Journal of Constructional Steel Research 127 (2016) 41–53

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Comparison of seismic design for steel moment frames in Europe, the

United States, Japan and China
Gang Shi ⁎, Fangxin Hu, Yongjiu Shi
Key Laboratory of Civil Engineering Safety and Durability of China Education Ministry, Department of Civil Engineering, Tsinghua University, Beijing 100084, PR China

a r t i c l e i n f o a b s t r a c t

Article history: Seismic provisions to guide the design of steel moment frames in Europe, the United States, Japan and China are
Received 28 September 2015 comprehensively examined. Seismic hazard levels and performance requirements, ground type classification,
Received in revised form 30 June 2016 magnitude and shape of elastic response spectra, seismic design force and distribution of required story shear
Accepted 11 July 2016
strength, local ductility requirements, and reduction factors are compared. The results show that the no-collapse
Available online xxxx
requirements in Eurocode and Japanese code correspond to a lower level of ground motion than the other two
Keywords:
codes. The unreduced elastic response spectra given in four codes are quite different in recognition of different
Seismic code ground types and seismicity, in particular, Japanese code generally specifies much larger elastic spectrum than
Seismic design other codes. Although local ductility requirements are quite similar, U.S. code specifies higher reduction factors
Steel structure than Eurocode and Japanese code, while Chinese code stipulates a constant reduction factor with a relatively
Steel moment frame small value regardless of the ductility level of structures. As a result of such over-conservatism, Chinese code de-
Earthquake engineering signed steel moment frames exhibit 20% to 150% larger lateral stiffness and resistance than Eurocode and U.S.
code in most cases, while the significant even larger lateral stiffness and resistance predicted by Japanese code
than that by Chinese code is mainly due to the larger seismic force from elastic response spectrum.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Steel moment frames were widely used for seismic design around
the world due to their highly regarded seismic performance under
earthquakes and relatively simple construction in practice. However,
such steel frames suffered much damage or even collapsed in the
1994 Northridge [1,2] and 1995 Kobe earthquakes [3,4], which
highlighted the need to thoroughly investigate seismic behavior of
steel framed structures. Since then, researchers and engineers have con-
tributed to giving a better understanding of those approaches in seismic
design and thus improving the provisions in seismic codes. Those efforts
have led to various improved seismic design practices for steel build-
ings, such as the innovative series of research by SAC to identify better
seismic details in beam-to-column connections [5], re-calibration of
seismic force reduction factor correlated with the expected ductility
[6] and new capacity design criteria to ensure a global plastic mecha-
nism [7]. It is notable that although current seismic codes in various
countries are based on similar fundamental seismic design principles,
the actual design procedures, detail strength and ductility requirements,
and consequently, the global seismic resistance and behavior exhibited
by steel structures can be quite different. It is therefore imperative that
design concepts and detailing rules in various seismic codes are ap-
praised and compared. Such information is deemed useful for
⁎ Corresponding author.
E-mail address: shigang@tsinghua.edu.cn (G. Shi).
researchers and engineers to understand better the seismic design prac-
tices in various countries. Some differences in the comparison may be
attributed to the engineering history, culture and economy, but others
may come from a lack of understanding or knowledge in the engineer-
ing philosophy for seismic design, even the recognition and experience
of practical earthquake damages. Thus, such a comparison can also re-
veal the potential opportunities to calibrate and re-evaluate the respec-
tive design provisions in different seismic codes.
To this end, this paper focuses on assessing and comparing the seis-
mic provisions adopted in Europe, the United States, Japan and China,
with emphasis on the seismic design approaches of steel moment
frames, since this structural system is the most basic seismic force
resisting system (SFRS). Some detailed assessment and comparison of
differences among the seismic codes in Europe, the United States and
Japan have been conducted by Uang [8], Mazzolani et al. [9], Nakashima
et al. [10], Tada et al. [11], Marino et al. [12] and Elghazouli [13,14].
However, as the seismic codes are constantly revised in light of recent
research findings and substantial developments, the previous compari-
son studies did not include seismic codes in Europe, the United States
and Japan simultaneously, and are not comprehensive, e.g. structural
performance requirement under different seismic hazard levels
(Section 2 below) and conversion relationships between the response
spectral parameters in different seismic codes (Section 3.2 below) are
not included. Moreover, there hasn't been any comparison including
the seismic code in China, which suffered much from several recent se-
vere earthquakes including the 2008 Wenchuan earthquake, the 2010

