Highlighting Differences between Force-Based and

Displacement-Based Design Solutions for Reinforced

Concrete Frame Structures
Timothy John Sullivan, Dr., University of Pavia—Civil Engineering and Architecture, Pavia, Italy.
Contact: [email protected]
DOI: 10.2749/101686613X13439149156958

Abstract of engineering seismology and struc-

tural dynamics improved, acceleration
Over the past two decades, the earthquake engineering community has become response spectra were proposed and it
increasingly aware of fundamental shortcomings in the current code force-based was recognised that the peak accelera-
design methods. In light of this, a large number of displacement-based seismic tion of an elastic structure should be
design (DBD) methods have been proposed, and the most developed of these is a function of its period of vibration.
currently the Direct DBD method. In order to illustrate how significantly design However, following major earthquakes
solutions obtained by Direct DBD can differ from those obtained using current around the 1950s and 1960s, it was also
code methods, design strengths obtained by the Eurocode 8 equivalent lateral seen that ductile structures withstood
force (ELF) method and the Direct DBD method are compared for over 1000 ground accelerations that were con-
regular case study reinforced concrete (RC) frame buildings. The research exa- siderably greater than the accelera-
mines buildings from 1 to 12 storeys in height, investigating the impact of struc- tion required to cause yield. Thus, by
tural proportions, material properties, design drift limits, gravity loads and design the 1970s, it was observed that seismic
intensity on the seismic design strengths required. The findings illustrate that the design forces should be a function of
design strengths required by the Eurocode 8 ELF method can be significantly the building mass, period of vibration
smaller but can also be significantly greater than the Direct DBD solution. The and structural ductility capacity, and
conceptual reasons for the observed differences are discussed. Furthermore, the this gave rise to the equivalent lat-
research uses the results of the large number of Direct DBD solutions to identify eral force (ELF) method for seismic
means of simplifying the Direct DBD procedure for regular RC frames so that it design, which appears in most interna-
can be applied with similar ease to the current code ELF approach. tional codes to date.
Keywords: displacement-based design; force-based design; equivalent lateral The ELF method for RC frame
force; substitute structure; concrete frame. structures is illustrated in Fig. 1. This
paper makes specific reference to
the Eurocode 8 ELF method,5 but
it is similar to both the US and New
Introduction ground to current force-based seismic Zealand ELF procedures.
design procedures, reviews the motives
Advancements in seismic design con- for alternative displacement-based As shown in Fig. 1a, the ELF procedure
tinue to be fuelled by the communi- design procedures and then highlights utilises an equivalent single degree of
ty’s expectation for high performance the potential differences between freedom (SDOF) representation of
building solutions and the increasing force-based and displacement-based multiple degree of freedom (MDOF)
recognition that current code design design methods by comparing design structures. The main structural char-
procedures possess a number of short- base shear strengths for a large set acteristic is the period of vibration,
comings. The past 20 years have seen of reinforced concrete (RC) frame T1, which, as per the EC8, can be esti-
the promotion of performance-based structures. The work should not only mated as:
earthquake engineering concepts1–3 help engineers appreciate the differ-
as well as many different proposals ences between force-based and dis- T1 = Ct ·H 3/4 (1)
for displacement-based seismic design placement-based design but also help
(DBD) procedures (see Ref. [4] for a identify the role of key structural char- where H is the building height (in m),
review of a number of different pro- acteristics for seismic design. and Ct is an empirical constant depen-
cedures). Such procedures promise to dent on the structural typology and for
give designers effective control of the A Brief Background to Seismic RC frame structures Ct = 0,075.
damage and losses that can be expected Design
in an earthquake, providing invaluable The period of vibration is then used
tools for the mitigation of seismic risk. Seismic design is not new, with Pliny to identify an elastic spectral accelera-
This paper briefly discusses the back- reportedly proposing safety measures tion value, as shown in Fig. 1b, which
against earthquakes in the year 60 AD. can be reduced by a behaviour factor,
However, it appears as though it was q, as shown in Fig. 1c. The EC8 speci-
Peer-reviewed by international ex- not until the start of the 20th century fies the maximum allowable behaviour
perts and accepted for publication
by SEI Editorial Board that seismic design requirements began factors for RC frames as a function of
to be codified, with the earliest codes their ductility class: q = 5,85 for high
Paper received: July 30, 2012 setting design forces as a function of ductility (DCH) multi-storey multi-
Paper accepted: September 19, 2012 building mass. As our understanding bay frames and q = 3,9 for medium

122 Scientific Paper Structural Engineering International 2/2013

 Accn. Force
Elastic
Sa Fe
Fi
q
Fd mp

T1 Fd Inelastic

T1 = Ct.H 3/4
T1 Period (s) Disp.

(a) (b) (c)

Fig. 1: Conceptual overview of the ELF design procedure (adapted from Ref. [6]). (a) Equivalent SDOF system with initial period T1;
(b) use of elastic acceleration response spectrum; (c) elastic force, Fe, reduced to an equivalent design force, Fd to allow for inelastic re-
sponse from the yield displacement, Dy, to the displacement capacity, Du.

