that they crack and later yield, the position of their neutral axis depth
shifts. Since a shift in neutral axis depth equates to a shift of the centre of rotation
in a section, the use of Giberson-beam elements up the centre of the walls may be
inaccurate. In particular, if a wall rotates about some point other its centre, then
this implies one side of the wall lifts or drops more than the other. This in turn
would imply that the curvature ductility demands on the beams would not be well
captured. However, a separate study [Sullivan, 2005] of two bre-element models
analysed in SeismoStruct [SeismoSoft, 2004] has shown that the curvature ductility
demands are only dierent over the lower stories of the building. Furthermore, the
overall dierence in energy dissipation between a model with shifting neutral-axis
depth and a model with constant neutral-axis depth is not signicant for these case
study structures because the discrepancies in beam curvature on either side of the
walls tend to cancel each other out.
Fig. 11. Illustration of model used to represent eight-storey frame-wall structure with link-beams.
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 115
The hysteretic behaviour of the concrete structures was represented using the
Takeda model [Otani, 1981], with 5% post-yield displacement stiness and the
unloading model of Emori and Schonbrich [1978]. Parameters for the Emori and
Schonbrich model included an unloading stiness factor of 0.5 for walls and columns
and 0.25 for beams, together with a reloading stiness factor of 0.0 and a reloading
power factor of 1.0 which were used for all the elements. Refer to the Ruaumoko
manual [Carr, 2004] for further details. The plastic hinge lengths associated with
the yielding elements were calculated using the recommendations from Paulay and
Priestley [1992].
The models developed in Ruaumoko use eective section properties up until
yield, obtained by taking the design strength and dividing by the yield curvature.
Approximations for yield curvature were obtained from expressions provided by
Priestley [2003]. The eective stiness for the ground storey columns at yield was
approximated using the strength under axial load from gravity only, divided by the
yield curvature. The elastic columns above the ground oor were modelled with the
same initial stiness.
Elastic damping was modelled for the structures using tangent stiness Rayleigh
damping with a 1st mode damping value set to provide the eect of 5% tangent sti-
ness damping for the MDOF structure, as recommended by Priestley and Grant
[2006]. Priestley and Grant [2006] provide an expression for this value that con-
siders the stiness and mass proportional components of the Rayleigh damping
equation. By specifying the same damping value at two dierent frequencies, where
the higher frequency is times the lower frequency, then the damping in the 1st
mode attributed to stiness proportional damping,
sp
, is given by Eq. (33). In
addition, the damping attributed to mass proportional damping,
mp
, in the 1st
mode is given by Eq. (34).

sp
=
1
+ 1
, (33)

mp
=

+ 1
. (34)
Having established the proportions of mass and stiness proportional damping,
Priestley and Grant [2006] recommend that for time-history analyses using Rayleigh
tangent stiness damping, the damping on the 1st mode,
1st
, should be set using
Eq. (35).