http://dx.doi.org/10.1016/j.jcsr.2016.07.009
0143-974X/© 2016 Elsevier Ltd. All rights reserved.
42 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53
Yushu earthquake, the 2013 Ya'an earthquake and the 2014 Ludian
earthquake. Therefore, the study in this paper provides a valuable over-
view and attempts to bridge differences of the design philosophy in
seismic codes in Europe, the United States, Japan and China.
A systematic and comprehensive comparison requires full charac-
terizations of the whole seismic design process, including the basic per-
formance requirements, seismic action for different limit states,
required strength, seismic force reduction factor and the corresponding
ductility requirements, and finally, drift limits. Currently, Eurocode 8
[15] (referred to as Eurocode hereinafter) and GB 50011-2010 [16] (re-
ferred to as Chinese code hereinafter) are the seismic codes applied in
Europe and China respectively, while in the United States, seismic de-
sign requirements of steel moment frames are included in ANSI/AISC
341-10 [17] and ASCE/SEI 7-10 [18] or IBC-2012 [19] (all referred to as
U.S. code hereinafter). Japan has a seismic design code adopted in
1981, which is called Building Standard Law [20], then the provisions
have been expanded to include limit state concepts and some revisions
have been made since then [21–24] (referred to as Japanese code here-
inafter). It should be stressed that the comparison is conducted on steel
moment frames located in a benchmark site of China, but designed per
different codes. Thus the difference in geography or geology of different
countries and its influence on design results can be excluded, leaving
behind only the difference in seismic design procedures or guidelines.
The ultimate goal of the presented study is to provide an insight into
the design philosophy in current seismic codes, in particular, the
trade-off between required strength and ductility, which is quite bene-
ficial for evaluating the strength and ductility capacity of steel moment
frames using high strength steels (especially in columns) and proposing
rational design alternatives for such steel frames that are both safe and
economical [25–27].
2. Performance requirements
Eurocode specifies two levels of ground motion. One is the reference
ground motion associated with a reference probability of exceedance
equal to 10% in 50 years or a return period of 475 years, under which
no local or global collapse of a structure is permitted (i.e. no-collapse re-
quirement which refers to ultimate limit state); the other one has a
probability of exceedance equal to 10% in 10 years or a return period
of 95 years and structures are designed to have sufficient resistance
and stiffness to maintain the function of vital services, without the oc-
currence of damage and the associated limitations of use (i.e. damage-
limitation requirement which refers to serviceability limit state).
U.S. code defines only one explicit level of seismic action based on a
recommended probability of exceedance of 2% in 50 years or a return
period of 2475 years, namely maximum considered earthquake (MCE)
ground motion, and stipulates that structures are designed to provide
an approximately uniform margin against collapse under such level of
ground motion throughout the United States [28]. The so-called seismic
margin is set at 1.5; consequently, the design level ground motion is de-
fined as 2/3 of MCE and is used to formulate the design response spec-
trum. Such design approach leads to a uniform margin against
collapse but not a uniform probability of the ground motion for seismic
design in different regions, which is quite different from Chinese code
and Eurocode employing the same level of design ground motion (i.e.
probability of exceedance of 10% in 50 years) in any region. For example,
the deign level ground motion in most regions of low-to-moderate seis-
micity in the central and eastern United States corresponds to a proba-
bility of exceedance of about 2% to 5% in 50 years, while in the western
United States of high seismicity (e.g. Los Angeles and San Francisco, Cal-
ifornia) the value of probability of exceedance is around 10% [29].
Japanese code explicitly considers two levels of ground motion. The
probability of exceedance of the Level 1 ground motion is equal to 50%
in 30 years (the return period is 43 years), while the probability of ex-
ceedance of the Level 2 ground motion is equal to 10% in 50 years (the
return period is 475 years). The former one is for serviceability
requirement and structural damage should be limited under this level
of ground motion. The latter one is for safety and no collapse of a struc-
ture is required.
Chinese code basically defines three-level seismic performance re-
quirements, i.e. “operational”, “damage-repairable” and “collapse-
prevention”, which essentially refer to serviceability, damageability
and survivability limit states respectively under seismic loading. The op-
erational and collapse-prevention requirement correspond to ground
motion based on a recommended probability of exceedance of 63%
and 2%–3% in 50 years or a return period of 50 and 1600–2400 years re-
spectively; whilst the values associated with the damage-repairable
level relate to a recommended probability of exceedance of 10% in
50 years or a return period of 475 years. To satisfy those requirements,
a two-phase design approach is employed. Phase 1 design is accom-
plished by performing an elastic analysis with the 63% in 50 years
ground motion and by assuring that critical structural elements are
below yield levels and the elastic inter-story drift should be within the
limiting values. Phase 2 design is required for some irregular or special
structures to ensure the inelastic inter-story drift under the 2%–3% in
50 years ground motion within the limiting value. Generally nonlinear
static (i.e. pushover analysis) or dynamic (i.e. time history analysis)
analysis is performed.
Those performance requirements by Eurocode, U.S., Japanese and
Chinese codes are compared in Fig. 1. It is clear that Chinese code is
the only one that defines an explicit three-level performance require-
ment; no-collapse requirement in Eurocode and Japanese code is less
stringent than that in the other codes, whilst the serviceability limit
state in Eurocode corresponds to a higher level of ground motion than
Chinese and Japanese codes. U.S. code doesn't consider any performance
requirement under frequent earthquakes.
3. Seismic action
In seismic design, an elastic pseudo-acceleration response spectrum
is developed to represent the ground motion, and such a spectrum is
correlated with the nature of supporting ground, i.e. different ground
types ranging from hard to soft soil result in different elastic response
spectra defined in each code. In order to compare seismic action levels
stipulated in different codes, correspondence among ground types is
established first; subsequently, elastic response spectra corresponding
to the equivalent ground type and the same level of ground motion
are compared.
3.1. Ground types
Although the descriptions of stratigraphic profile for various ground
types are more or less different, a common parameter which is used to
quantitatively classify ground conditions is the average shear wave ve-
locity Vs, which is computed as follows,
d
Vs ¼ ð1Þ
X
n
hi
i¼1
Vi
where n is the number of soil layers, hi and Vi denote respectively the
thickness and shear wave velocity of the i-th soil layer, and d represents
the total depth of considered soil layers. In Chinese code, d is taken as
the thickness of overburden soil layer or 20 m, whichever is less, and
five ground types named I0, I1, II, III, and IV are specified on the basis
of Vs; while in Eurocode, d equals to 30 m, then ground types A, B, C
and D are classified according to Vs, while ground type E corresponds
to a particular kind of ground stratigraphy in which a soft surface
layer (type C or D) is placed over a hard soil (type A), and there are an-
other two ground types S1 and S2 representing deposits or highly
liquefiable or sensitive soils. U.S. code also sets d as 30 m when
 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53 43

Fig. 1. Seismic hazard levels and performance requirements.

computing Vs, and defines five Site Classes A, B, C, D and E to classify 3.2. Elastic response spectra
ground types; another particular Site Class F is identified which requires
specific site response analysis. Eurocode defines elastic response spectrum Se (in terms of pseudo-
Another parameter to classify soil types in Japanese code is the fun- acceleration) as a function of the building natural period T, which is for-
damental period of foundation soil Tg, which is evaluated as follows, mulated as follows,
 
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
T
u 0≤T ≤T B Se ¼ ag S 1 þ ð2:5η−1Þ ð3aÞ
u H i−1 þ H i TB
u h
u X n i
2
T g ¼ t32 2
ð2Þ T B bT ≤T C Se ¼ 2:5Sηag ð3bÞ
i¼1 V i
 
TC
where n represents the number of soil layers, and hi, Hi and Vi represents T C bT ≤T D Se ¼ 2:5Sηag ð3cÞ
T
respectively the thickness, depth and shear wave velocity of the i-th soil
 
layer. Then three ground types are defined: type I (hard soil) if Tg ≤ 0.2, TCTD
type II (medium soil) if 0.2 b Tg ≤ 0.75 and type III (soft soil) if Tg N 0.75. T D bT ≤4 Se ¼ 2:5Sηag ð3dÞ
T2
Ground types defined by four codes are compared in Fig. 2, where
only ground types that can be quantified by Vs are included. Since differ- ag ¼ γ I ag;R ð3eÞ
ent total depths (d) are used by the former three codes and a totally dif-
ferent parameter is used by Japanese code to classify ground conditions, where ag,R is the reference PGA for ground type A with reference to the
the comparison has been carried out with reference to a unique soil 10% in 50 years ground motion (i.e. for no-collapse requirement in Fig.
layer with thickness equal to 30 m, placed over the hard rock. On the 1), and is established in national annexes on the basis of seismic risk
basis of this hypothesis, Vs and Vi defined in each code are identical. maps; γI is an importance factor depending on the importance class of
By substituting the limit values of Tg given by Japanese code into Eq. the considered structure, and is assumed to equal to 1.0 for ordinary
(1), the soil classification stated in Japanese code has been reformulated buildings; S is a soil factor and TB, TC, TD are characteristic periods of
in terms of Vs. It is therefore notable that due to quite different natural the response spectrum determined by ground type; η is damping cor-
site conditions, Chinese code gives finer classification for medium rection factor with a reference value of 1 for 5% viscous damping,
ground types, and the widespread ground type II in China incorporate which is commonly accepted for steel structures under design earth-
soils equivalent to type II in Japanese code, but with intermediate char- quakes. Table 1 shows the values of S, TB, TC and TD for different ground
acteristics between type B and C in Eurocode, or between type C and D types corresponding to Type 1 elastic response spectrum, which charac-
in U.S. code. terizes well the earthquakes with a surface-wave magnitude Ms N5.5.

Fig. 2. Ground type classification.