ductility (DCM) multi-storey multi- ceptual seismic design is simplified if displacement capacities of the wall and
bay frames. Behaviour factors are 15% seismic actions on buildings can be per- frame are greater than the non-struc-
smaller for single storey buildings ceived as accelerations multiplied by tural displacement capacity. With these
because it is argued that fewer hinges mass. For these reasons, and arguably points in mind, a few of the important
are required to reach instability in because the multi-modal version of shortcomings with current force-based
such structures. The period-dependent the approach was easily implemented design methods can be noted from
reduced “design” spectral acceleration in computer software, the method has Fig. 2:
value, Sd(T1), is then multiplied by the been included in seismic design codes
• Force-reduction factors should not
equivalent SDOF mass to obtain the from the 1970s to date. However, as
be set independent of expected duc-
design force, Fd, also referred to as the pointed out by Priestley,7,8 there are
tility demand. For the non-structural
design base shear, Vbase: a number of conceptual shortcomings
displacement limit shown, the duc-
with the methodology, and just a few of
tility demand on the system is a little
(2) these are explained with reference to
more than 2,0. The ductility capacity
Fig. 2, which depicts the lateral force–
where λ is a factor used to relate the of neither the wall nor the frame is
displacement response (right) of a dual
total mass, m, to the participating exhausted at this point, and there-
system RC frame-wall structure (left).
(SDOF) mass (mp in Fig. 1a); accord- fore it appears inappropriate to set
ing to the EC8, it adopts a value of 1,0 The force–displacement response of reduction factors as a function of
for one- or two-storey buildings and the frame-wall structure in Fig. 2 is the ductility capacity of structural
0,85 for buildings of three storeys or plotted for the combined frame-wall elements.
higher. system up to the attainment of a dis- • The use of elastic stiffness for the pre-
placement limit for non-structural diction of inelastic force distributions.
The base shear is used to identify a elements. The control of damage to In the elastic range of response, note
set of “equivalent lateral forces”, Fi, non-structural limits is a key consid- that the frame appears to be carry-
as per Eq. (3), which are then used as eration for performance-based seismic ing a little more than a quarter of the
part of an elastic structural analysis of design, which should take into account shear in the wall. However, because
the building in order to obtain design the performance of the whole build- the frames have a greater yield dis-
member strengths. The analysis should ing and not that of the structure alone. placement than the walls, they carry a
also be used to estimate the peak Note that for the case depicted, the much greater proportion of the total
displacements and storey drifts that
should be checked against code limits:
Felastic
mi hi
Fi = Fd
∑ mi h i (3) Force
Reduction by
behaviour
factor “q”
where mi is the mass and hi is the Frame Wall System yield
point
height above ground level of storey i.
Disp. capacity
The main benefit of this force-based Fy,sys
System wall
design method is its simplicity; only Disp. capacity
a couple of equations are required Wall frame
Fy,Wall
to obtain the design base shear. Frame
Fy,Frame
Furthermore, the method does address
the three historical observations iden-
tified earlier as the seismic design y,Wall y,Frame Displacement
force depends on the building mass, Disp. limit for non-
period of vibration and ductility capac- structural elements

ity (through the behaviour factor). In

(a) (b)
addition, engineers are familiar with
designing with forces for other load Fig. 2: Idealisation of a dual system RC frame-wall structure (a) and force–displacement
cases (gravity, wind, etc.), and con- points that characterise its lateral response (b) (after Refs. [9–11])

Structural Engineering International 2/2013 Scientific Paper 123

 shear at the system displacement
limit, illustrating that elastic analyses me Fu
may not predict inelastic force distri- F Fn rKi
butions correctly.
• The relationships used to relate elas-
tic displacement to inelastic displace- He Ki Ke
ment response. Figure 2 illustrates the
elastic system stiffness extrapolated
up to what would be a peak elastic
force demand, Felastic. As explained
earlier for the ELF method, this
elastic force can be reduced by the
behaviour factor q to obtain inelastic (a) (b)
design forces. In addition, note that
the EC8 recommends (for medium 0,25
= 0,05
0,5
period structures) that the peak Elasto-plastic
inelastic displacements should be Steel frame
0,2 0,4 = 0,10
assumed equal to the elastic displace-

Displacement (m)
← Concrete frame
Damping ratio,

ment demand. Other international = 0,15

codes make different recommenda- 0,15 Concrete bridge 0,3 = 0,20
= 0,30
tions, illustrating the uncertainty in
this approximation. In reality, the 0,1 0,2
Hybrid prestress →
relationship between the elastic
and inelastic spectral displacement 0,05 0,1
demand should depend on the hys-
↑ T
↓ e
teretic properties of the system. As 0 0
such, a system that dissipates more 0 2 4 6 0 1 2 3 4 5
energy (with fatter hysteretic loops) Displacement ductility Period (s)
should be expected to have a smaller
inelastic to elastic spectral displace- (c) (d)
ment ratio. However, the benefits of Fig. 3: Conceptual overview of the direct DBD procedure (reproduced with permission
energy dissipation on displacement from Ref. [8]). (a) SDOF simulation; (b) effective stiffness Ke ; (c) equivalent damping and
demands are not recognised in the ductility; (d) design displacement spectra
ELF method.
There are several other limitations to
force-based design methods but for peak displacement response (as shown 2,0–2,5% will typically govern design
the sake of brevity they are not dis- in Fig. 3b); the effective height, He; the at the ultimate limit state.8 Equation
cussed here, and interested readers effective mass, me and the equivalent (4) also includes w θ, a higher mode
should instead refer to Ref. [8] for fur- viscous damping, x. After selecting pre- drift reduction factor, which is recom-
ther details. liminary member sizes, the first step is mended12 to be set equal to 1,0 for
to set a system design displacement, Δd, frame buildings of up to six storeys
in order to satisfy target performance height and reduces linearly to 0,91 for
Direct Displacement-Based Design criteria (that could be storey drift 12 storey frame buildings (this paper
of RC Frame Buildings limits, section curvature limits, chord focuses on regular frame buildings of
rotation limits or even residual defor- up to 12 storeys).
In order to address the shortcomings
mation limits; see Ref. [12]). To relate
of force-based design methods a large The design displacement profile of Eq.
local deformation demands to system
number of DBD methods have been (4) is used to determine the substitute
displacement demands for RC frame
proposed in the literature.4 This paper structure characteristics of design dis-
structures, the following expression for
focuses on the most developed DBD placement, effective mass and effective
the displaced shape at peak response
approach, known as the Direct DBD height, as shown in Eqs. (5–7):
is used (when inelastic response is
method, which has been published both
expected):
as text8 and in a model code format12.
In addition, note that Direct DBD
Δd =
∑ mi Δ2i (5)
software is currently under develop-
ment13 for RC structures.
(4) ∑ mi Δi
An overview of the Direct DBD where hi is the height of level i above
me =
∑ mi Δi (6)
approach8 is shown in Fig. 3. The the base, Hn is the total building Δd
MDOF frame structure is represented height, h1 is the height of the first sto-
by an equivalent SDOF system, as
shown in Fig. 3a, in line with the substi-
rey, qc is the critical storey drift limit
that should be set to satisfy drift lim- He =
∑ mi Δi hi
(7)
tute structure concept of Gulkan and its for non-structural elements, section ∑ mi Δ i
Sozen14 and Shibata and Sozen.15 Key curvature limits and possibly residual
characteristics of the equivalent SDOF drift limits (see Ref. [12] for guidance). where Δi is the design displacement, mi
system include the effective stiffness, For well-detailed RC frames, drift is the seismic mass and hi is the height
Ke, equal to the secant stiffness at limits for non-structural elements of of level i.