1st
= (
mp

0.75
+
sp
) 5%, (35)
where is the displacement ductility of the equivalent SDOF system.
The initial periods of the case-study structures were obtained from eigen-value
analysis, and the ratio between the 1st and 2nd mode frequencies were calculated to
give . These frequency ratios were then used together with the system displacement
ductility values to obtain the 1st mode elastic damping values for time-history
analysis shown in Table 7.
April 28, 2006 13:47 WSPC/124-JEE 00274
116 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
Table 7. 1st mode elastic viscous damping values for
time-history analysis.
T
1
(s) T
2
(s)
1st
4 storey Without L-beams 0.656 0.129 5.07 1.88
With L-beams 0.674 0.133 5.08 1.87
8 storey Without L-beams 1.324 0.266 4.98 2.08
With L-beams 1.539 0.311 4.95 2.14
12 storey Without L-beams 2.216 0.452 4.90 2.35
With L-beams 2.570 0.587 4.38 2.49
16 storey Without L-beams 3.234 0.708 4.57 2.71
With L-beams 3.814 1.038 3.67 2.70
20 storey Without L-beams 4.502 1.039 4.33 2.97
With L-beams 4.835 1.270 3.81 2.92
The dynamic equation of equilibrium is integrated by the unconditionally sta-
ble implicit Newmark Constant Average Acceleration (Newmark = 0.25) method
[Chopra, 2000]. The time-step for this form of integration method should be less
than 0.1 of the period of the highest mode of free vibration that contributes sig-
nicantly to the response of the building. Consequently, for these case studies a
time step of 0.005 s has been adopted. The results of the analyses, presented in the
following section, were output from Ruaumoko for post-processing every 0.01 s.
4.5. Results of time-history analyses
The two groups of case studies have been analysed under the suite of accelerograms
and the results have been processed to obtain displacements, shears, moments and
storey drifts. Results examined in this paper focus on the performance of the pro-
posed methodology with respect to its ability to control drifts and as such, only
displacements and drifts are included. In [Sullivan, 2005], the shears and moments
developed in the structures during the time history analyses are presented and used
to verify the performance of new capacity design recommendations.
4.5.1. Review of maximum recorded displacements
Figures 12, 13 and 14 present the maximum oor displacements recorded during
time-history analysis using the seven dierent accelerograms. These are compared
with the target displacement prole associated with the design drift limit for the
various height structures.
The scatter in the results obtained for the seven dierent records is in general
quite small. This indicates that the equivalent viscous damping approach was able
to maintain relatively uniform inelastic demands for the accelerograms used. For
the taller structures, record A4 tends to impose large displacements which are
foreseeable if the displacement spectra in Fig. 10 are examined closely. The target
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 117
Target
A1
A2
A3
A4
A5
A6
A7
0.0 0.1 0.2 0.3
Displacement (m)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.0 0.1 0.2 0.3 0.4
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.4
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
(i)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0 0.2 0.4 0.6 0.8
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0 0.2 0.4 0.6 0.8
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
Target
A1
A2
A3
A4
A5
A6
A7
(ii)
Fig. 12. Maximum recorded displacements compared with target displacements for the
(i) four-storey and (ii) eight-storey structures with (right) and without (left) link-beams.
April 28, 2006 13:47 WSPC/124-JEE 00274
118 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.0 0.5 1.0 1.5
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.0 0.5 1.0 1.5
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
Target
A1
A2
A3
A4
A5
A6
A7
(i)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.0 0.5 1.0 1.5
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.0 0.5 1.0 1.5
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
Target
A1
A2
A3
A4
A5
A6
A7
(ii)
Fig. 13. Maximum recorded displacements compared with target displacements for the
(i) 12-storey (ii) 16-storey structures with (right) and without (left) link-beams.
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 119
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.0 0.5 1.0 1.5
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
Target
A1
A2
A3
A4
A5
A6
A7
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.0 0.5 1.0 1.5
Displacement (m)
R
e
l
a
t
i
v
e