44 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53

Table 1 1:6T C
Characteristic parameters classified by ground types (in Eurocode). 2T C ≤T Rt ¼ ð5cÞ
T
Ground
A B C D E Generic spectrum where TC is determined according to ground type as shown in Table 3.
type
In Chinese code different regions are characterized by Mercalli in-
S 1.00 1.20 1.15 1.35 1.40
tensity scale to describe the effects of seismic action on structures. Seis-
TB 0.15 0.15 0.20 0.20 0.15 mic active regions are categorized into 4 groups with an intensity of 6, 7,
TC 0.40 0.50 0.60 0.80 0.50 8 and 9 respectively. For the purpose of comparison, elastic response
TD 2.00 2.00 2.00 2.00 2.00 spectra for far-field earthquakes corresponding to intensity-7 (i.e. low
seismicity), -8 (i.e. moderate seismicity) and -9 (i.e. high seismicity) re-
gions are considered. The reference peak ground acceleration PGA,
U.S. code incorporates elastic response spectrum Sa (in terms of ac-
which corresponds to the 10% in 50 years ground motion (i.e. for dam-
celeration of gravity) as a function of building natural period T, which
age-repairable requirement in Fig. 1), is specified as 0.1 g, 0.2 g and 0.4 g
is indicated as follows,
respectively in three regions. The spectrum value α is given as a fraction
  of acceleration of gravity and is defined as a function of building natural
T
0≤T ≤T 0 Sa ¼ SDS 0:4 þ 0:6 ð4aÞ period T, by means of the following expressions,
T0
 

0≤T ≤0:1 α ¼ 0:45 þ 10η2 −4:5 T α max ð6aÞ

T 0 bT ≤T S Sa ¼ SDS ð4bÞ
0:1 ≤T ≤T g α ¼ α max ð6bÞ
SD1
T S bT ≤T L Sa ¼ ð4cÞ  γ
T Tg
T g ≤T ≤5T g α¼ η2 α max ð6cÞ
T
SD1 T L
TNT L Sa ¼ ð4dÞ
T2  

5T g ≤T ≤6 α ¼ η2 0:2γ −η1 T−5T g α max ð6dÞ

2 2
SDS ¼ SMS ¼ F a SS ð4eÞ where αmax is maximum value of α depending on the level of ground
3 3
motion and intensity, and is set at 0.23, 0.45 and 0.90 in low, moderate
2 2 and high seismicity regions (i.e. intensity-7, -8 and -9) respectively with
SD1 ¼ SM1 ¼ F v S1 ð4fÞ reference to the 10% in 50 years ground motion; Tg is characteristic pe-
3 3
riod of the response spectrum depending on ground type; γ, η1 and η2
where SS and S1 are mapped MCE parameters at short period and 1 s pe- are damping correction factors (equal to 0.9, 0.02 and 1 for 5% viscous
riod respectively, while SMS and SM1 are adjusted values of SS and S1 for damping). Table 4 shows values of Tg for different ground types.
ground type effects, and SDS and SD1 are reduced values of SMS and SM1 In order to compare those elastic response spectra on the basis of the
corresponding to design earthquake; Fa and Fv are site coefficients de- same ground type and the same level of ground motion, conversion re-
termined by ground type; T0 and Ts are characteristic periods of the lationships have been established among spectrum parameters in four
ground motion and are computed as 0.2SD1/SDS and SD1/SDS respective- codes [30]. Because PGAs in Chinese code are determined with refer-
ly; TL is mapped long-period transition period. Table 2 shows values of ence to ground type II, the equivalent ag in Eurocode and SS, S1 in U.S.
Fa and Fv for different ground types. Note that both site coefficients de- code can be expressed as follows,
pend on values of mapped MCE parameters Ss and S1.
Japanese code stipulates the following expressions for elastic re- ag ¼ PGA=S ð7aÞ
sponse spectrum Rt (in terms of acceleration of gravity) with reference
SS ¼ 2:5PGAs =F a ð7bÞ
to the 10% in 50 years ground motion,
S1 ¼ 2:5T g PGAs =F v ð7cÞ
0≤TbT C Rt ¼ 1 ð5aÞ