124 Scientific Paper Structural Engineering International 2/2013

Equation (8) then provides an estimate required effective period from the Comparison of Design
of the system ductility demand associ- highly damped design displacement Strengths Obtained from
ated with the design displacement: spectrum. The required effective stiff-
ness, Ke, and design base shear, Vb, Direct DBD with Those
Δ Δ are then obtained from the effective Obtained from the EC8
= d = d (8)
Δy y He
period, Te, and effective mass, me, as Equivalent Lateral-Force
per Eq. (12): Method
where Δy is the yield displacement of
the RC frame, approximated by mul- In order to highlight the differences
tiplying the effective height He by the one could expect between Direct
frame yield drift qy, which can be esti- (12) DBD and FBD solutions, this section
mated from: compares design base shear values,
where the right term in the base shear normalised by the building weight,
y Lb computation is for P–d effects, with C for a large number of RC frame con-
y = 0,5 being a coefficient equal to 0,5 for RC figurations. Design solutions are only
hb (9)
structures, and g is the acceleration developed for the ultimate limit state,
where ey is the yield strain of the lon- due to gravity. All the other symbols assuming that this will be the critical
gitudinal reinforcement in the beams, have been defined earlier. limit state for RC frame buildings. In
Lb is the beam length between col- applying the EC8 ELF method, the
umn centres and hb is the beam sec- The base shear can then be distrib- height-dependent period expression
tion depth. The drifts predicted by uted as a set of ELFs using Eq. (13) of Eq. (1) is used and solutions are
Eq. (9) have been shown8 to correlate and used in the structural analysis of obtained for frames of both medium
well with the RC frame yield drifts the frame to obtain design strengths of (DCM) and high (DCH) ductility
recorded experimentally. plastic hinges. Frame elements should classes. The maximum behaviour fac-
be modelled with effective (secant) tors permitted by the EC8 (reported
A ductility-dependent equivalent vis- stiffness properties for this analysis earlier in Section A Brief Background
cous damping value is then obtained (see Ref. [8]): to Seismic Design) are adopted for the
as per Fig. 3c. Note that the ductility- design, as would be typical of engi-
dependent equivalent viscous damp- mi Δi neering practice.
Fi = Vb
ing expressions used in modern Direct
DBD procedures are calibrated to
∑ mi Δi (13)
In applying the Direct DBD approach
the results of numerous nonlinear the design displacement profile of Eq.
dynamic analyses undertaken for spe- For frame structures, Researchers8 (4) is used, but higher mode effects are
cific hysteretic models (see Ref. [16] also recommend that Eq. (13) be neglected by setting wq equal to 1,0.
for discussion). For RC frames, it is modified such that 10% of the design This was considered appropriate for
recommended8 that the equivalent vis- base shear is lumped at roof level with this comparison exercise, given that
cous damping be calculated from: the remainder distributed as per Eq. higher mode effects are also ignored in
(13). Capacity design procedures8,17,18 the basic ELF method. P–d effects are
should then be followed to establish assumed here to be negligible for both
(10) design forces for other elements and the Direct DBD and ELF procedures,
also for capacity-protected actions in and although they would need to be
where m is the frame ductility demand plastic hinge regions. checked in practice, this assumption
from Eq. (8). is not expected to change the general
This brief description of the Direct
As shown in Fig. 3d, the design displace- DBD methodology demonstrates that trends highlighted by this comparison
ment spectrum is scaled to the design it is relatively straightforward but in study.
value of equivalent viscous damping. comparison with the ELF methodo-
There are various expressions in the Characteristics of the Case Study RC
logy, it does require a greater num-
literature proposed for such scaling, Frame Buildings
ber of computations. Conceptually,
and care is required to ensure that however, the Direct DBD procedure In order to adequately highlight the
the adopted scaling expression used permits more rational seismic design potential differences between force-
in Direct DBD is representative of the since the limitations associated with based and displacement-based design,
ground motions used to develop the force-based design are overcome; a large number of RC frame configu-
equivalent viscous damping curves, as ductility demands are used instead of rations are examined in this work.
explained in Ref. [16]. For the damp- ductility capacities, inelastic force dis- Figure 4 depicts a general case-study
ing expression of Eq. (10), the fol- tributions are implicitly considered frame, with a number of key variables
lowing scaling expression is deemed by using effective (secant) stiffness highlighted.
appropriate: properties and the effect of hysteretic
As indicated in Fig. 4, the variables
properties on inelastic displacement
0,5 considered for the case study designs
7 demands are accounted for with the
= include the following:
(11) equivalent viscous damping approach.
2+
The question that then follows is, “how • Number of storeys: 1, 2, 3, 4, 5, 6, 7, 8,
significant are these conceptual differ- 9, 10, 11, 12;
where x is the design value of the
ences for the final design strengths?” • Beam aspect ratio, Ar = Lb/hb = 6, 9,
equivalent viscous damping.
The next section of this paper exam- 12, 15;
As shown in Fig. 2d, the design dis- ines this question for the case of RC • Longitudinal reinforcement strength,
placement is then used to read the frame structures. Fy = 300, 400, 500 MPa;