H
e
i
g
h
t

(
h
i
/
H
)
Fig. 14. Maximum recorded displacements compared with target displacements for the 20-storey
structures with (right) and without (left) link-beams.
displacement prole (shown dashed) lies either in the centre or conservatively to
the right of the recorded displacements suggesting that the design method has
worked well.
The excellent correlation between the recorded and anticipated displacements is
very convincing. However, the ability of the design method to control the damage
that the structures are subject to will be better gauged by comparison of the design
storey drift with the average maximum recorded storey drift.
4.5.2. Review of maximum storey drifts
Figures 15, 16 and 17 present the average of the maximum storey drifts recorded
during time-history analysis using the seven dierent accelerograms. These are com-
pared with the design drift prole associated with the various height structures. Also
shown for the 12, 16 and 20 storey structures is a dashed line that represents the
limiting drift. The design drift is less than the drift limits in these cases because of
the adjustment made to account for the eects higher modes have on storey drifts.
The results of the time-history analyses indicate that the design approach has
been very successful in limiting the storey drifts. Most encouragingly, the design
drift prole again provides an excellent match to the average maximum drift prole.
As would be expected, the storey drifts are correlated closely with the displacements
and therefore similar trends are observed.
April 28, 2006 13:47 WSPC/124-JEE 00274
120 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
0
1
2
3
4
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
1st mode
Target
T-history
Average
0
1
2
3
4
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
(i)
0
1
2
3
4
5
6
7
8
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
0
1
2
3
4
5
6
7
8
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
1st mode
Target
T-history
Average
(ii)
Fig. 15. Average of maximum recorded storey drifts compared with the target drift prole for
the (i) four-storey and (ii) eight-storey structures with (right) and without (left) link-beams.
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 121
0
2
4
6
8
10
12
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
1st mode
Target
T-history
Average
Drift limit
0
2
4
6
8
10
12
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
(i)
0
2
4
6
8
10
12
14
16
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
0
2
4
6
8
10
12
14
16
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
1st mode
Target
T-history
Average
Drift limit
(ii)
Fig. 16. Average of maximum recorded storey drifts compared with the target drift prole for
the (i) 12-storey and (ii) 16-storey structures with (right) and without (left) link-beams.
April 28, 2006 13:47 WSPC/124-JEE 00274
122 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
0
2
4
6
8
10
12
14
16
18
20
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
0
2
4
6
8
10
12
14
16
18
20
0.0% 1.0% 2.0% 3.0%
Storey drift
L
e
v
e
l
1st mode
Target
T-history
Average
Drift limit
Fig. 17. Average of maximum recorded storey drifts compared with the target drift prole for
the 20-storey structures with (right) and without (left) link-beams.
The adjustment for higher modes is most easily assessed through review of
the displacements and storey drifts for the 12- and 16-storey structures with link-
beams and the 20-storey structure without link-beams. For these structures, the
recorded and predicted displacement proles were closely matched and therefore any
dierences in storey drifts will highlight the eects of higher modes. Consequently,
it is clear that the higher mode adjustment that was made during design using
Eq. (8) has performed satisfactorily, especially for the 16-storey structure. For the
20-storey structure with link-beams, drifts are greater than desired over the top
storeys, suggesting that the higher mode reduction factor should have been greater
and as such, future research could aim to improve Eq. (8).
5. Conclusions
In conclusion, these case studies have clearly illustrated that the new design pro-
cedure for frame-wall structures provides excellent control of storey drifts and dis-
placements for buildings of up to 20 storeys in height. The interaction that takes
place between frame and wall elements has been successfully accounted for and
it has been shown that the approach works well when structures with or without
link-beams are considered.
The recommendations made for prediction of the displacement prole are con-
sidered valid for structures possessing RC walls with aspect ratio greater than three.
If the method were to be applied to structures possessing walls with aspect ratio
April 28, 2006 13:47 WSPC/124-JEE 00274
Direct Displacement-Based Design of Frame-Wall Structures 123