 2 where PGA is the peak ground acceleration in Chinese code correspond-

T ing to the 10% in 50 years ground motion, which is set at 0.1 g, 0.2 g and
T C ≤Tb2T C Rt ¼ 1−0:2 −1 ð5bÞ
TC 0.4 g for low- (intensity-7), moderate- (intensity-8) and high-seismicity
(intensity-9) regions respectively, while PGAs is the one corresponding
to the 2% in 50 years ground motion, which is set at 0.22 g, 0.4 g and
Table 2 0.62 g for low-, moderate- and high-seismicity regions respectively; S
Characteristic coefficients classified by ground types (in U.S. code).
equals to 1.2 if ground type B in Eurocode is considered to be equivalent
Ground type A B C D E Generic spectrum to ground type II in Chinese code (see Fig. 2 and Table 1); Fa and Fv are
SS ≤ 0.25 0.80 1.00 1.20 1.60 2.50 taken as 1.2 and 1.65 respectively for low-seismicity, 1.0 and 1.5 respec-
tively for moderate-seismicity, and 1.0 and 1.3 respectively for high-
SS = 0.5 0.80 1.00 1.20 1.40 1.70
seismicity regions, if ground type C in U.S. code is considered to be
Fa SS = 0.75 0.80 1.00 1.10 1.20 1.20 equivalent to ground type II in Chinese code (see Fig. 2 and Table 2);
SS = 1.0 0.80 1.00 1.00 1.10 0.90 Tg is 0.45 for ground type II in Chinese code (see Table 4). In addition,
SS ≥ 1.25 0.80 1.00 1.00 1.00 0.90 Japanese code has introduced a seismic zone factor, Z, to consider differ-
ent seismicity all over the country, then ZRt is used for comparison with
S1 ≤ 0.1 0.80 1.00 1.70 2.40 3.50
the elastic spectrum in other codes considering different seismicity.
S1 = 0.2 0.80 1.00 1.60 2.00 3.20 Equivalent spectrum parameters have been obtained with the results
Fv S1 = 0.3 0.80 1.00 1.50 1.80 2.80 shown in Table 5. Comparison of elastic response spectra in four
S1 = 0.4 0.80 1.00 1.40 1.60 2.40 codes, therefore, is carried out with reference to hard soil (i.e. ground
types A, B, I, I0 in Eurocode, U.S., Japanese and Chinese codes respective-
S1 ≥ 0.5 0.80 1.00 1.30 1.50 2.40
ly), medium soil (i.e. ground types B, C, II and II in Eurocode, U.S.,
 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53
Table 3
Characteristic periods classified by ground types (in Japanese code).
Ground
type I II III Generic spectrum
Tc 0.4 0.6 0.8
2Tc 0.8 1.2 1.6
Japanese and Chinese codes respectively) and soft soil (i.e. ground types
D, E, III and IV in Eurocode, U.S., Japanese and Chinese codes respective-
ly) in low-, moderate- and high-seismicity (i.e. PGA = 0.1 g, 0.2 g and
0.4 g respectively) regions, as shown in Fig. 3. The abscissa is fundamen-
tal period (T1) and the ordinate is elastic response spectrum value (Se)
in terms of acceleration of gravity, which corresponds to the 10% in
50 years ground motion.
With respect to low or moderate-seismicity regions (see Fig. 3(a)
and (b)), Japanese code always specifies the largest elastic spectrum,
followed by U.S. code, and differences in seismic force are notable, e.g.
for medium soil in low-seismicity regions, the spectrum in U.S. code is
from 1.4 to 1.6 times as large as those in Chinese code and Eurocode
for structures with T1 b 2 s, and this difference is more significant for
soft soil (1.8 to 2.4 times) than that for hard soil (1.1 to 1.4 times); the
Japanese spectrum is even 3 to 6 times as large as the Chinese one for
hard or medium soils. With reference to high-seismicity regions (see
Fig. 3(c)), elastic spectra provided by four codes seem more comparable,
e.g. for hard or medium soils and structures with T1 b 2 s, seismic forces
stipulated in both Eurocode and U.S. code are either smaller or larger by
15% than that in Chinese code, and the Japanese spectrum is about 1 to 2
times as large as the Chinese one. But for soft soil, elastic spectrum in
Eurocode is larger by 17% to 30% than those in the other two codes for
structures with very short periods (T1 b 0.5 s), and then shows a sub-
stantial decrease with the increase of period, e.g. the elastic spectrum
in Eurocode can be smaller by 15% to 50% for 1.5 s b T1 b 3.0 s.
4. Required strength and ductility
Taking into account the ductile behavior of a steel structure, i.e. the
ability to deform plastically and cyclically without major loss of resis-
tance and to dissipate energy, a seismic force reduction factor (i.e. be-
havior factor q in Eurocode, response modification factor R in U.S.
code, or structural characteristic factor Ds in Japanese code) is incorpo-
rated in seismic design to reduce the seismic force determined by elastic
response spectrum which the structure should sustain if its response is
elastic. The value of reduction factor depends on the level of ductility
available. In this case, design criteria and detailed requirements aiming
to provide the structure with necessary ductility are stipulated. In what
follows, design procedures for required strength and ductility in
Eurocode, U.S., Japanese and Chinese codes are summarized. Distribu-
tion of seismic design forces along the building height, reduction factors
specified for steel moment frames, and provisions related to the ductil-
ity of individual structural members (limits on material properties of
steel and width-to-thickness ratios of cross-sections) are compared.
Table 4
Characteristic periods classified by ground types (in Chinese code).
Ground
I0 I1 II III IV Generic spectrum
type
Tg 0.30 0.35 0.45 0.65 0.90
5Tg 1.50 1.75 2.25 3.25 4.50
45
4.1. Distribution of seismic design forces
In order to give a simplified but quantitative comparison of seismic de-
sign forces in different codes, the lateral force method of analysis is con-
sidered herein. According to Eurocode, the lateral force method can only
be used for buildings meeting the criteria for regularity in elevation and
having a fundamental period (T1) not greater than 4TC (see Table 1 for
the definition of TC) or 2.0 s. The seismic base shear force (Fb) used for
elastic design corresponding to the 10% in 50 years ground motion, is
given by,
F b ¼ Sd mλ ð8Þ
where m is the total mass of the building estimated by taking into account
the presence of dead gravity load and a fraction of live gravity load, Sd is
the ordinate of design spectrum corresponding to fundamental period
of the building (T1), which is obtained by reducing the ordinate of elastic
response spectrum Se (see Eqs. (3a)–(3e)) by q-factor, and λ is a correc-
tion factor accounting for that in multi-story buildings the effective
modal mass of fundamental mode of vibration is smaller than total
mass. In particular, λ is equal to 0.85 if the building has more than two
stories and T1 b 2TC, or 1.0 otherwise. The seismic design forces (Fi) are
distributed along the height according to an inverted triangular distribu-
tion, and are given as follows,
zi mi
Fi ¼ Fb ð9Þ
X
N
z jm j
j¼1
where N is the number of stories, and mi and zi are the i-th story mass and
height measured from the foundation level.
According to U.S. code, generally the lateral force method is permit-
ted for structures with no vertical irregularities and not exceeding
48.8 m in height, or exceeding 48.8 m in height but with a fundamental
period (T1) not N3.5Ts (Ts = SD1/SDS, see Eqs. (4e) and (4f)). The seismic
base shear force used for elastic design corresponding to design earth-
quake, is determined by the following equation,
V ¼ Cs W ð10aÞ
Sa S1
Cs ¼ ≥ maxf0:044SDS I e ; 0:01g and 0:5 ðif S1 ≥0:6g Þ ð10bÞ
R=Ie R=I e
where Cs is a seismic response coefficient, W is the effective seismic
weight including dead load and a fraction of live load, Sa is the ordinate
of elastic response spectrum and SDS and S1 are spectrum parameters
(see Eqs. (4a)–(4f)), R is a response modification factor determined in
accordance with the ductility level of structures, Ie is an importance fac-
tor and is assumed to be 1.0 for ordinary buildings. The seismic design
force (Fx) induced at any level x is calculated as follows,
k
hx wx
Fx ¼ V ð11Þ
X
N
k
hi wi
i¼1
where N is the number of stories, and wx and hx are the x-th story seis-
mic weight and height measured from the foundation level; the expo-
nent k, which takes into account higher mode effects, is related to
fundamental period (T1) of the structure. For T1 ≤ 0.5 s, k = 1; for
T1 ≥ 2.5 s, k = 2; for 0.5 s b T1 b 2.5 s, k shall be 2 or determined by linear
interpolation between 1 and 2.
According to Japanese code, a two-level design is adopted to calcu-
late required strength. The stresses due to frequent earthquakes (see
Fig. 1) should not exceed allowable stresses (Level 1 design). In addi-
tion, ultimate lateral strength of each story of the structure has to be
larger than design story shear due to moderate earthquakes (Level 2
46 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53

Table 5 X
N
wj
Equivalent spectrum parameters in four codes. αi ¼ ð13bÞ
j¼i
W
Chinese code Eurocode U.S. code Japanese code

Seismicity α ag (g) SS (g) S1 (g) Z where W is total seismic weight of the building, and higher mode effects
Low 0.23 0.083 0.458 0.150 0.7
are taken into account. The Level 2 seismic shear forces, or the required
Moderate 0.45 0.167 1.000 0.300 0.8 ultimate strength are calculated as,
High 0.90 0.333 1.550 0.537 1.0
V 2un;i ¼ Ds;i F es;i V 2i ð14aÞ
design). The Level 1 seismic shear force of the i-th story (V1i ) is stipulat-
ed as follows, X
N
V 2i ¼ C 2i wj ð14bÞ
j¼i
1
X
N
V 1i C i wj ð12aÞ
j¼i C 2i ¼ ZRt Ai C 20 ð14cÞ

C 1i ¼ ZRt At C 10 ð12bÞ where Ds,i is a structural characteristic factor which is conceptually the
reciprocal of q-factor in Eurocode or that of R-factor in U.S. code, Fes,i is
where N is the number of stories, wj is the j-th story seismic weight, C1i is a shape factor which is taken as 1.0 for regular structures, C2i is a Level
a Level 1 i-th story shear coefficient, C10 is equal to 0.2, Z is a seismic zone 2 i-th story shear coefficient, and C20 is equal to 1.0.
factor (see Table 5), Rt is the elastic spectrum value corresponding to According to Chinese code, such a method is applicable to structures
fundamental period of the structure (T1), and Ai is a height distribution not exceeding 40 m in height, with dominant shear drift and uniformly
factor which is given as, distributed story mass and stiffness along height. The seismic base shear
  force (FEk) used for elastic design is determined as follows,
1 2T 1
Ai ¼ 1 þ pffiffiffiffiffi −α i ð13aÞ
αi 1 þ 3T 1 F EK ¼ αGeq ð15Þ

Fig. 3. Elastic response spectra.