Structural Engineering International 2/2013 Scientific Paper 125

 Beam Uniform bay length aspect ratios. The results are shown for
depth hb Lb frames designed to a storey drift limit
Design done for different
number of storeys: of 2,0% for ag = 0,3 g and with longi-
Design done for different
n = 1 to 12 storeys tudinal reinforcement strength of Fy
beam aspect ratios,
Ar = Lb/hb = [6, 9, 12, 15] = 500 MPa. The beam aspect ratio is
defined as the ratio of the beam length
Storey mass variable since Constant to the beam depth, as shown in Fig. 4.
design base shear normalised by storey height, From Section A Brief Background to
building weight hs = 3,3 m Seismic Design it is seen that the nor-
Three different design intensities malised design base shear in the ELF
considered: PGA = 0,2, 0,3 & 0,4g procedure is independent of the beam
aspect ratio, varying only as a function
Three different design intensities Number of bays
assumed variable
of the building height (which is used
considered: ag = 0,2, 0,3 & 0,4 g
to find the building period), the duc-
Fig. 4: Elevation of generic case study RC frame building tility class (behaviour factor) and the
number of storeys. As such, only two
ELF curves are plotted in Fig. 5 for the
• Seismic design intensity: EC8 type are designed for the same drift limit medium and high ductility classes.
1 spectrum, ground type C and ag = at the same earthquake intensity then The results shown in Fig. 5 permit some
0,2, 0,3, 0,4 g; the normalised design base shear for interesting observations to be made.
• Storey drift limit, qc = 1,5, 2,0, 2,5%. the two buildings would be expected Firstly, note that the Direct DBD
to be the same. method generally leads to greater
As such, the Direct DBD and ELF
methods for both medium (DCM) As reported above, design solutions design base shear forces than the
and high (DCH) ductility classes have have been developed for the EC8 type EC8 ELF method. This is effectively
been applied to a total of 1296 (12 × 4 1 response spectrum for ground (soil) because the Direct DBD approach
× 3 × 3 × 3) design scenarios. Note that type C, and for three different values evaluates the effect of energy dissi-
design has only been conducted to the of peak ground acceleration—0,2, 0,3 pation associated with the ductility
point that gives the design base shears, and 0,4 g. In setting the design spectral demand at the drift limit, whereas the
and the design base shears have been shape, however, it was decided to adopt ELF method assumes that forces can
normalised by the building weight. a spectral displacement corner period, be reduced by the full strength reduc-
The normalisation by the building TD, equal to 8,0 s since the value of tion factor. In addition, for low-rise
weight actually implies that the exact 2,0 s recommended in the EC8 may be buildings the spectral acceleration pla-
building layout, including the num- non-conservatively low.8 teau tends to limit the design forces
ber of bays, tributary floor area and for the ELF method more significantly
beam and column dimensions are not than for the Direct DBD method. As
Impact of Structural Proportions the ELF method uses a height-depen-
required and that the design solutions
developed are applicable to any frame The first parameter to be varied to dent period expression here, there is
configuration that fits the general highlight differences between Direct no relationship between the build-
parameters shown in Fig. 4. In other DBD and the ELF methods are the ing stiffness and the design forces. As
words, if two buildings with very dif- structural proportions. Figure 5 pres- such, if the cracked stiffness is over-
ferent mass but with the same num- ents the normalised design base shear estimated when analysing the frame
ber of storeys, beam aspect ratios and for RC frame buildings of 1 to 12 sto- under the ELFs obtained from Eq. (3),
longitudinal reinforcement strength reys height and four different beam the peak displacement demands will

0,90 0,30
normalised by building weight (Vb/Wt)

normalised by building weight (Vb/Wt)

0,80

0,70
DBD Ar = 15
Design base shear

Design base shear

0,60 0,20
DBD Ar = 12
0,50
DBD Ar = 9
0,40 DBD Ar = 6

0,30 ELF (DCM)

0,10
ELF (DCH)
0,20

0,10

0,00 0,00
1 2 3 4 5 6 6 7 8 9 10 11 12
Number of storeys Number of storeys

Fig. 5: Comparison of design base shear forces (normalised by building weight) from DBD with those obtained from the EC8 ELF
method, highlighting the influence of the beam aspect ratio Ar = Lb/hb (reinforcement yield strength Fy = 500 MPa, design drift limit
qc = 2,0% and PGA = 0,3 g)

