Professional Documents
Culture Documents
40 Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014
seismic design regulations are quite few.
Moreover, most of the codes are not regu-
larly updated. Some basic provisions of the
few codes available are obsolete and some
are unrealistic. Cases in point include the
Kenyan seismic code, which has never been
updated since its issuance in 1973, whereas
the knowledge of earthquake engineering
has shown significant strides ever since.
Also, most of the sub-Saharan African codes
employ obsolete site-effect provisions that
lead to unsafe design and that are no more in
use in most parts of the world.
Quality control in the construction indus-
try is in a dismal state. Building regulations,
if available, are not often properly adhered to,
even for static gravity loads. Reports of build-
ings collapsing during and after construction
have become common news in Africa. In
2011 alone, the collapse of five buildings in
different locations of Kenya was reported
(http://www.a4architect.com/). In Kampala,
the capital of Uganda, at least 60 people were
reported dead from the collapse of 11 build-
ings within the recent few years (http://www.
ugpulse.com/). In February this year, a 16-sto-
rey building under construction in Dar es
Salaam, Tanzania, collapsed costing close to
40 lives. Equally tragic incidents have become
a major concern in Nigeria (Oloyede et al
2010). Similar situations can be expected to
exist in many other African countries where
the building industry has been wrongly left as
an informal economic sector, with negligible
public control and enforcement of regulations. Figure 1 The seismic zoning map of Kenya in terms of MMI scale (MWK 1973)
Imagining how buildings of such quality
would behave during strong earthquakes is period of design earthquake. It focuses on The Kenyan Code – 1973
extremely scary, to say the least. the pertinent provisions of the selected rep- The Kenyan seismic code, probably one of
A number of shortcomings can be cited resentative seismic codes. The corresponding the pioneers on the continent, was issued in
of the rather few seismic codes available in provisions in these codes are compared 1973 by the Kenyan Ministry of Works and
some sub-Saharan African countries. Some with one another and with recent issues of uses the Modified Mercalli intensity (MMI)
of the fundamental pitfalls are related to American and European codes. scale to map the seismic hazard of the coun-
the definition of the design ground motion. try, as shown in Figure 1 (MWK 1973). The
Primarily, there exists a lack of uniformity in map divides the country into four seismic
the return period of the design earthquake BASIC GROUND MOTION zones: Zone V, VI, VII and VIII–IX, where
specified by the codes. For example, the PROVISIONS OF THE the Roman numbers are in accordance with
Ethiopian seismic code, EBCS 8, (MWUD AFRICAN SEISMIC CODES the MMI scale. This method of seismic haz-
1995) provides for a return period of 100 The selected codes include the Ethiopian, ard mapping using earthquake intensity is
years, whereas the South African code Kenyan and Ugandan codes to represent seldom in use nowadays for design purposes.
(SABS 2010) has adopted 475 years, while the most seismic region of the EARS, the The code neither states the recurrence
none of the other codes explicitly state the Ghanaian code to represent the less seismic period of the design earthquake nor its prob-
return period. An equally important issue western Africa, and the South African ability of exceedence. Explicit numeric values
is the fact that the approach employed to code to represent the southern region. The of parameters like PGA, which are more rele
account for the amplification potential of site provisions on the two basic issues are briefly vant to engineering design, are not provided.
soils in almost all these codes is obsolete, presented in the following section, to be However, these can be inferred from the
with the exception of the recently updated followed by comparisons among themselves relation provided by the code for estimating
South African code (SABS 2010), which has and with selected recent American and the seismic coefficient (see Equation (1)), and
adopted the recent approach introduced European codes. The National Earthquake the accompanying explanation that the PGA
in the European code to account for site Hazard Reduction Program (NEHRP) corresponding to Zones VIII–IX, VII and VI
amplification. is selected to represent the American are 0.05 g, 0.025 g and 0.0125 g respectively.
This paper presents a study of these two codes (BSSC 2004), whereas the Eurocode According to the map, populous Kenyan
fundamental issues related to input ground (Eurocode 8 2004) automatically qualifies as cities within the eastern branch of the EARS,
motion, namely site effect and recurrence representative of European codes. like Nakuru, belong to Zone VIII–IX; the
Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014 41
4 disregards the inherent wide variation in the
dynamic behaviour of natural soil deposits,
and is not in agreement with even the earli-
est site-dependent response spectra devised
3 in the 1970s, as will also be discussed later.
The seismic coefficient, Se, in Equation (1)
Se (normalised)
42 Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014
Seismic coefficient 3 3
2 2
Sd
1 1
0 0
0 1 2 3 0 1 2 3
T (s) T (s)
EBCS 8, Class A EBCS 8, Class B EBCS 8, Class C EBCS 8, Class A EBCS 8, Class B EBCS 8, Class C
(a) (b)
Figure 4 Site dependent design spectra of EBCS 8:1995 (a) for ESF analysis, and (b) for dynamic analysis (MWUD 1995; Worku 2001, 2011)
Subsoil Type II
0.06
Subsoil Type I
Cd
0.04
0.02
0
0 0.4 0.5 1.0 1.5 2.0 2.5 3.0
where I is the importance factor of the Subsoil Class C comprises loose cohesion- Standards in 2003, specifies the following
building that is assigned values of 0.8 to less soil deposits and soft to medium stiff design spectrum, Ca, for both the ESF and
1.4; W is the weight of the building; β is the cohesive soils with vs < 200 m/s; and vs is the modal analysis (UNBS 2003):
design response factor of the structure that average shear-wave velocity of the geological
accounts for site effect; and γ is the behav- formation of the site. Plots of Equation (3) are Ca(Ti) = C(Ti )ZIK(4)
iour factor which accounts for ductility, non- given in Figure 4(a).