 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53
where Geq is equivalent seismic weight which is taken as 100% and 85%
of total seismic weight including dead load and a fraction of live load for
single- and multi-degree-of-freedom systems respectively; α is the or-
dinate of elastic response spectrum (see Eqs. (6a)–(6d)) corresponding
to fundamental period of the structure (T1), but is determined with ref-
erence to the 63% in 50 years ground motion. The seismic design forces
(Fi) are calculated by,
Hi Gi
F i ¼ F Ek ð1−δn Þ
X
N
H jG j
j¼1
ΔF n ¼ δn F Ek
where N is the number of stories, Gi and Hi are the i-th story seismic
weight and height measured from the foundation level, ΔFn is the
added seismic design force at the top of the structure, and δn is such a
factor accounting for higher mode effects determined by Table 6,
where the characteristic period Tg is defined in Table 4.
For ten-story buildings located in medium soil with uniform mass and
stiffness distribution along height, the vertical distribution of story shear
strength required by Eurocode, U.S., Japanese and Chinese codes have
been evaluated and the comparison is shown in Fig. 4 for three funda-
mental periods (T1 = 0.5 s, 1.0 s and 2.0 s). The abscissa is required
story shear strength normalized by base shear force (Vi/Vb), and the ordi-
nate is number of stories. It shows that for a short-period building (T1 =
0.5 s), distributions of design story shear strength in Eurocode, U.S. and
Chinese codes are the same and they are very close to that in Japanese
code; however, for intermediate- and long-period buildings (T1 = 1.0 s
and 2.0 s), both U.S., Japanese and Chinese codes always specify larger
story shear strength than Eurocode at any story, with the largest differ-
ence observed at the top story, e.g. for T1 = 1.0 s, the top story shear
strength provided by U.S., Japanese and Chinese codes are 1.41, 1.11 and
1.56 times as large as that in Eurocode respectively. This is because
Eurocode doesn't consider higher mode effects in calculating the distribu-
tion of seismic design forces along height in this case. In addition, since
Chinese code adds an individual seismic design force at the top story, it
provides the largest top story shear strength for long-period buildings
(T1 = 2.0 s).
4.2. Reduction factor
In Eurocode, earthquake resistant steel moment frames are classified
into three ductility classes, i.e. low (DCL), medium (DCM) and high
(DCH) ductility classes according to the ductility level of dissipative mem-
bers. This classification is based on two design concepts that can be used
in seismic design. The first design concept is low dissipative structural be-
havior, and the structure belongs to DCL with a q-factor not N 1.5 in recog-
nition of a minimal level of inherent over-strength and ductility. In this
case, global elastic analysis can be utilized and resistance of members
and connections are evaluated according to Eurocode 3 [31] without
any additional seismic requirements. The second design concept is dissi-
pative structural behavior which takes into account the capability of a
structure to resist earthquakes through inelastic behavior of its members
or connections, and the structure belongs to DCM or DCH for which ca-
pacity design procedures and corresponding detailing are required. q-fac-
tor is then established on the basis of structural type and ductility class as
summarized in Table 7, where the ratio αu/α1 depends on the ultimate-
Table 6
Values of δn (Chinese code).
T1 N 1.4Tg T1 ≤ 1.4Tg
Tg ≤ 0.35 s 0.08T1 + 0.07 0.0
0.35 s b Tg ≤ 0.55 s 0.08T1 + 0.01
Tg N 0.55 s 0.08T1 − 0.02
47
to-first plasticity resistance ratio representing the redundancy of the
structure. The value of this ratio is recommended according to geometri-
cal scheme of the structure, i.e. 1.1 for portal frames, 1.2 for single-bay and
multi-story steel moment frames, and 1.3 for multi-bay and multi-story
steel moment frames, or it can be determined from pushover analysis
but should not exceed 1.6.
In U.S. code, steel moment frames are categorized into three types,
i.e. ordinary (OMF), intermediate (IMF) and special (SMF) moment
frames according to the level of inelastic deformation capacity in their
ð16aÞ
members and connections. OMF is expected to provide minimal inelas-
tic deformation capacity and can be designed according to ANSI/AISC
360-10 [32] only. This is similar to DCL in Eurocode. IMF and SMF are de-
ð16bÞ signed to provide limited and significant inelastic deformation capacity
respectively. R-factors assigned in U.S. code for each type of steel mo-
ment frames are summarized in Table 7.
In Japanese code, steel moment frames are classified according to the
ductility level which depends on the width-to-thickness ratio of beams
and columns. In particular, Japanese code identifies four moment frame
classes, namely FA, FB, FC and FD, which correspond to high, good, fair
and poor ductility, respectively. Ds-factor is established based on this clas-
sification, with the value of its reciprocal shown in Table 7 for the purpose
of comparison.
However, in Chinese code the reduction factors for seismic force in
various types of structures corresponding to different ductility levels are
not explicitly defined. As introduced in Section 2, Chinese code introduces
a two-phase design approach in which the first phase is accomplished by
global elastic analysis with the 63% in 50 years ground motion directly,
rather than with a reduced one from the 10% in 50 years ground motion.
Chinese code stipulates that the maximum elastic response spectrum
values (αmax) under the former ground motion are approximately 1/3
of those under the latter one. This implies that reduction factors are
taken as 3 for all kinds of steel frames. Obviously, this is neither reasonable
nor economical enough since ductility level differs indeed with various
types of systems, e.g. steel moment frames and concentrically braced
frames. Hence, there are no correspondence between the reduction factor
and seismic detailing requirements for members and connections as well.
Reduction factor with its constant value of 3 in Chinese code is also includ-
ed in Table 7 for the purpose of comparison.
For multi-bay and multi-story steel moment frames, reduction fac-
tors stipulated in four codes are compared in Fig. 5. It shows a significant
difference in steel moment frames with relatively high ductility. Reduc-
tion factors specified in Eurocode and U.S. code can be 1.7 to 2.7 times as
large as those in Japanese and Chinese codes.
4.3. Reduced seismic forces
Since elastic response spectra and reduction factors in different codes
have been clarified in previous sections, the reduced seismic forces calcu-
lated by Eqs. (8)–(16a) and (16b) are compared in Fig. 6 for typical ten-
story buildings located in medium soil of moderate-seismicity regions.
The ordinate is seismic design base shear force (Vb) normalized by seismic
weight (W), while the abscissa indicates different types of moment
frames. It is apparent that Japanese code always specifies the largest seis-
mic design force, except for low-ductility short-period buildings (T1 =
0.5 s) in which case seismic design force in Eurocode is slightly larger
(see Fig. 6(a)). For moderate- and high-ductility buildings, Eurocode stip-
ulates comparable seismic design force to U.S. code, and Chinese code
gives larger seismic design force than Eurocode and U.S. code, in particu-
lar, for short- or intermediate-period buildings with high ductility (see
Fig. 6(a) and (b)).
4.4. Local ductility
Generally, structural steels that are explicitly permitted for use in seis-
mic applications should have good weldability and impact toughness, and
moreover, a substantial inelastic deformation capacity which can be
48 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53

Fig. 4. Distribution of required story shear strength of ten-story buildings along height.