126 Scientific Paper Structural Engineering International 2/2013

be underestimated and designers may The results in Fig. 6 show that Direct sities corresponding to the EC8 type
be unaware that ELF method provides DBD design shear forces tend to 1 response spectrum for ground (soil)
insufficient strength to adequately increase with increasing longitudinal type C scaled to peak ground accelera-
control the storey drift limits. reinforcement strength. This occurs tions of 0,2, 0,3 and 0,4 g. The results
because the yield strain increases are again shown for frames of 1 to 12
The second trend that should be
proportionally to the reinforcement storeys, with a beam aspect ratio of
observed from Fig. 5 is the influence of
strength, reducing the expected duc- Ar = 9,0 and reinforcement strength
structural proportions on the required
tility demand and energy dissipation of Fy = 500 MPa, designed to a storey
design strength. In Direct DBD, it is
for the drift limit and subsequently drift limit of 2,0%.
recognised that the structural propor-
increasing the required design shear.
tions will affect the yield displacement The results shown in Fig. 7 reveal
This trend, recognised previously by
and, consequently, the ductility demand an interesting influence of seismic
Priestley,8 suggests that there may be
for a given drift limit. As the energy intensity on design strengths for the
little advantage of using high-grade
dissipation is proportional to the duc- force-based and displacement-based
reinforcement for seismic design. This
tility demand, it follows that frames methods. As the design intensity
is not the case for force-based design
with large beam aspect ratios, which increases from 0,2 to 0,4 g the Direct
where the design base shear forces are
are subject to low ductility demands, DBD design base shear increases by
unaffected by the grade of reinforce-
require greater design strengths than roughly a factor of 4,0, whereas the
ment, and as such, the use of high-
systems with low beam aspect ratios ELF design shear forces increase only
grade reinforcement leads to smaller
that tend to yield at smaller displace- by a factor of two. As such, while the
reinforcement quantities in the ELF
ment demands, dissipating a greater Direct DBD method requires lower
method. However, the effect of the
amount of energy for the same drift design forces than the ELF method for
reinforcement grade on the design
limit. The results in Fig. 5 show that regions of low seismicity, the approach
base shears does not appear to be very
the beam aspect ratio can change the requires considerably higher design
large, with differences of around 25%
required design base shear by a factor forces for regions of high seismicity.
obtained for the Direct DBD solutions
of two in Direct DBD but has no effect This trend occurs because in the ELF
for Fy = 500 and 300 MPa.
on the ELF design forces. method the design force is directly
Interestingly, the design base shear is proportional to the seismic intensity
Impact of Material Properties assumed to be independent of the con- (as per Eq. (2)), whereas in the Direct
crete compressive strength for both DBD method the base shear is pro-
In force-based design, material proper-
Direct DBD and force-based design of portional to the square of the seismic
ties are not expected to alter the design
RC frame structures. In Direct DBD intensity because the required effec-
forces. In Direct DBD, however, it is
this is because it was observed (experi- tive period is inversely proportional
recognised (see Eq. (9)) that the yield
mentally and analytically8) that the to the intensity and the base shear is
drift of RC frame structures will be
yield deformations (and therefore duc- inversely proportional to the effective
affected by the longitudinal reinforce-
tility demands and energy dissipation) period squared (as per Eq. (12)).
ment strength and that this, in turn, will
are relatively unaffected by the con-
affect the ductility and energy dissipa-
crete compressive strength, depending Impact of the Design Drift Limit
tion for a given drift limit. Figure 6 com-
mainly on the reinforcement strength
pares the normalised design base shear Figure 8 reveals how the design drift
and structural proportions.
for RC frame buildings of 1 to 12 sto- can affect the normalised design base
reys height and three different grades shear forces for displacement-based
Impact of Design Intensity
of longitudinal reinforcement—Fy = design and compares the design
300, 400 and 500 MPa. The results are Figure 7 shows the normalised design shears with those obtained from the
shown for frames with a beam aspect base shear forces obtained from both ELF method. The three design drifts
ratio of Ar = 9,0, designed to a storey force-based and displacement-based considered for the Direct DBD solu-
drift limit of 2,0% for ag = 0,3 g. design for three different design inten- tion are 1,5, 2,0 and 2,5%.

0,50 0,20
0,45
Design base shear normalised by

Design base shear normalised by

0,40
building weight (Vb/Wt)

building weight (Vb/Wt)

0,15
0,35 Fy = 500 MPa
0,30 Fy = 400 MPa
0,25 Fy = 300 MPa 0,10
0,20 ELF (DCM)
0,15 ELF (DCH)
0,05
0,10
0,05
0,00 0,00
1 2 3 4 5 6 6 7 8 9 10 11 12
Number of storeys Number of storeys

Fig. 6: Comparison of design base shear forces (normalised by building weight) from DBD with those obtained from the EC8 ELF
method, highlighting the influence of the reinforcement yield strength (beam aspect ratio Ar = Lb/hb = 9, design drift limit q = 2,0% and
PGA = 0,3 g)

Structural Engineering International 2/2013 Scientific Paper 127

 0,90 0,30

Design base shear normalised by

Design base shear normalised by

0,80
0,25
building weight (Vb/Wt)

building weight (Vb/Wt)

0,70
DBD ag = 0,4 g
0,60 0,20
DBD ag = 0,3 g
0,50 DBD ag = 0,2 g
0,15
0,40 ELF ag = 0,4 g
0,30 ELF ag = 0,3 g 0,10
0,20 ELF ag = 0,2 g
0,05
0,10
0,00 0,00
1 2 3 4 5 6 6 8 10 12
Number of storeys Number of storeys

Fig. 7: Comparison of design base shear forces (normalised by building weight) from DBD with those obtained from the EC8 ELF
(DCH) method, highlighting the influence of design intensity (reinforcement yield strength of Fy = 500 MPa, beam aspect ratio
Ar = Lb/hb = 9 and design drift limit q = 2,0%)