linear response and the influence of damping EBCS 8 also provides the site-dependent where C(Ti ) is the ordinate of the “basic”
other than 5%. elastic design spectra of Figure 4(b) for response spectrum given for three different
The influence of the site soil is embed- dynamic analysis which are not identical to soil groups as shown in Figure 5; Z is the
ded in the design response factor (or elastic the spectra of Figure 4(a). Whereas the right- seismic zone factor given for the three seis-
design spectrum) given by hand segment varies according to 1/T2/3 mic zones of the country shown in Figure 6
in Figure 4(a), it varies in accordance with assuming the values of 1, 0.8 and 0.7 for
S ≤ 2.5 (3) 1/T in the spectra of Figure 4(b). Similarly
β = 1.2 Zones 1, 2 and 3 respectively; I is the struc-
T 2/3 the amplification factors in Figure 4(b) for tural importance factor that takes values of 1
S is the site coefficient, which can assume Classes B and C are 1.5 and 2.25 instead of up to 2; and K is the structural performance
the values of 1.0, 1.2 or 1.5 for Subsoil Class 1.2 and 1.5. The curves in Figure 4(b) are factor which is dependent on the type of the
A, B or C, respectively. Subsoil Class A almost identical to the ATC-3 Spectra, which structural system and materials. The recom-
includes rock or similarly competent forma- are based on the early findings of the empiri- mended minimum values of K vary from 1
tions and dense deposits of sand, gravel cal studies of Seed et al (1976) presented in for the most carefully designed ductile struc-
or over-consolidated clay characterised by Figure (11) and will be discussed later. tural systems to 4 for structural systems of
vs ≥ 800 m/s; Subsoil Class B consists of minimal ductility.
deep deposits of medium dense sand, gravel The Ugandan Code – US 319:2003 The design spectrum in Figure 5, which
or medium stiff clays with thicknesses of at The seismic code of Uganda, US 319:2003, is defined as the “basic” design spectrum by
least several tens of metres and vs ≥ 200 m/s; issued by the Ugandan National Bureau of the code, is the minimum design inelastic
Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014 43
3.0
2.5
2.0
1.5
Sd
1.0
0.5
0
0 1 2 3
T (s)
S1 S2 S3 S3 for PGA >0.3g
(a) (b)
Figure 7 (a) Seismic hazard map of Ghana, and (b) the seismic elastic design spectra according to the Ghanaian seismic code (BRRI 2010)
spectrum specified for the most seismic selected points, the amplification factors can three seismic zones of 1, 2 and 3. Each zone
zone, Zone 1, and for the most ductile be inferred to be 1.25 and 2.5 respectively. is assigned a constant peak rock-surface
structural system, for which K is assigned Even though these site-dependent design acceleration of 0.15 g, 0.25 g and 0.35 g
the value of unity. In contrast to this, the spectral curves look similar in shape to those respectively. This hazard map seems to have
basic design spectrum in most seismic design of ATC-3, they also exhibit some differences been influenced a lot by the deterministic
codes is conventionally defined as the elastic in the values of the amplification factors, study of Amponsah et al (2009) that is based
design spectrum before any factor is applied especially in the case of the softest soil class, on a single event and appears to have already
to it to account for inelastic response and for which the Ugandan code assigns a value accounted for site amplification. Also, the
ductility. This corresponds to the spectrum of 2.5 instead of 2.25. recurrence period of the design earthquake
obtained from Figure 5 by applying the max- is not stated, indicating the influence of the
imum value of the structural performance The Ghanaian Code – 1990 deterministic approach employed in the work
factor of K = 4 specified for the least ductile As in most sub-Saharan African countries, of Amponsah et al (2009).
system in the same seismic zone. recorded strong earthquake ground motions The rather large values of PGA specified
Thus, the maximum ordinate of the in Ghana are nonexistent. A few studies by the code are comparable to values speci-
elastic design spectrum corresponding to the conducted since recently indicate that the fied for the highly seismic western coastal
flat part of the spectral curves in Figure 5 Accra metropolitan area is one of the most region of the US (including California) pro-
obtained in this manner will have the value of seismic-prone areas. Earthquake magnitudes posed for a return period of 475 years. It is
0.32. This ordinate is as a rule 2.5 times the of up to 6.5 on the Richter scale have been considered unlikely that these values apply to
rock-level normalised PGA (Newmark & Hall recorded (Amponsah 2004; Amponsah et the comparatively less seismic region of West
1969). Therefore, the corresponding mini- al 2009; Allotey 2010; Oteng-Ababio 2012). Africa, unless corroborated by more detailed
mum rock-level PGA for Zone 1 can easily be Using a hybrid deterministic seismic hazard probabilistic studies that account for the
back-calculated through dividing 0.32 by 2.5 assessment technique, Amponsah et al range of earthquake magnitudes expected to
to obtain 0.13 g. The corresponding minimum (2009) simulated the 6.5 magnitude earth- occur in the region.