less than three, some account for shear deformations should be made. Furthermore,
it has been necessary to make simplifying assumptions as to the cracked elastic
stiness of wall elements. As such, future work could aim to reduce the uncertainty
associated with this part of the methodology.
Future work should verify that three-dimensional eects do not jeopardise
the ability of the method. Testing conducted at UC Berkeley indicated that 3-
dimensional eects tend to increase the overturning resistance of a structure. This
implies that such eects should not require any changes to the DBD process but
may need to be accounted for during capacity design. Maximum forces that develop
in the structures and recommendations for their capacity design are reported else-
where [Sullivan, 2005]. Such capacity design guidelines, together with the DBD
process presented here, complete a set of recommendations that provide designers
with a simple, rational and eective means of conducting seismic design of frame-
wall structures.
References
Blandon, C. A. and Priestley, M. J. N. [2005] Equivalent viscous damping equations
for direct displacement based design, Journal of Earthquake Engineering 9 (Special
Issue 2), 257278.
Carr, A. J. [2004] Ruaumoko3D A Program for Inelastic Time-History Analysis, Depart-
ment of Civil Engineering, University of Canterbury, New Zealand.
CEN [1998] Eurocode 8 Design Provisions for Earthquake Resistance of Structures,
prEN-1998-1:200X, Revised Final PT Draft (preStage 49), Comite Europeen de Nor-
malization, Brussels, Belgium.
Chopra, A. K. [2000] Dynamics of Structures (Pearson Education, USA).
Emori, K. and Schonbrich, W. C. [1978] Analysis of reinforced concrete frame-wall struc-
tures for strong motion earthquakes, Civil Engineering Studies, Structural Research
Series, No. 434, University of Illinois, Urbana, Illinois, USA.
Grant, D. N., Blandon, C. A. and Priestley, M. J. N. [2005] Modelling Inelastic Response
in Direct-Displacement Based Design, Research Report ROSE 2005/3, IUSS Press,
Pavia, Italy.
Gulkan, P. and Sozen, M. [1974] Inelastic response of reinforced concrete structures to
earthquake motions, ACI Journal 71(12), 604610.
NZS1170.5:2004 [2004] Structural Design Actions Part 5: Earthquake Actions
New Zealand (Standards New Zealand, Wellington, N.Z).
NZS3101 [1995] Concrete Structures Standard. Part 1 The Design of Concrete Struc-
tures, and Part 2 Commentary (Standards New Zealand, Wellington).
Otani, S. [1981] Hysteresis models of reinforced concrete for earthquake response analy-
sis, Journal of the Faculty of Engineering, University of Tokyo XXXVI(2), 125159.
Paulay, T. and Goodsir, W. J. [1986] The capacity design of reinforced concrete hybrid
structures for multistorey buildings, Bulletin of NZ National Society for Earthquake
Engineering, New Zealand 19(1), 117.
Paulay, T. and Priestley, M. J. N. [1992] Seismic Design of Reinforced Concrete and
Masonry Buildings (John Wiley & Sons, Inc., New York).
Priestley, M. J. N. [2003] Myths and Fallacies in Earthquake Engineering, Revisited, The
Mallet Milne Lecture (IUSS Press, Pavia, Italy).
April 28, 2006 13:47 WSPC/124-JEE 00274
124 T. J. Sullivan, M. J. N. Priestley & G. M. Calvi
Priestley, M. J. N. and Grant, D. N. [2006] Viscous damping for analysis and design,
Journal of Earthquake Engineering, Special Edition, printing in progress.
Priestley M. J. N. and Kowalsky M. J. [1998] Aspects of drift and ductility capacity of can-
tilever structural walls, Bulletin of the New Zealand National Society for Earthquake
Engineering, New Zealand National Society for Earthquake Engineering, Silverstream
31(2).
Priestley, M. J. N. and Kowalsky, M. J. [2000] Direct displacement-based seismic design of
concrete buildings, Bulletin, NZ National Society for Earthquake Engineering, New
Zealand 33(4), 421444.
Priestley, M. J. N. and Paulay, T. [2002] What is the stiness of reinforced con-
crete walls, SESOC Journal, Structural Engineering Society of New Zealand 15(1),
3034.
SeismoSoft [2004] SeismoStruct A computer program for static and dynamic nonlinear
analysis framed structures, Available on-line from URL: http//www.seismosoft.com.
Shibata, A. and Sozen, M. A. [1976] Substitute structure method for seismic design in
reinforced concrete, Journal of the Structural Division, ASCE 102(ST1).
Sullivan, T. J. [2005] Seismic Design of Frame-Wall Structures, PhD thesis, ROSE school,
Universit`a degli studi di Pavia, Italy.
Sullivan, T. J., Priestley, M. J. N. and Calvi, G. M. [2004] Displacement shapes of frame-
wall structures for direct displacement based design, Proceedings of Japan-Europe
5th Workshop on Implications of Recent Earthquakes on Seismic Risk, Bristol.
Sullivan, T. J., Priestley, M. J. N. and Calvi, G. M. [2005] Development of an innovative
seismic design procedure for frame-wall structures, Journal of Earthquake Engineer-
ing 9 (Special Issue 2), 279307.

You might also like