0,70
0,50
normalised by building weight (Vb/Wt)

normalised by building weight (Vb/Wt)

0,60
0,40
0,50
DBD Dr = 1,5%
Design base shear

Design base shear

0,40 DBD Dr = 2,0% 0,30
DBD Dr = 2,5%
0,30 ELF (DCM)
0,20
ELF (DCH)
0,20

0,10
0,10

0,00 0,00
1 2 3 4 5 6 6 7 8 9 10 11 12
Number of storeys Number of storeys

Fig. 8: Comparison of design base shear forces (normalised by building weight) from DBD with those obtained from the EC8 ELF
method, highlighting the influence of the design drift limit on the DBD solution (obtained for reinforcement yield strength of
Fy = 500 MPa, beam aspect ratio Ar = Lb/hb = 9 and a PGA = 0,3 g)

The results shown in Fig. 8 illustrate design base shear was obtained and tic analysis in the EC8 ELF approach
that the design drift limit can affect the that elastic structural analysis would would underestimate the drift demands
Direct DBD shear forces significantly, then follow to check that the expected would be if the equal-displacement
with shear forces changing by a factor drifts do not exceed code limits. As rule were non-conservative (this is the
of nearly three for a change in design such, one could argue that the design case for certain hysteretic properties as
drift from 1,5 to 2,5%. This occurs shear values shown in Fig. 8 for the shown by Priestley et al.8 among oth-
because larger design displacements ELF method are not the final design ers). If, however, the cracked stiffness
permit higher levels of energy dissi- shear forces. However, note that the is overestimated (note that the EC8
pation and also lead to significantly ELF procedure has been applied here recommendation to consider 50% the
longer effective periods. This illus- using a period estimate that is indepen- gross section stiffness is likely to over-
trates how improving the deformation dent of the structural stiffness. As the estimate the actual cracked stiffness8),
capacity of a building can significantly generalised case study configurations then the drift demands obtained from
improve its seismic resilience. Note examined here do not restrict the total elastic analysis in the ELF approach
that for the shorter frame systems number of bays, column dimensions or will underestimate the actual drift
designed to 1,5% drift, low system duc- beam widths, it is clear that a code-com- demands, and engineers are likely to be
tility demands are expected and there- pliant configuration could be found for unaware that drift limits are not being
fore the Direct DBD base shear forces all the EC8 ELF shear values shown. respected.
are considerably greater than the ELF Nevertheless, the drifts estimated in
(DCH) forces that have been obtained the ELF approach are dependent on Potential Impact of Gravity Loads
by reducing the elastic shear force by a the cracked stiffness values assumed
for the elastic analysis. If accurate In accordance with the EC8, the beam
factor of up to 5,85.
cracked section stiffness values are moment obtained for the seismic load
Recall that the ELF procedure was set as a function of the design shear case should be added to the beam
undertaken only to the point that the strength, then the only reason that elas- moment obtained from unfactored

128 Scientific Paper Structural Engineering International 2/2013

gravity loads. In Direct DBD, how- 9000
ever, it is recommended8 that grav-
8000
ity moments not be added to seismic
moments. The reason for this is that 7000
Direct DBD relies on providing a

Design base shear (kN)

sufficient effective stiffness at the 6000
design displacement. If beam strengths
5000
obtained from Direct DBD were
added to those obtained from a static 4000
gravity analysis, this would increase
plastic hinge strengths and unneces- 3000
sarily increase the effective stiffness.
2000
As such, in Direct DBD the final beam
strengths should be set equal to the 1000
maximum of unfactored seismic loads
(obtained from Direct DBD) or fac- 0
tored gravity loads (and other poten- 1 2 3 4 5 6 7 8 9 10 11 12
tially critical load cases such as wind). Number of storeys
Consequently, even though the results Fig. 9: Direct DBD base shear forces obtained for case study frame buildings possessing
presented thus far in Figs. 5–8 indi- the same storey weight (of 1000 T) and member proportions
cate that Direct DBD often results in
greater design base shear forces than
the ELF method, final beam strengths it may well be possible to simplify the In addition to the design displacement
may still be greater using the code design procedure. Similar observations and effective mass, the only parameter
approach depending on the magnitude have been made in Refs. [8, 19, 20]. required to obtain the design base
of gravity loads. This is also why engi- shear is the equivalent viscous damp-
neers often observe a great level of Simplified Expressions for Substitute ing. As was shown earlier in Eq. (10),
overstrength when conducting push- Structure Characteristics equivalent viscous damping is a func-
over analyses of frames designed using tion of the ductility demand m, and as
the EC8 ELF approach. In order to arrive at a simplified Direct such, Eq. (18) is proposed to permit
DBD procedure, simplified expres- quick estimation of the ductility and
sions are proposed here for the sub- therefore damping, without computa-
Potential Simplifications stitute structure characteristics. It was tion of the design displacement profile:
to the Direct DBD Procedure shown in Direct DBD of RC Frame
for RC Frames Buildings that the substitute structure
design displacement, effective mass
The previous section has illustrated (18)
and effective height are all a function
that by accounting for a considerably of the design displacement profile. For
increased number of design param- RC frames, it was further reported that where εy is the yield strain of longitu-
eters, Direct DBD can lead to very the design displacement profile (Eq. dinal reinforcement and Ar is the beam
different design base shear values com- (4)) is empirical. As such, it seems fea- aspect ratio (Lb/hb); the other symbols
pared to the ELF method. However, sible that one could directly formulate have been defined earlier.
it is also apparent that in the process, empirical expressions for the substi- In order to gauge the accuracy of the
the Direct DBD procedure requires tute structure characteristics without proposed expressions, Fig. 10 com-
a greater number of calculations and the need to make reference to a design pares predictions made from Eqs.
is more time consuming than the ELF displacement profile. After examining (14) to (18) with the rigorous values
method, at least for the determination trends in the substitute structure char- obtained from the full Direct DBD
of the design base shear. In this sec- acteristics for the case study frames, procedure. The results shown corre-
tion, the possibility of simplifying the the following empirical expressions spond to case study frame structures
Direct DBD procedure for RC frames are proposed for the substitute struc- possessing a reinforcement strength
is explored. ture characteristics of RC frames: of Fy = 500 MPa designed for a seis-
There are good reasons to think that 1,5 mic intensity of ag = 0,3 g. It can be
Δd = c Hn fs (14) seen that all the design parameters
the Direct DBD procedure can be sim-
plified for RC structures. Consider, for me = m· fs are approximated well by the simpli-
(15)
example, the design base shears shown fied expressions. The only significant
in Fig. 9 that are obtained for the 1 to He = Hn fs (16) discrepancies tend to occur for the two
12 storey buildings that possess a uni- and three storey buildings where dif-
form storey mass of 1000 T. The frame where θc is the design storey drift limit, ferences of up to around 3% occur in
buildings are all characterised by the Hn is the total building height, m is the the predictions of effective height and
same reinforcement strength and beam total building mass and fs is an empiri- effective mass.
aspect ratio and have been designed to cal substitute structure factor that can
limit the peak storey drift to 2,0% for a be found simply as a function of the Direct Calculation of the Design
design intensity of 0,3 g. It is apparent number of storeys, n, as: Base Shear
from Fig. 9 that the design base shear
is relatively constant for buildings of (17) Having established simplified expres-
four storeys or greater, suggesting that sions for the substitute structure charac-