PGAs for Zone 2 and Zone 3 are obtained as quake to predict peak horizontal and vertical As shown in Figure 10(a) of the GSHAP
0.10 g and 0.09 g by applying the zone factors ground accelerations in the ranges of 0.14 g map for Africa, the maximum PGA in the
of 0.8 and 0.7 respectively, given by the code. to 0.57 g, and 0.2 g to 0.34 g respectively, entire continent for a return period of 475
Comparison of these values with those in with a maximum amplification factor of 4 for years is 0.24 g, and the maximum possible
the recent continental GSHAP map of Africa the horizontal motion in the period range of PGA specifically for Ghana is 0.16 g, which is
presented in Figure 10 suggest that these 0.2 s to 0.5 s. The highest amplifications are applicable to the zone containing Accra and
PGA values tally with a 475-year return estimated at sites of unconsolidated or poorly its environs. This is based on a probabilistic
period of the design earthquake, even though consolidated deposits. Such magnitudes of seismic hazard assessment study conducted
the code does not mention the length of the shaking are quite large in the context of by Grünthal et al (1999).
recurrence period. the known history of seismic activity in the The design spectra of the Ghanaian code
The code also fails to clearly specify the African continent. are given in Figure 7(b). The spectra for
values of amplification factors associated The latest Ghanaian code is the Code for PGAs less than 0.3 g and the corresponding
with Subsoil Types II and III of Figure 5. Nor the Seismic Design of Concrete Structures amplification factors are similar to those
does it indicate how the descending seg- issued in 2010 by the Ghanaian Building of the Ethiopian seismic code, EBCS 8,
ments on the right-hand side of the spectral and Road Research Institute (BRRI 2010). It given in Figure 4(b) or to the ATC 3, 1978
curves vary with period. Nonetheless, using provides the seismic hazard map shown in spectrum given in Figure 11(b), though the
ordinates directly read from the curves at Figure 7(a), which divides the country into right-hand segment in the Ghanaian code
44 Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014
Figure 10(a) Seismic hazard map of Africa
for a return period of 475 years
according to GSHAP (data and
plotting tool from http://gmo.gfz-
Potsdam.de/)
Figure 8 The seismic hazard map of South Africa according to SANS 10160-4:2010 (SABS 2010)
4.00
Ground Type 1
3.50
Normalised spectral acceleration
Ground Type 2
3.00
Ground Type 3
2.50
Ground Type 4
2.00
1.50
1.00
0.50
0
0 TB TC 1.0 1.5 TD 2.5 3.0 3.5 4.0
Period, T (s) Figure 10(b) Seismic hazard map of EARS
region for a return period of 475
Figure 9: Normalised elastic design spectra of SANS 10160-4:2010 (SABS 2010) years according to GSHAP (data
and plotting tool from http://
varies according to 1/T2/3. Apparently the The South African Code – gmo.gfz-Potsdam.de/)
code uses the same spectra for both the ESF SANS 10160-4:2010
and dynamic methods of analysis. The code The recent revision of the South African seismic activities. Zone I is assigned ag = 0.1 g,
reduces the ordinate of the constant part standard for seismic actions, SANS 10160- whereas Zone II can assume larger values.
from 2.5 to 2 for soil group S3 when the PGA 4:2010 (SABS 2010), was issued in June 2010 With regard to site effect, the code has
is larger than 0.3 g, as shown in Figure 7(b). and makes up one of the eight parts of the directly adopted Type 1 Spectrum of the
This measure seems to have been intro- South African National Standard SANS recent European code (Eurocode 8 2004)
duced in order to account for the increased 10160:2010 (Wium 2010), which replaces the together with the four Ground Types A to D,
damping in soils during high-intensity older version of SABS 0160:1989. omitting Ground Type E and softer sites. The
shaking that tends to reduce soil amplifica- Seismic hazard in SANS 10160-4:2010 is corresponding elastic design spectra normal-
tion, even though such a level of shaking presented in the form of the seismic hazard ised with respect to the PGA are provided
is a very remote possibility in Ghana, as map of Figure 8 in terms of the reference in Figure 9. The amplification factor S varies
explained above. peak ground acceleration ag for rock site and from 1 to 1.35 depending on the ground
The provisions for design ground motions a return period of 475 years. Notably, this type. The amplification factors in all cases
in the 2010 edition (BRRI 2010) described return period has been introduced in 1989, are constant over the entire period range and
above are practically identical to those in the when the earlier edition, SABS 0160, was are much smaller than those introduced in
1990 edition (BRRI 1990) and do not account issued. Two major zones are distinguishable current US codes. Similarly, the transition
for recent findings on site amplification in the map: Zone I of natural seismic activities periods, TB and TC, are dependent on the
effects, which are discussed in a later section. and Zone II of predominantly mining-induced ground type, whereas TD, the transition to
Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014 45
4 4
(maximum ground acceleration) Spectra for 5% damping
Spectral acceleration/PGA
3 Soft to medium clay and sand 3 Soil Type S2
Deep cohesionless soils Soil Type S1
Stiff site conditions
2 Rock 2
1 1
0 0
0 1 2 3 0 1 2 3
Period (s) Period (s)
(a) (b)
Figure 11 ( a) Site-dependent average spectra by Seed et al (1976); (b) Site-dependent design spectra modified from Seed et al (1976) and specified by
ATC-3 (Dobry et al 2000)
the long-period range, is constant, taken as introduced in US codes since more than period in this code is most likely 475 years,
2 seconds. three decades ago, and subsequently adopted though not explicitly stated in the code.