described by the yield-to-tensile strength ratio (fy/fu), elongation after
rupture (δ) and ultimate strain (εu) corresponding to the ultimate
strength (fu) as shown in Table 8, where εy is the yield strain correspond-
ing to the nominal yield strength (fy). It is obvious that ductility require-
ments for structural steels in U.S. and Chinese codes are basically the
same and more stringent than those in Eurocode. Besides, it is worth
stressing that, the maximum nominal yield strength of structural steels
allowed in Chinese code is 420 MPa; while in Eurocode and U.S. code,
some high strength steels with fy ≥ 460 MPa have been introduced, such
as S690 and ASTM A514 steels, but there are additional limitations. In
Eurocode, for high strength steels with fy N 460 MPa individual ductility
requirements different from those in Table 8 are recommended; the lim-
iting value of the ratio fu/fy is now 1.05, the elongation (δ) should not be
b10%, but the ultimate strain (εu) should still be greater than 15εy. How-
ever, U.S. code stipulates that the specified nominal yield strength of
structural steels shall not exceed 345 MPa for IMFs and SMFs, while this
limit is 380 MPa for OMFs, which directly prohibits the use of high
strength steels in earthquake resistant steel moment frames. Japanese
code doesn't explicitly stipulate the ductility requirement for structural
steels used in seismic design. But in general, according to the steel produc-
tion standard in Japan, fy/fu should not exceed 0.8, while the elongation
(δ) should be at least 20% for most structural steels like SN400, SN490
and high performance steels SA440.
Local ductility in steel members and connections should be consistent
with the expected demand under severe earthquakes, and thus shall be
closely related to the ductility class and reduction factor adopted in seis-
mic design. According to Eurocode, member cross-sections are similarly
classified in four classes, namely 1, 2, 3 and 4 on the basis of width-to-
thickness ratios. For buildings belonging to DCL, any class of cross-sections
can be used; however, for buildings belonging to DCM and DCH, cross-
section classes allowed depend on the q-factor adopted (for 1.5 b q ≤ 2,
class 1, 2 and 3 are allowed; for 2 b q ≤ 4, class 1 and 2 are allowed; for
q N 4, only class 1 is allowed). U.S. code introduces two types of ductile
members, namely moderately ductile (MD) and highly ductile (HD)
based on the cross-sectional width-to-thickness ratios. For OMFs, there
are no limitations on members except those basis requirements in
ANSI/AISC 360-10; however, for IMFs and SMFs, requirements for moder-
ately and highly ductile members shall be satisfied respectively. In Japa-
nese code, the member cross-sections in type FA, FB, FC and FD frames
have to fulfill different limits for the width-to-thickness ratio. In Chinese
code, although the implicit reduction factor adopted in seismic design is
set at 3 for all kinds of steel moment frames, the member cross-sections
are still classified in five grades, namely 1, 2, 3, 4 and 5, with reference
to the width-to-thickness ratio of flanges and web. Those cross-sectional
grades are related to seismic design category of the building, and the
higher and more important is the building, to the lower grade are mem-
ber cross-sections assigned.
Width-to-thickness requirements provided by four codes are com-
pared in Fig. 7 for wide flange cross-sections. The symbols c, tf, d and tw
represent the half width and thickness of flange, the inner depth and
thickness of web, respectively. The symbol ε is defined as follows,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ε¼ 235=f y ð17Þ
where fy is nominal yield strength of the used steel (in MPa). In Chinese
code, the limiting value of width-to-thickness ratio of beam webs is

Table 7
Reduction factors.

Seismic design code Reduction factor

Eurocode 1.5 (DCL) 4 (DCM) 5αu/α1 (DCH)

U.S. code 3.5 (OMF) 4.5 (IMF) 8 (SMF)
Japanese code 2.5 (FD) 3.33 (FB) 4 (FA)
2.86 (FC)
Chinese code 3
Fig. 5. Reduction factors for steel moment frames.
 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53
given as a function of the ratio of beam axial force to its axial strength (nb),
while this case is specified for webs of both beams and columns in
Eurocode and U.S. code, where the stress distribution within the cross-
section (i.e. α or φ) and the ratio of beam or column axial force to its
axial strength (i.e. nb or nc) are both used to define the limits of width-
to-thickness ratio, respectively. Note that α represents the ratio between
the depth of web part in tension and the outer depth of the cross-section
for class 1 or 2 cross-sections, varying from 0 (pure axial compression) to
0.5 (pure flexure), while φ is the ratio of the maximum stress of web part
in tension to that of web part in compression for class 3 or 4 cross-sec-
tions, varying from 1 (pure axial compression) to −1 (pure flexure).
Generally, four codes provide substantially similar limits with regard
to the width-to-thickness ratio of flanges of both beams and columns,
while the limits in Eurocode and Japanese code are somewhat less strin-
gent for flanges that are not so ductile (see Fig. 7(a) and (b)). Limits of
the width-to-thickness ratio for webs in Chinese code are basically the
strictest, except for columns under pure compression in which case
Eurocode gives the most stringent requirements. It is noteworthy that,
the current Chinese and Japanese codes doesn't relate limits for column
webs to the axial load ratio (nc), which is not rational. Nevertheless, ener-
gy dissipation of steel moment frames is developed primarily by yielding
of beam flanges; hence comparable member ductility is expected for var-
ious cross-sectional classifications in four codes.
In addition to member ductility, both Eurocode and U.S. code specify
minimum requirements in deformation capacity of connections in ductile
steel frames. In Eurocode, the rotation capacity of plastic hinge region (θp)
shall not be b0.025 rad for structures of DCM and 0.035 rad for structures
of DCH with q-factor adopted N2, while U.S. code stipulates that, for IMFs
and SMFs, beam-to-column connections shall be capable of accommodat-
ing a story drift angle of at least 0.02 rad and 0.04 rad respectively, and the
flexural resistance of connections determined at the column face shall
equal to at least 80% of the full plastic moment resistance of connected
beam at this story drift angle for both frame types. Those requirements
for connection ductility are quite similar in recognition of the elastic rota-
tion which is generally in the order of 0.005 to 0.01 rad. However, in Chi-
nese or Japanese code no such requirements have been provided.
5. Lateral stiffness of steel moment frames
According to Chinese code, the elastic story drift due to the 63% in
50 years ground motion (frequent earthquakes in Fig. 1) and the inelas-
tic one due to the 2% in 50 years ground motion (severe earthquakes in
Fig. 6. Seismic design forces of ten-story buildings.
49
Fig. 1) cannot be greater than the corresponding limiting values as
shown in Table 9. But in Eurocode, such limits depend on the type of
non-structural elements for their ability to accommodate the story
drifts without damage, and are corresponding to the 10% in 10 years
ground motion (see Table 9). Analogously, Japanese code provides
story drift limits due to the Level 1 seismic forces (i.e. the 50% in
30 years earthquakes shown in Fig. 1). Two limit values are given, and
the former one is stricter and commonly used, whereas the latter ap-
plies when non-structural elements can tolerate relatively large defor-
mations of the structure without damage, as shown in Table 9. U.S.
code, however, provides limits that have to be fulfilled due to design
earthquakes (see Fig. 1) regardless of types of non-structural elements
(see Table 9). It follows that the elastic story drift requirement in Chi-
nese code (0.4%) is comparable to that with brittle non-structural ele-
ments in Eurocode (0.5%) and the commonly used one in Japanese
code (0.5%), while the inelastic limit in Chinese code (2%) is much
more stringent than that in U.S. code (2%) since the former code stipu-
lates much more severe ground motion in this case.
By combing the story drift requirements in Table 9 with the seismic
force introduced in Section 3.2, the ratio of required stiffness (K) in four
codes is evaluated for ordinary buildings (i.e. importance factor equals
to 1.0) as follows (superscripts C, E, A and J refer to Chinese code,
Eurocode, U.S. code and Japanese code, respectively),
0:85  α=3 v  λSe =g C d =R  Sa =g ZRt C 10
KC : KE : KA : K J ¼ : : : ð18Þ
ΔC ΔE ΔA ΔJ
where α, Se, Sa, Rt, λ, R, Z and C10 are defined in Sections 3.2, 4.1 and 4.2; v
is a reduction factor which takes into account the lower level of seismic
force associated with damage-limitation requirement and is set at 0.5
for ordinary buildings; Cd is an displacement amplification factor
which takes into account the larger deformation under design earth-
quakes than the elastic one calculated by reduced seismic forces, and
its value is 3.0, 4.0 and 5.5 for OMFs, IMFs and SMFs respectively; ΔC,
ΔE, ΔA and ΔJ are limiting story drifts in Chinese code, Eurocode, U.S.
code and Japanese code respectively, and are taken as 0.40%, 1.00%,
2.00% and 0.50% without consideration of non-structural elements
(see Table 9). Eq. (18) is shown in Fig. 8 with reference to hard, medium
and soft soils in low-, moderate- and high-seismicity regions, where the
ratio of lateral stiffness is a function of fundamental period of the struc-
ture (T1). The results make it clear that, Japanese code is always the
most stringent in low- or moderate-seismicity regions, followed by
50 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53

Table 8
Ductility requirements for structural steels.