Structural Engineering International 2/2013 Scientific Paper 129

 (a) (b)
1,200 1,200
Ratio of effective height to total height, He/Hn

Ratio of effective mass to total mass squared

BAr6Dr1,5
BAr9Dr1,5
1,000 1,000
BAr12Dr1,5
BAr15Dr1,5
0,800 BAr6Dr2,0 0,800
BAr9Dr2,0

(me/m)2
0,600 BAr12Dr2,0 0,600
BAr15Dr2,0
0,400 BAr6Dr2,5 0,400
BAr9Dr2,5
0,200 BAr12Dr2,5 0,200
BAr15Dr2,5
Eqtn. 17
0,000 0,000
1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12
Number of storeys Number of storeys

(c) (d)
1,00 4,00
BAr6Dr1,5

Ductility demand estimated from Eq.(8)

BAr9Dr1,5
Design displacement from Eq.(5)

3,50
0,80
BAr12Dr1,5
BAr15Dr1,5
3,00
0,60 BAr6Dr2,0
BAr9Dr2,0
2,50
BAr12Dr2,0
0,40 BAr15Dr2,0
2,00
BAr6Dr2,5
0,20 BAr9Dr2,5
1,50
BAr12Dr2,5
BAr15Dr2,5
0,00 1,00
0,00 0,20 0,40 0,60 0,80 1,00 1,0 1,5 2,0 2,5 3,0 3,5 4,0
Design displacement from Eq. (14) Ductility demand estimated from Eq. (18)

Fig. 10: Accuracy of simplified expressions for the substitute structure characteristics: (a) effective height ratios, (b) effective mass ratios,
(c) design displacements and (d) ductility demands (Units: m)

teristics of regular RC frames, it is also 1,20

Ratio of base shear from Eq. (12) to base

BAr6Dr1,5
proposed that the design base shear, Vb,
BAr9Dr1,5
can be found directly using Eq. (19): 1,00
BAr12Dr1,5
shear from Eq. (19)

BAr15Dr1,5
0,80
(19) BAr6Dr2,0
BAr9Dr2,0
0,60
BAr12Dr2,0
where TD is the spectral displacement BAr15Dr2,0
corner period (4,0 s for the spectra 0,40
BAr6Dr2,5
shown in Fig. 3d), SD is the elastic spec- BAr9Dr2,5
tral displacement demand at the cor- 0,20 BAr12Dr2,5
ner period (approximately 0,5 m for BAr15Dr2,5
the spectrum shown in Fig. 3d), h is 0,00
0 1 2 3 4 5 6 7 8 9 10 11 12
the damping dependent spectral scal-
ing factor from Eq. (11) (obtained by Number of storeys
inserting the ductility from Eq. (18) into Fig. 11: Ratio of design base shears obtained using the simplified approach with those
the damping expression of Eq. (10)) obtained from the rigorous Direct DBD approach from Ref. [8] (beam aspect ratios and
and PGA is the peak ground accelera- design drift limits shown in legend).
tion for the site; all the other symbols
have been defined in the previous sec-
tion. Note that this expression does not Figure 11 compares the ratio of design DBD approach. The results are again
include an allowance for P–d effects base shear determined from Eq. (19) shown for case study frame structures
and, therefore, should be adjusted if P–d with the design base shear obtained possessing a reinforcement strength
effects are deemed to be significant. from Eq. (12) using the rigorous Direct of Fy = 500 MPa and designed for a