Given the relatively stable seismic nature as a standard design-level hazard worldwide. According to the GSHAP map, the west-
of South Africa compared to other seismic- The data base of GSHAP is accessible ern side of the sub-Saharan Africa is not as
prone regions, especially those in the EARS, to users (http://www.seismo.ethz.ch/). A seismic as the eastern side. The maximum
the stringent requirements of this code seismic hazard map of Africa prepared using PGA in the western region is 0.16 g, which
are quite exemplary for future revisions of an online tool (http://gmo.gfz-Potsdam. is assigned to limited areas in countries
seismic design codes of other sub-Saharan de/) is given in Figure 10(a). A similar map like Gabon, Congo, Cameroon and Ghana
African countries. of the EARS region – the most earthquake- (Figure 10(a)). Thus, the PGA of 0.15 g, 0.25 g
prone region of Africa – is also given in and 0.35 g specified in the Ghanaian local
Recent seismic hazard map Figure 10(b). Compared with these regional code for its three seismic zones are obviously
of sub-Saharan Africa maps, the current local seismic hazard maps too large for a return period of 475 years.
The Global Earthquake Model (GEM) initia- of some East African countries like Ethiopia This indicates that the present local seismic
tive is an on-going collaborative worldwide and Kenya, given in Figures 1 and 3, under- map of Ghana significantly overestimates
effort launched in 2009 with the aim of estimate the seismicity of their cities and the seismicity and would result in unneces-
building a heightened public understanding towns by an average of about a half. PGAs sarily conservative and expensive design.
and awareness of seismic risk, leading to of up to 0.24 g are assigned by GSHAP to Obviously, the seismic hazard map of Ghana
increased earthquake resilience worldwide some localities in East Africa, particularly needs reassessment, even though it was
through sharing of earthquake data, models to the Afar region of Ethiopia and to an area updated very recently, in 2010.
and knowledge; through application of around Arusha in Tanzania. A number of The return period in the South African
GEM tools and software to inform decision- populous towns and cities are also assigned code of SANS 10160-4:2010 is explicitly stat-
making for risk mitigation and management; PGAs up to 0.16 g (coloured yellow). ed as 475 years. In fact, the seismic hazard
and through expansion of the science and Kisumu, located on the northeastern map given in Figure 9 specifies PGA values a
understanding of earthquakes (http://www. shore of Lake Victoria and the third most little in excess of those recommended by the
globalquakemodel.org). However, tangible populous city in Kenya, has probably seen GSHAP map (on the average by 25%).
results from this programme, like updated one of the extreme discrepancies – from
seismic hazard maps, will only be available in 0.0125 g in the local Kenyan code to 0.12 g
a couple of years’ time. in the GSHAP map – with a nearly tenfold SITE EFFECT ON DESIGN SPECTRA
In contrast, the seismicity of the entire difference. This would imply that, whereas
globe was compiled in the 1990s for the first seismic loading is practically negligible for Site-dependent design spectra:
time as part of the Global Seismic Hazard the design of structures in Kisumu according early studies
Assessment Program (GSHAP), which was to the existing Kenyan code, it is an impor- Figure 11(a) shows some of the earliest site-
launched by the International Lithosphere tant consideration if the GSHAP map is to be dependent average spectra from the pioneer-
Program (ILP) and endorsed as a demonstra- followed. ing work of Seed et al (1976). Figure 11(b)
tion programme in the framework of the One of the reasons for such discrepancies gives the ATC-3 simplified spectra introduced
United Nations International Decade for is obviously the difference in the adopted in 1978 for routine design purposes based on
Natural Disaster Reduction (UN/IDNDR). return period of design earthquakes. Even the curves of Seed et al (1976), whereby the
The project was active from 1992 to 1999 though no return period is stated in the fourth site class is omitted, as the associated
(http://www.seismo.ethz.ch/). Kenyan code, it can be inferred to be not more data were not of satisfactory quality, as cau-
The GSHAP global seismic hazard map is than 100 years. In the Ethiopian code the tioned by the authors themselves.
given in terms of rock-surface PGA for a 10% return period is explicitly stated as 100 years. The spectral curves for Soil Types S2 and
exceedence in 50 years, which is equivalent In contrast, the local seismic hazard map of S3 of Figure 11(b) in the descending branch
to a return period of 475 years for the design Uganda in US 319:200 provides PGAs com- are obtained by simply raising the spectral
earthquake. This level of hazard has been parable with GSHAP hinting that the return curve of Soil Type S1 (rock) by a single factor
46 Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014
in each case. This factor, referred to as the 1.0
ratio of response spectra (RRS), takes the Damping = 5%
values of 1.5 and 2.2 for Soil Types S2 and S3
respectively. The descending segments cor- Soft soil – South of San Francisco
responding to the velocity-sensitive period
3.5 6
3.0 5
2.5
4
2.0
3
Fa
Fv
1.5
2
1.0
0.5 1
0 0
0 500 1 000 1 500 0 500 1 000 1 500
vs (m/s) vs (m/s)
A1 = 0.1 g 0.2 g 0.3 g 0.4 g A1 = 0.1 g 0.2 g 0.3 g 0.4 g
Figure 13 V
ariation of spectral amplification factors versus vs for short and long period ranges and for a range of intensity of rock shaking
(re-plotted after Borcherdt (1994))
Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014 47
proportional to the mean shear-wave veloc- Table 1 Site soil classes as per the recent NEHRP editions (BSSC 1995, 1998, 2004)