Ductility index Eurocode U.S. code Japanese code Chinese code

fy/fu ≤1/1.1 (≈0.91) ≤0.85 (for A992 and A913 steels), with a pronounced yield plateau ≤0.80 (for most steels) ≤0.85, with a pronounced yield plateau
δ ≥15% ≥20% ≥20% (for most steels) ≥20%
εu ≥15εy – – –

Chinese code, e.g. for 1.0 s b T1 b 2.0 s and for medium soil in moderate-
seismicity regions, the stiffness required by Japanese code is 80% to
100% larger than that required by Chinese code, which is still 30% to
70% larger than those required by Eurocode and U.S. code. Such a signif-
icant difference is due to much larger elastic spectrum values given by
Japanese code (see Fig. 3(a) and (b)). In high-seismicity regions (see
Fig. 8(c)), for short- and long-period systems (T1 b 0.5 s and
T1 N 2.0 s), Chinese code is more stringent than other codes, while for in-
termediate-period systems (0.5 s b T1 b 2.0 s), Japanese code is still the
most stringent. Generally, Eurocode is comparable to U.S. code.
steel moment frames is governed by story drift requirements due to
their great flexibility. As a result, lateral resistance (V) of steel moment
frames designed according to Chinese code, Eurocode, U.S. code and Jap-
anese code can be evaluated. Supposing that lateral stiffness (K) and lat-
eral resistance (V) is proportional to the geometrical moment of inertia
and the plastic modulus of members respectively, according to Marino
et al. [12], the ratio between the actual lateral resistances in four codes
is given by (superscripts C, E, A and J refer to Chinese code, Eurocode,
U.S. code and Japanese code, respectively),
h i34   γ0 −1
34 h i3 h i3
V C : V E : V A : V J ¼ K C : K E 1− : KA : K J
4 4

0
s ð19Þ
6. Lateral resistance of steel moment frames γ

In reality, based on the required strength and lateral stiffness evalu- where s represents the contribution of columns to total lateral deforma-
ated in previous sections, it is generally accepted that seismic design of tion and is assumed to equal to 0.2 based on the best practices followed

Fig. 7. Width-to-thickness requirements for wide flange cross-sections.

 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53 51

Table 9 the strongest in most cases. In low-seismicity regions (see Fig. 9(a)), Jap-
Story drift requirements. anese code designed steel moment frames are 1.6 to 2.4 times as strong as
Code Permitted story drift Scenario those designed by Chinese code, while this difference is 1.0 to 1.6 times
Eurocode 0.5% Brittle non-structural element, under
and 0.7 to 1.2 times in moderate- and high-seismicity regions (see Fig.
the 10% in 10 years earthquakes 9(b) and (c)), respectively. Generally Eurocode and U.S. code designed
0.75% Ductile non-structural element, under steel moment frames exhibit about 20% to 60% lower strength than that
the 10% in 10 years earthquakes designed by Chinese code, except for structures with T1 b 0.6 s and for
1% No interfering, under the 10% in
soft soil in low-seismicity regions (see Fig. 9(a)), in which case the lateral
10 years earthquakes
U.S. code 2% Under design earthquakes, i.e. 2/3MCE resistance required by U.S. code is about 1.34 times that required by Chi-
Japanese code 0.5% Commonly used, under Level 1 earthquakes nese code. It is also apparent that Eurocode designed moment frames ex-
0.83% Drift tolerant, under Level 1 earthquakes hibit the lowest lateral resistance.
Chinese code 0.4% Under the 63% in 50 years earthquakes As discussed in Section 4.3, local ductility requirements that control
2% Under the 2%–3% in 50 years earthquakes
deformation capacity of members are similar among Eurocode, U.S., Japa-
nese and Chinese codes. However, Chinese code stipulates a constant re-
duction factor (i.e. 3) for all steel frames with different level of ductility
by design firms, γ′ represents the amplification factor of the moment of requirements, and this value is much smaller than those in Eurocode
inertia of columns involved by capacity design criteria in Eurocode and and U.S. code for ductile steel moment frames, as noted in Section 4.2.
is assumed to equal to 1.53 [12]. Thus Eq. (19) can also be reported in The contradictory situation comes from over-conservatism in the trade-
Fig. 9 with reference to hard, medium and soft soils in low-, moderate- off between strength and ductility adopted by Chinese code. Such over-
and high-seismicity regions, where the ratio of lateral resistance is a conservatism is the major source of remarkable larger lateral stiffness
function of fundamental period of the structure (T1). and resistance exhibited by steel moment frames designed to Chinese
Fig. 9 shows that, as a consequence of significantly larger spectrum code than those designed to Eurocode and U.S. code, as shown in Figs. 8
values in Japanese code, steel moment frames designed by that code are and 9. The significant much larger lateral stiffness and resistance

Fig. 8. Lateral stiffness of steel moment frames.