130 Scientific Paper Structural Engineering International 2/2013

seismic intensity of ag = 0,3 g. The beam base shears required by the Eurocode wall structures. J. Earthquake Eng. 2011; 7(15):
aspect ratios and drift limits considered 8 ELF method tend to be smaller than 1083–1116.
are shown in the legend. It is apparent those obtained through Direct DBD [7] Priestley MJN. Myths and fallacies in earth-
that the approximate approach per- for low-rise RC frames, but differences quake engineering – conflicts between design
forms well, with the design base shear in gravity load considerations and and reality. Bull. NZ Nat. Soc. Earthquake Eng.
well predicted for all building heights. cracked stiffness assumptions imply 1993; 26(3): 328–341.
For two and three storey frame struc- that consistent trends in final differ- [8] Priestley MJN, Calvi GM, Kowalsky MJ.
tures, discrepancies of up to 13% of ences may be difficult to predict. The Displacement Based Seismic Design of Structures.
the design base shear are observed for final part of the research has taken IUSS Press: Pavia, 2007; 721.
some design scenarios, and, therefore, the results of the large number of case [9] Sullivan TJ, Priestley MJN, Calvi GM.
some conservatism could be adopted study solutions to identify means of Development of an innovative seismic
(by using the full mass rather than simplifying the Direct DBD procedure design procedure for frame-wall structures.
J. Earthquake Eng. 2005; 9(Special Issue 2)
the effective mass for example) when for regular RC frames. The main sim-
279–307.
using the approach for low-rise frame plification proposed is to directly for-
buildings. mulate empirical expressions for the [10] Sullivan TJ, Priestley MJN, Calvi GM.
Direct displacement based design of frame-wall
substitute structure characteristics of
The results of this section suggest that structures. J. Earthquake Eng. 2006; 10(Special
regular RC frames without construct- Issue 1) 91–124.
the Direct DBD procedure could cer-
ing a design displacement profile. By
tainly be simplified by using substitute [11] Calvi GM, Sullivan TJ. Development of
comparing design base shear forces
structure factors, at least for regular RC a Model Code for Direct Displacement Based
obtained from the two methods, it is Seismic Design, Prodotto ReLUIS 2005-2008,
frame structures. Note also that a sim-
shown that the simplified procedure Linea 4, 2009. www.reluis.it.
plified displacement-based design pro-
provides a good approximation to the
cedure that uses substitute structure [12] Sullivan. TJ, Priestley MJN, Calvi GM (eds).
more rigorous Direct DBD approach.
factors has been recently proposed6 A Model Code for the Displacement-Based
for RC wall structures. Given that the Seismic Design of Structures, DBD12. IUSS
simplifications proposed in this paper Acknowledgements Press: Pavia, 2012; 105. ISBN: 978-88-6198-072-3.
do not greatly affect the accuracy of [13] Sullivan TJ, Bono F, Magni F, Calvi GM.
The author wishes to recognise that without Development of a computer program for direct
the Direct DBD procedure and permit thorough discussions with Nigel Priestley,
consideration of the same number of displacement-based design. Proceedings of 15th
Gian Michele Calvi and Mervyn Kowalsky, World Conference on Earthquake Engineering,
variables and structural characteris- this paper would not have been written. Lisbon, paper 4135, 2012.
tics, it appears that such simplifications
warrant further examination as part of [14] Gulkan P, Sozen M. Inelastic response of
References reinforced concrete structures to earthquake
future research. motions. ACI J. 1974; 71(12): 604–610.
[1] (SEAOC) Structural Engineers Association
of California, Vision 2000. A Framework for [15] Shibata A., Sozen M.A., Substitute struc-
Conclusions Performance-Based Engineering. Structural ture method for seismic design in reinforced
Engineers Association of California: Sacramento, concrete, J. Struct. Div., ASCE 1976; 102 (ST1).
Advancements in seismic design have 1995.
[16] Pennucci D, Sullivan TJ, Calvi GM.
seen the development of the Direct
[2] Porter K. An overview of PEER’s perfor- Displacement reduction factors for the design
DBD procedure, which is capable of mance-based earthquake engineering meth- of medium and long period structures. J.
considering a larger number of build- odology. Ninth Conference on Application of Earthquake Eng. 2011; 15(Suppl 1): 1–29.
ing characteristics and design param- Statistics and Probability in Civil Engineering.
[17] Sullivan TJ, Priestley MJN, Calvi GM.
eters than traditional force-based San Francisco, 2003.
Estimating the higher mode response of duc-
methods. By comparing design base [3] Fardis M, ed. Advances in Performance-Based tile structures. J. Earthquake Eng. 2008; 12(3):
shears obtained using the Eurocode Earthquake Engineering. Dordrecht, Springer 456–472.
8 ELF method and the Direct DBD Netherlands. 2010. ISBN 978-90-481-8745-4.
[18] Sullivan TJ. Capacity design considerations
method for over 1000 regular case [4] Sullivan TJ, Calvi GM, Priestley MJN, for RC frame-wall structures. Earthquakes
study RC frame buildings, this work Kowalsky MJ. The limitations and performances Struct. 2010; 1(4): 391–410.
has illustrated that the differences of different displacement based design methods.
[19] Priestley MJN. Performance based seismic
between the methods can be very large. J. Earthquake Eng. 2003; 7(S1): 201–241.
design. Proceedings 12th World Conference on
The research has shown that struc- [5] CEN EC8, Eurocode 8 – Design Provisions Earthquake Engineering, Auckland, paper 2831,
tural proportions, material properties, for Earthquake Resistant Structures, EN-1998- 2000.
design drift limits and design intensity 1:2004: E, Comite Europeen de Normalization,
[20] Sullivan TJ, Priestley MJN, Calvi GM.
all affect the Direct DBD more sig- Brussels, 2004.
Seismic Design of Frame-Wall Structures,
nificantly than the ELF approach. The [6] Sullivan TJ. An energy factor method for Research Report ROSE-2006/02. IUSS Press:
findings also illustrate that the design the displacement-based seismic design of RC Pavia, 2006.

Structural Engineering International 2/2013 Scientific Paper 131