ity, vS, of the upper 30 m thickness raised
Pre-1994 New
to a certain negative exponent, which is SPT blow
site class NEHRP Description vS (m/s) Su (kPa)
count, N
dependent on the period band and the (approximate) site class
intensity of the rock acceleration (Borcherdt A Hard rock > 1 500 – –
1994; Rodriguez-Marek et al 1999; Borcherdt S1
B Rock 760 – 1 500 – –
& Fumal, 2000; Dobry et al 2000; Dobry
& Susumu 2000; Borcherdt 2002; Crouse C Soft rock / very dense soil 360 – 760 > 50 > 100
& McGuire 2002; Stewart et al 2003). It S1 and S2
D Stiff soil 180 – 360 15 – 50 50 – 100
was thus found important that site soils
are classified on the basis of this important E Soft soil < 180 < 15 < 50
parameter. S3 and S4
F Soils requiring site-specific study – – –
The empirical studies of Borcherdt
(1994; 2002) in particular showed that,
for the low-amplitude rock accelerations Table 2 Values of the site coefficient Fa and Fv according to NEHRP 1994 (BSSC 1995)
not exceeding 0.1 g recorded in California
Soil Fa for shaking intensity Aa Fv for shaking intensity Av
Bay Area during the 1989 Loma-Prieta profile
earthquake, the amplification factors for the type ≤ 0.1 0.2 0.3 0.4 ≥ 0.5 ≤ 0.1 0.2 0.3 0.4 ≥ 0.5
acceleration-sensitive and velocity-sensitive A 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8
period regions, denoted by Fa and Fv, are
B 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
approximately proportional to vs–0.4 and
vs–0.6 respectively. The statistically estab- C 1.2 1.2 1.1 1.0 1.0 1.7 1.6 1.5 1.4 1.3
lished values of Fa and Fv for low-intensity
D 1.6 1.4 1.2 1.1 1.0 2.4 2.0 1.8 1.6 1.5
ground motions were used to calibrate one-
E 2.5 1.7 1.2 0.9 b 3.5 3.2 2.8 2.4 b
dimensional analytical site response analy-
sis programmes, which in turn were used to F b
48 Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014
3.0 0.8
2.5
0.6
2.0
CSe
CSe/Ca
0.4
1.5
1.0 0.2
0.5
0
0 0 1 2 3
0 0.5 1.0 1.5 2.0 T (s)
T (s) Class A Class B Class C Class D Class E
(a) (b)
Figure 14 Elastic design spectra according to NEHRP 1994 (a) Basic (BSSC 2004), and (b) for Aa = Av = 0.1
3 4
3
2 Se/ag
Se/α
1
1
0 0
0 1 2 3 0 1 2 3 4
T (s) T (s)
Class A Class B Class C Class A Class B Class C Class D Class E
(a) (b)
Figure 15 Normalised elastic response spectra: (a) EC 8: 1994 (ENV 1998, 1994); (b) EC 8: 2004 – Type 1 (re-plotted from EN 1998-1 (2004))
1. The original three (later four) site categories 2002; BSSC 2004). For this reason, no major against period normalised with respect to Ca
are replaced by six new categories A to F. changes have been made so far to these val- is given in Figure 14(a) for Cv/Ca = 1.
2. The older qualitative site classification ues since their first introduction in 1994. Equation (7) is plotted in Figure 14(b) for
method is replaced by a new unambiguous the five soil classes A to E for a seismic zone
and more rational classification method characterised by Aa = Av = 0.1. As pointed
using a unique value of vs of the upper SITE-DEPENDENT DESIGN out earlier, the amplification occurs over the
30 m geological formation. Alternatively, SPECTRA IN RECENT US entire period range. Similar sets of spectral
average SPT blow counts and/or un- AND EUROPEAN CODES curves can be prepared for other seismic
drained shear strength can be used. zones. These spectral curves for various
3. Two seismicity-dependent site coef- Site effect provisions in the soil classes are entirely different from those
ficients, Fa and Fv, replace the single recent NEHRP series proposed by ATC-3, as given in Figure 11(b),
site coefficient, S, of the old system. The recent site amplification factors especially in the short-period region.
Both factors increase with decreasing described in the foregoing sections were first In more recent versions of NEHRP, the
shaking level due to the associated introduced in conjunction with the basic amplification factors remained basically
decreased damping. This leads to larger design spectrum of NEHRP 1994 given by the same, whereas some changes have been
seismic design forces for many classes of the following relationship (BSSC 2004); introduced related to the seismic hazard
structures in low-seismic regions, like in parameters, namely Aa and Av, which have
1.2Cv
Africa, founded on soft formations. CSe = ≤ 2.5Ca; Cv = FvAv; Ca = Fa Aa(7) meanwhile been replaced by spectral acceler-
T2/3
It is important to point out that results of ations at short period and at 1-second period,
more recent studies based on an enlarged The short-period amplification factor Fa is SS and S1 respectively. Subsequent changes
database including records from more applied on the constant part of the spectra, made to the basic shape of the design
recent earthquakes, like Northridge 1994, whereas the intermediate-period amplification spectrum shown in Figure 14(a) include the
have not indicated significant changes to factor Fv is applied on the descending seg- reintroduction of the linearly rising left part
the values of the above site amplification ment. In order to show the basic shape of the and the change made to the descending right
factors (Borcherdt & Fumal 2000; Borcherdt elastic design spectrum, a plot of Equation (7) side from T–2/3 to T–1.
Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014 49
Site effect provisions in 3.0
the recent Eurocode
The first edition of the European seismic 2.5
code (Eurocode 8 1994) employed only three
Normalised spectre
2.0
site classes, A, B and C, similar to those in
ATC-3, 1978, which were also adapted by 1.5
almost all African codes as presented above.
EC 8 1994 strangely specifies a smaller 1.0
maximum value for the softest site class C
than for the stiffer sites A and B, as shown 0.5
50 Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014
the European code, do not include this class 6
of rock.