52 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53
Fig. 9. Lateral resistance of steel moment frames.
predicted by Japanese code, however, is mainly due to larger seismic
forces, i.e. the elastic response spectra as shown in Fig. 3.
7. Conclusions
This paper examines and compares the seismic provisions stipulated
in Eurocode, U.S., Japanese and Chinese codes for steel moment frames.
The following concluding remarks can be made.
(1) Chinese code is the only one that adopts an explicit three-level
performance requirement, and the no-collapse requirement in
Eurocode and Japanese code is less stringent than that in the
other two codes. The serviceability limit state in Eurocode, how-
ever, corresponds to a higher level of ground motion than Japa-
nese and Chinese codes.
(2) Hard ground types (type A in Eurocode, type A and B in U.S. code,
type I in Japanese code and type I0 in Chinese code) and soft
ground types (type D in Eurocode, type E in U.S. code, type III in
Japanese code and type IV in Chinese code) include substantially
the same soils respectively in four codes, while Chinese code
gives finer classification for medium soil.
(3) The unreduced elastic response spectra corresponding to the
moderate earthquakes (i.e. ground motion with a probability of
exceedance of about 10% in 50 years) given in four codes show
notable differences. In particular, Japanese code always specifies
much larger elastic spectrum than other codes in low- or moder-
ate-seismicity regions; while in high-seismicity regions, elastic
spectra provided by four codes become more comparable.
(4) For relatively short-period buildings (T1 ≤ 0.5 s), the distributions
of required story shear strength in Eurocode, U.S. and Chinese
codes are the same and they are close to that in Japanese code;
however, for intermediate- and long-period buildings (T1 =
1.0 s and 2.0 s), U.S., Japanese and Chinese codes always require
larger story shear strength than Eurocode, especially at the top
story.
(5) Local ductility requirements including the material properties
and width-to-thickness limits of flanges stipulated in four
codes are similar. Chinese code generally gives the most strin-
gent width-to-thickness limit for webs, but without consider-
ation of the effect of axial load on columns.
(6) U.S. code always specifies larger reduction factors than Eurocode
and Japanese code for steel moment frames, while the constant
reduction factor in Chinese code ignores the different required
strength based on different ductility requirements and its rela-
tively small value is apparently over-conservative for ductile mo-
ment frames.
(7) Japanese code always specifies the largest seismic design force in
most cases, while for moderate- and high-ductility buildings
 G. Shi et al. / Journal of Constructional Steel Research 127 (2016) 41–53
Eurocode stipulates comparable seismic design force to U.S. code.
Chinese code gives larger seismic design force than Eurocode and
U.S. code especially for short- or intermediate-period buildings
with high ductility.
(8) Concerning lateral stiffness requirement, Japanese code is always
the most stringent in low- or moderate-seismicity regions,
followed by Chinese code. But in high-seismicity regions, for
short- and long-period systems Chinese code is more stringent
than other codes, while for intermediate-period systems, Japa-
nese code is still the most stringent. Eurocode is comparable to
U.S. code.
(9) Since seismic design of steel moment frames is generally con-
trolled by serviceability requirements, the lateral resistance exhib-
ited is highly correlated with the lateral stiffness. Japanese code
designed moment frames are the strongest in most cases, and
those designed by Eurocode and U.S. code generally exhibit
about 20% to 60% lower strength than the one designed by Chinese
code.
(10) Generally, the over-conservatism in small values of reduction fac-
tor stipulated in Chinese code is the major source of the remark-
able larger lateral stiffness and resistance of steel moment frames
than those designed to Eurocode and U.S. codes, while the signifi-
cant even much larger lateral stiffness and resistance predicted by
Japanese code is mainly due to the larger reduced seismic forces
from the elastic response spectrum.
Acknowledgements
The authors acknowledge the support for this work, which was
sponsored by the National Natural Science Foundation of China (No.
51478244), and the Excellent Young Scientist Programme of the Nation-
al Natural Science Foundation of China (No. 51522806).
The corresponding author also gratefully acknowledges the Japan
Society for the Promotion of Science (JSPS) for giving him generous fi-
nancial support to visit Japan as a JSPS short-term invitation fellow
with the ID No. S-14070 and conduct the research reported in this
paper.
Prof. Masayoshi Nakashima from Disaster Prevention Research Insti-
tute, Kyoto University, Japan, has given many valuable suggestions to
this paper, and his contribution is gratefully appreciated.
References
[1] D.K. Miller, Lessons learned from the Northridge earthquake, Eng. Struct. 20 (4–6)
(1998) 249–260.
[2] S.A. Mahin, Lessons from damage to steel buildings during the Northridge earth-
quake, Eng. Struct. 20 (4–6) (1998) 261–270.
[3] M. Nakashima, K. Inoue, M. Tada, Classification of damage to steel buildings ob-
served in the 1995 Hyogoken-Nanbu earthquake, Eng. Struct. 20 (4–6) (1998)
271–281.
[4] E. Watanabe, K. Sugiura, K. Nagata, Y. Kitane, Performances and damages to steel
structures during the 1995 Hyogoken-Nanbu earthquake, Eng. Struct. 20 (4–6)
(1998) 282–290.
53
[5] FEMA-350, Recommended Seismic Design Criteria for New Steel Moment-Frame
Buildings, SAC Joint Venture, Sacramento, California, U.S., 2000
[6] L. Sanchez-Ricart, A. Plumier, Parametric study of ductile moment-resisting steel
frames: a first step towards Eurocode 8 calibration, Earthq. Eng. Struct. Dyn. 37
(7) (2008) 1135–1155.
[7] F.M. Mazzolani, V. Piluso, Plastic design of seismic resistant steel frames, Earthq. Eng.
Struct. Dyn. 26 (2) (1997) 167–191.
[8] C.M. Uang, Comparison of seismic force reduction factors used in USA and Japan,
Earthq. Eng. Struct. Dyn. 20 (4) (1991) 389–397.
[9] F.M. Mazzolani, V. Piluso, Theory and Design of Seismic Resistant Steel Frames, E&
FN Spon, London, U.K., 1996
[10] M. Nakashima, C.W. Roeder, Y. Maruoka, Steel moment frames for earthquakes in
United States and Japan, J. Struct. Eng. ASCE 126 (8) (2000) 861–868.
[11] M. Tada, T. Fukui, M. Nakashima, C.W. Roeder, Comparison of strength capacity for
steel building structures in the United States and Japan, Earthq. Eng. Eng. Seismol.
4 (1) (2003) 37–49.
[12] E.M. Marino, M. Nakashima, K.M. Mosalam, Comparison of European and Japanese
seismic design of steel building structures, Eng. Struct. 27 (6) (2005) 827–840.
[13] A.Y. Elghazouli, Assessment of capacity design approaches for steel-framed struc-
tures, Int. J. Steel Struct. 5 (5) (2005) 465–475.
[14] A.Y. Elghazouli, Assessment of European seismic design procedures for steel framed
structures, Bull. Earthq. Eng. 8 (1) (2010) 65–89.
[15] Eurocode 8, Design of structures for earthquake resistance - part 1: general rules,
seismic actions and rules for buildings, EN1998-1-2004, European Committee for
Standardization, Brussels, 2004.
[16] GB 50011-2010, Code for Seismic Design of Buildings, China Architecture & Building
Press, Beijing, 2010 (in Chinese).
[17] ANSI/AISC 341-10, Seismic Provisions for Structural Steel Buildings, AISC, Chicago, IL,
2010.
[18] ASCE/SEI 7-10, Minimum Design Loads for Buildings and Other Structures, ASCE,
Reston, Virginia, 2010.
[19] IBC-2012, International Building Code, International Code Council, Country Club
Hills, IL, 2011.
[20] T. Nakano, The Historical Development of Seismic Design Load for Buildings and its
Up-to-date Trends, NTT Building Technology Institute, 2006 (in Japanese).
[21] BCJ, Structural Provisions for Building Structures - 1997 Edition, Building Center of
Japan, Tokyo, Japan, 1997 (in Japanese).
[22] AIJ, Guidelines for Limit State Design of Steel Structures, The Architectural Institute
of Japan, Tokyo, Japan, 1998 (in Japanese).
[23] Steel Design Manual, JFE Steel Corporation, Japan, 2010 (in Japanese).
[24] M. Nakashima, M. Kato, T. Okazaki, Collaboration between practice and research for
development of steel construction in Japan, Int. J. Steel Struct. 4 (2004) 249–262.
[25] G. Shi, F.X. Hu, Y.J. Shi, Recent research advances of high strength steel structures
and codification of design specification in China, Int. J. Steel Struct. 14 (4) (2014)
873–887.
[26] G. Shi, J.J. Wang, Y. Bai, Y.J. Shi, Experimental study on seismic behavior of 460 MPa
high strength steel box-section columns, Adv. Struct. Eng. 17 (7) (2014) 1045–1059.
[27] G. Shi, M. Wang, Y. Bai, F. Wang, Y.J. Shi, Y.Q. Wang, Experimental and modeling
study of high-strength structural steel under cyclic loading, Eng. Struct. 37 (4)
(2012) 1–13.
[28] E.V. Leyendecker, R.J. Hunt, A.D. Frankel, K.S. Rukstales, Development of maximum
considered earthquake ground motion maps, Earthquake Spectra 16 (1) (2000)
21–40.
[29] G.J.P. Nordenson, G.R. Bell, Seismic design requirements for regions of moderate
seismicity, Earthquake Spectra 16 (1) (2000) 205–225.
[30] K.H. Luo, Y.Y. Wang, Research on conversion relationships among the parameters of
ground motions in seismic design codes of China, America and Europe, Build. Struct.
36 (8) (2006) 103–107 (in Chinese).
[31] Eurocode 3, Design of steel structures - part 1.1: general rules and rules for build-
ings, EN 1993-1-1-2005, European Committee for Standardization, Brussels, 2005.
[32] ANSI/AISC 360-10, Specification for Structural Steel Buildings, AISC, Chicago, IL,
2010.