5
African versus NEHRP spectra:
4
soil sites
The design spectra of the NEHRP 2003
Sa/g
3
(BSSC 2004) for soil classes C to E are plot-
ted in Figure 17 for a selected PGA of 0.05 g 2
together with the design spectra of the
African codes for soil sites. This particular 1
value of PGA is selected for consistency, with
0
the comparison made in the next section 0 0.5 1.0 1.5 2.0 2.5 3.0
to study the influence of return period. To T (s)
avoid further congestion, and because of EC 8 / SANS Class B EC 8 / SANS Class E Kenya: Soft ground
their similarities, the South African and the EC 8 / SANS Class C EBCS Class B / Ghana S2 Uganda: Type II
Eurocode spectra are plotted separately in EC 8 / SANS Class D EBCS Class C / Ghana S3 Uganda: Type III
the next section.
The plots in Figure 17 show that the spec- Figure 18 C
omparison of normalised design spectra of sub-Sahara African codes for soil sites with
tra of most sub-Saharan African codes can Eurocode 8 (2004) Type 1 spectra
be smaller by up to 60% of those of NEHRP.
The spectra of the Kenyan code are even It should be emphasised that, compared 0.16 g. Addis Ababa of Ethiopia, Nakuru of
much smaller, being only 25% of NEHRP to the Eurocode 8 amplification factors, the Kenya and Accra of Ghana belonging to this
spectra in an important range of periods amplification factors of NEHRP are support- zone are selected for comparison purposes
typical of commonly built types of buildings. ed by evidence from a significant number of the combined influence of site soil and
In general, the spectra of the sub-Saharan of independent research works on a larger return period.
African codes are deficient over a wide database enriched by ground motion records According to the GSHAP map of Figures
range of period up to at least 1 second. This from the 1989 Loma-Prieta and the 1994 11(a) and 11(b), a representative PGA of 0.1 g
range encompasses most common-purpose Northridge strong earthquakes (Borcherdt may be assumed for the design earthquake
structures built in African cities, including 1994; Rodriguez-Marek et al 1999; Borcherdt in all of these places. In contrast, according
individual residential houses, schools, apart- & Fumal 2000; Dobry et al 2000; Dobry & to the respective local codes, Addis Ababa
ments, office flats, public offices, hotels, Susumu 2000; Borcherdt 2002; Crouse & belongs to Zone 2 with a PGA of 0.05 g,
hospitals, etc, with heights reaching up to McGuire 2002; Stewart et al 2003). Nakuru belongs to Zone VIII–IX with a PGA
around 15 storeys. In the case of Soil Class of 0.05 g, and Accra belongs to Zone 3 with a
E, buildings of almost any height would be The combined influence of PGA of 0.35 g.
under-designed by the provisions of sub- return period and site soil The NEHRP spectra corresponding to a
Saharan African codes. Of the five seismic codes of sub-Saharan PGA of 0.1 g for soil sites are presented in
The spectral discrepancies in Figure 17 African nations considered in this study, Figure 19 in comparison with the spectra
clearly show the inadequacy of the old the South African and Ugandan codes for soil sites according to the African local
approach, on which most African codes are adopted a return period of 475 years for the codes. The comparisons are made for rock
based, to account for site effect. Comparisons design earthquake, though the latter does and soil sites separately.
for other seismic zones can be made analo- not state this explicitly. On the other hand, The plots in Figure 19(a) for rock sites
gously, but the trend remains similar. the Ethiopian code unambiguously states a show that the design spectra of the East
return period of 100 years only. The return African codes of Ethiopia and Kenya can
African versus Eurocode 8 spectra: period employed by the Kenyan code is not result in less than 50% of the spectra of
soil sites clearly stated, but one can infer from its NEHRP adapted to the GSHAP seismic
Comparison of the design spectra in the seismic map and the associated zone factors hazard map. Since the site effect is zero for
sub-Saharan African codes with the spectra that its return period cannot be more than rock sites, the differences in these spectral
of the European code is more direct for- 100 years. Similarly, the Ghanaian code curves reflect the influence of the return
ward, as all of them use PGA to characterise does not explicitly state its return period. period of the design earthquake only. On
seismicity. This fact enables the comparison However, when compared to the GSHAP the other hand, the spectra of the Ghanaian
of normalised spectra regardless of the seis- map of Figure 10(a), the PGAs reaching up code can be close to threefold of the NEHRP
micity of regions. Comparison with Type to 0.35 g in this local code suggest that the spectra and five to tenfold of the Ethiopian
1 spectra of Eurocode 8 (2004) are given in corresponding return period should be much and Kenyan spectra respectively.
Figure 18. larger than 475 years. Similarly, the plots in Figure 19(b) for soil
The plots show that the spectra of most In order to study the combined effect of sites exhibit a more pronounced difference
sub-Saharan African codes can be smaller by return period and site effect, we shall focus between the NEHRP spectra and those of
up to 30% of Type 1 spectra of Eurocode 8, on the Ethiopian, Kenyan and Ghanaian the East African countries of Ethiopia and
whereas in the Kenyan case, this can reach codes, which adopt a return period different Kenya. The NEHRP spectra could be as
58%. These discrepancies are still substantial, from 475 years. For this purpose, a site is large as fivefold of the East African spectra.
but comparatively smaller than the discrep- selected in each of these countries within This suggests that the design ground motion
ancies observed with respect to the NEHRP the same zone of the GSHAP map of Figures provisions of the East African codes are
spectra, whose amplification factors are 10(a) and 10(b). This zone is selected as the highly inadequate to ensure safety. The
consistently larger than those of Eurocode 8. one shaded in yellow with PGAs of 0.08 to Ghanaian spectra are still larger than the
Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014 51
0.8 result in design seismic forces that are 30%
to 75% smaller than would be obtained in
accordance with the provisions of American
0.6
and European codes. This is without consid-
Sa/g
0.4
incorporated the state of the art of both site
effect and return period. However, it is not
clear why the European site-dependent spec-
0.2
tra are favoured over the NEHRP spectra,
while the NEHRP spectra and the corre-
0
0 0.5 1.0 1.5 2.0 2.5 3.0 sponding site classification and amplification
T (s) factors are supported by a larger number
NEHRP Class C EBCS 8 Class B / Ghana S2 Ghana S2 of independent studies on a large database,
NEHRP Class D EBCS 8 Class C / Ghana S3 Ghana S3 compared to the European spectra.
NEHRP Class E Kenya: Soft ground Apparently both the 1990 and 2010 edi-
(b) tions of the Ghanaian code equally suffer
from flaws in the background seismic hazard
Figure 19 C
omparison of NEHRP design spectra adapted to GSHAP zoning of the region with the assessment study that led to the unrealistic
spectra of local African codes for (a) rock sites, and (b) soil sites seismic zone factors equivalent to PGA val-
ues ranging from 0.15 g to 0.35 g. In addition,
NEHRP spectra. This is despite the small and the current European code. With the even the 2010 edition failed to incorporate
site amplification factors of a maximum of exception of the Ethiopian and the South the current state of knowledge of site effect.
1.5 employed by this code, compared to up African codes, the return periods are not Based on the observations made above
to 3.5 proposed by NEHRP. The results are clearly stated in the codes. The Ethiopian from the perspective of design ground
indicative of flaws likely to have occurred code clearly states 100 years as its return motion alone, it may be concluded that most
in the background work of seismic hazard period for the design earthquake, whereas of the sub-Saharan seismic codes need care-
assessment that led to the local seismic haz- the South African code explicitly states 475 ful revision, and in some cases a thorough
ard map of Ghana. years. Through inference, the Kenyan and overhaul, to provide adequate safety to
According to the plots in Figure (19), Ugandan codes seem to employ 100 years human life and to the badly needed infra-
nearly all ranges of buildings (T of up to 3 s) and 475 years, respectively. The Ghanaian structure and building structures. It is hoped
built on any soil formation would be severely code seems to have a major flaw in its seis- that the GEM initiative will come up soon
under-designed by the inadequate provisions mic hazard map, because its zone factors are with important tools to rectify most of the
of the East African codes of Ethiopia and too large to suit the seismicity of the region. identified shortcomings observed in the local
Kenya. On the extreme opposite end, the In terms of site effect provisions, all, codes and to provide a common platform
Ghanaian code demands unjustifiably too except for the recent South African code, use that can lead to a continental harmonisation
large design forces for structures of all types site-dependent design spectra that are based of approaches.
on all categories of soil sites. on the outdated average spectra proposed
by Seed et al (1976). The South African code
directly adopted the design spectra specified REFERENCES
CONCLUSION by the current European code by simply omit- Allotev, N K, Arku, G & Amponsah, P E 2010.
The status of the design ground motion ting the softest site group. It is interesting to Earthquake-disaster preparedness: The case of
provisions of representative sub-Saharan note that the recently revised Ghanaian and Accra. International Journal of Disaster Resilience in
African codes has been assessed in terms Ugandan codes have not incorporated the the Built Environment, 1:140–156.
of two basic issues of site effect and return state of the art in site amplification potential. Amponsah, P E 2004. Seismic activity in Ghana:
period. The provisions are compared among For this reason, most of the sub-Saharan Past, present and future. Annals of Geophysics,
the five selected codes and against NEHRP African site-dependent design spectra can 47(2/3):539–543.
52 Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014
Amponsah, P E, Banoeng-Yakubo, B K, Panza, G F Eurocode 8: 1994. Design Provisions for Earthquake risks in Accra. Journal of Housing and the Built
& Vaccari, F 2009. Deterministic seismic ground Resistance for Structures. Brussels: European Environment, 27:187–205.
motion modeling of the greater Accra metropolitan Committee for Standardisation. Rey, J, Faccioli, E & Bommer, J 2002. Derivation of
area, southeastern Ghana. South African Journal of Eurocode 8: 2004. Design of Structures for Earthquake design soil coefficients (S) and response spectral
Geology, 112:317–328. Resistance. Brussels: European Committee for shapes for Eurocode 8 using the European Strong-
Borcherdt, R 1994. Estimates of site-dependent Standardisation. Motion Database. Journal of Seismology, 6:547–555.
spectra for design – Methodology and justification. Girdler, R W & McConnell, D A 1994. The 1990 to Rodriguez-Marek, A, Bray, J & Abrahamson, N 1999.
Earthquake Spectra, 10(4):617–653. 1991 Sudan earthquake sequence and the extent Characterization of site response general categories.
Borcherdt, R & Fumal, T 2000. Empirical evidence of the East African Rift System. Science Magazine, Berkley, CA: University of California, Pacific
from the Northridge earthquake for site-specific 264(5155):67–70. Earthquake Engineering Research Center, PEER
amplification factors used in US building codes. Ghosh, S 2001. Trends in the seismic design provisions Report 1999/2003.
Proceedings, 12th World Conference on Earthquake of US building codes. PCI Journal, Sept–Oct:98–102. SABS (South African Bureau of Standards) 2010. SANS
Engineering, Auckland, New Zealand, pp 1–6. Ghosh, S 2004. Update on the NEHRP provisions: The 10160-1: Basis of Structural Design and Actions for
Borcherdt, R 2002. Empirical evidence for site resource document for seismic design. PCI Journal, Buildings and Industrial Structures. Pretoria: SABS.
coefficients in building code provisions. Earthquake May–June:96–102. Seed, H B, Ugas, C & Lysmer, J 1976. Site-dependent
Spectra, 18(2):189–217. Grünthal, G, Bosse, C, Sellami, S, Mayer-Rosa, D spectra for earthquake-resistant design. Bulletin of
BRRI (Building and Road Research Institute) 1990. & Giardini, D 1999. Compilation of the GSHAP the Seismological Society of America, 66(1):221–244.
Code for the Seismic Design of Concrete Structures. regional seismic hazard for Europe, Africa and the Sousa, P M, Barros, R C & Bommer J J 2008.
Kumasi, Ghana: Council for Scientific and Industrial Middle East. Annals of Geophysics, 42:1215–1223. Assessment of seismic hazard and structural safety
Research (CSIR-INSTI). http://www.a4architect.com/2011/09/18/ of buildings in Beira and Chimoio. Proceedings, 14th
BRRI (Building and Road Research Institute) 2010. why-buildings-collapse-in-kenya/ World Conference on Earthquake Engineering, 12–17
The Seismic Design of Concrete Structures. Kumasi, http://www.globalquakemodel.org/ October, Beijing, China.
Ghana: Council for Scientific and Industrial http://gmo.gfz-Potsdam.de/pub/GSHAP_Map_Online/ Stewart, J, Liu, A & Choi, Y 2003. Amplification factors
Research (CSIR-INSTI). gshap_map_online_frame.html for spectral acceleration in tectonically active
BSSC (Building Seismic Safety Council) 1995. National http://www.iaee.or.jp/ regions. Bulletin of the Seismological Society of
Earthquake Hazard Reduction Program (NEHRP): http://www.seismo.ethz.ch/static/gshap/Gshap98-stc. America, 93(1):332–352.
Recommended Provisions for Seismic Regulations for html UNBS (Uganda National Bureau of Standards) 2003.
New Buildings. Washington DC: BSSC, FEMA 222A http://www.ugpulse.com/uganda-news/business/ US 319:2003: Seismic Code of Practice for Structural
and 223A (Provisions and Commentary). engineers-concerned-about-collapsing- Designs. Kampala, Uganda: UNBS.
BSSC (Building Seismic Safety Council) 1998. National buildings/9053.aspx Wium, J 2010. Background to SANS 10160 (2009).
Earthquake Hazard Reduction Program (NEHRP): http://www.usgs.org/ Part 4: Seismic loading. Journal of the South African
Recommended Provisions for Seismic Regulations for MWK (Ministry of Works of Kenya) 1973. Code of Institution of Civil Engineering, 52(1):20–27.
New Buildings and Other Structures. Washington Practice – Earthquake Design for Kenya. Nairobi: Worku, A 2001. Comparison of seismic provisions of
DC: BSSC, FEMA 302 and 303 (Provisions and MWK. EBCS 8 and current major building codes pertinent
Commentary). MWUD (Ministry of Works and Urban Development to the equivalent static force analysis. Zede, Journal
BSSC (Building Seismic Safety Council) 2004. National of Ethiopia) 1995. EBCS 8: Design of Structures of the Ethiopian Engineers and Architects, 18:11–25.
Earthquake Hazard Reduction Program (NEHRP): for Earthquake Resistance. Addis Ababa: MWUD, Worku, A 2011. Recent developments in the definition
Recommended Provisions for Seismic Regulations for Ethiopian Building Code Standard. of design earthquake ground motions calling for
New Buildings and Other Structures, Washington Newmark, N M & Hall, W J 1969. Seismic design a revision of the current Ethiopian Seismic Code
DC: BSSC, FEMA 450. criteria for nuclear reactor facilities. Proceedings, – EBCS 8: 1995. Zede, Journal of the Ethiopian
Crouse, C & McGuire, J 2002. Site-response studies 4WCEE, Santiago, Chile, Vol. II, B5, pp 1–12. Engineers and Architects, 28:1–15.
for the purpose of revising the NEHRP seismic Newmark, N M & Hall, W J 1982. Earthquake
provisions. Earthquake Spectra, 12(2):407–439. spectra and design. Engineering Monographs on
Dobry, R, Borcherdt, R, Crouse, B, Idriss, I, Joyner et Earthquake Criteria, Structural Design and Strong
al 2000. New site coefficients and site classification Motion Records, Vol 3, Oakland, CA: Earthquake
system used in recent building seismic code Engineering Research Institute (EERI).
provisions. Earthquake Spectra, 16(1):41–67. Oloyede, S A, Omoogun, C B & Akinjare, O A 2010.
Dobry, R & Susumu, I 2000. Recent development in Tackling causes of frequent building collapse
the understanding of earthquake site response in Nigeria. Journal of Sustainable Development,
and associated seismic code implementation. 3(3):127–132.
Proceedings, GeoEng2000, International Conference Oteng-Ababio, M. 2012. Neglected vulnerabilities in a
on Geotechnical & Geological Engineering, rapidly urbanizing city: Reflections on earthquake
Melbourne, Australia, pp 186–129.
Journal of the South African Institution of Civil Engineering • Volume 56 Number 1 April 2014 53