PROPOSED CONSIDERA TIONSFOR REVISION OF EBCS-8: 1995 FORCONSERVATIVE SEISMIC ZONING AND STRINGENT REQUIREMENTS FOR

TORSIONALLY IRREGULAR BUILDINGS

Samuel KindeHeQri Samueli School of EngineeringUniversity of California - Irvine, USA

ABSTRACT

There is a significant variation in seismic zoningadopted in Ethiopia for urban areas between theseismic building code [1]' and those reported inother research publications. F.urther, it is alsobelieved that the torsional loads recommended bythe current seismic building code to account foraccidental torsion are not conservative,.enough fora seismically active country such as Ethiopia whereth,re is inadequate quality control in constructionpractices. This paper deals with proposals forimplementing a conservative seismic zoning forurban areas such as Addis Ababa andimplementing a more stringent torsional loadrequirements in the current Ethiopian seismicbuilding code [2J.

INTRODUCTION

The seismic zone of an area detennines the level ofequivalent static forces that will be applied in abuilding frame in seismic design using the ESL(Equivalent Static Load) procedure. Likewise, theseismic zone also detennines the level of dynamicloads to be considered for design based on responsespectra analysis. However, among the publishedseismic zone maps for Ethiopia, there are distinctand significant differences in the predicted peakground accelerations (pGA) for various seismically .important regions of the country. It is thereforeargued that, the variations in predicted PGA aresignificant enough to warrant a review of theseismic zoning of urban areas such as AddisAbaba.

On the other hand, to account for the presence oftorsional motions in torsionally irregular buildingsan~ also to account for torsional ground motion andthe shift of the centers of mass due to a variety ofreasons such as poor construction quality, codesoften stipulate provisions for additional torsionalmomehts. These code provisions for accidentaltorsion detennine the location of where the lateralseismic loads are applied in a building and hence

the magnitude of the torsional moments thebuilding is designed for. The torsional moments ateach floor are typically calculated as the floor shearforce times the most unfavorable eccentricity.Adopting a conservative value for theseeccentricity values is, therefore, crucial f,?rparticularly in places such as Ethiopia wherequality control in the construction industry is not atadequate level yet.

SEISMIC ZONING FOR ETHIOPIA

Gouin [3] who used probabilistic approach iscredited for the initial attempts in producing thefirst seismic hazard map of Ethiopia as shown inFig. 1. Gouin's work served as a basis for theseismic zoning adopted by the EBCS-l 1983 [2]building code of Ethiopia (see Fig. 2). Since theproduction of Gouin's map, quite a large number ofdestructive earthquakes have occurred in thecountry causing damages both to property andhuman life. Further, destructive earthquakes thatoccurred in the neighboring countries were notincluded in the production of the first map in 1976.Subsequently, Kebede [4,5], panza et al [6]produced a new seismic hazard map of Ethiopiaand its northern neighboring countries to account

.for these additional earthquake records. Unlikeprevious works, the seismic zoning of Ethiopia andthe Horn of Africa reported by Kebede [4], Kebedeand Asfaw [7] also account for ground motionattenuation in addition to newer data obtained fromsuch sources as the US National EarthquakeInformation Service (NEIS). The works of Kebede[4,5], Kebede and Asfaw [7] along with the worksof L. Asfaw [8] served as a basis for the seismiczoning adopted by the current Ethiopian buildingcode [1] as shown in Fig. 3..Further, there havebeen other attempts on seismic zoning of some ofthe country's important economic regions such asthe city of Addis Ababa. The work of the RADIUSproject [9] is a notable example..

Journal of EEA, VoL 20, 2003

16 Samuel Kinde

Variations in Seismic Zoning

A close look at these various seismic zonings forthe country reveals significant differences in thepeak ground accelerations (pGA) assigned to suchimportant locations such as the city of AddisAbaba, tlle Afar Triangle and the main Ethiopianrift (MER) (fable 1). For example, for tlle MERarea, tlle peak ground accelerations seem to·compare well witll tllOse predicted by Gouin [3]using a probabilistic approach whereas for tlle AfarTriangle area, Gouin's hazard map gives highervalues [6]. Further, tlle seismic zoning adoptedcurrentIy by EBCS-8: 1995 departs significantlyfrom tllOse reported elsewhere such as results fromtile RADIUS project [9]. For example, tlle city ofAddis Ababa is classified as Zone 2 by EBCS­8:1995 (Fig. 3) witll a peak ground acceleration of0.07g whereas tlle report from tlle RADIUS projectdivides Addis Ababa into various microzones with

peak ground accelerations varying from O.13g ­0.50g as shown in Fig. 2. Interestingly, the widelyadopted building international codes such as UBC(UBC 1997) and mc (mC 20(00) classify majorcities in the Horn of Africa particularly AddisAbaba, Etlliopia as Zone 3 which corresponds to apeak ground acceleration ofO.3g as opposed to the0.07g assigned by EBCS-8: 1995 ..

A number of reasons could be sited for tllese

differences among the various 'reported seismiczonings for the country.

1. The first reason is the consideration of local

site conditions. For instance, tlle seismic

zoning for the whole country as reported byGouin does not take into account local site

conditions and ground motion attenuations. Aprimary reason may be the lack of such datadue to its prollibitive expense. The effect ofsoil conditions on PGA is highlighted byAsfaw who reported that site amplificationshave resulted in significant recorded higherintensities of earthquakes in the Northwestcomer and central part of Addis Ababa (nearFilowha area) as compared to other areas ofthe city [8]. He attributes topographicprominences and alluvium deposits as thefactors contributing to site amplifications.Looking closely, however, there is no adequateinformation on how Kebede [4] hasdetermined local site conditions. The same is

true for the study by the RADIUS project thatreported a seismic zoning for the city of Addis

Journal of EEA, Vol 20, 2003

Ababa (Fig. 4) which does not provideverifiable evidence on how the site conditions

were determined. In fact, the seismic zone mapof Addis Ababa according to the RADIUSproject suggests that the Northwest region ofthe city of Addis ~baba has one of the lowestPGAs. This is, however, in din~ct conflict withwhat is reported by Asfaw [iO] where theNorthwest part of the city is credited as havingsome of tile strongest earthquake intensitiesrecorded due to its topographic prominence.Further, tlle PGA reported by tile RADIUSproject seem to be much higher than any of tilereports in tlle literature suggesting that it is toodifficult to argue tllat tlle site conditions alonewarranty the use of more stringentclassifications to the city of Addis Ababa withmaximum PGA of 0.5g for southeast AddisAbaba. The issue of local soil and otller site

conditions is, therefore, an area that requiresfurther investigations.

H. The second source of variations in seismic

zoning among the various reports is theselected return period of the earthquakeconsidered for design. The return period usedboth in tile current Etlliopian building codeEBCS-8: 1995 and the previous code, ESCP­1: 1983 is 100 years whereas the twointernational building codes UBC 1997 and.mc 2003 use a return period of 475 years.Mekonnen [11] points that the PGA values forth¢ different regions of tile country are almosttwice when a return period of 475 years is usedas opposed to a return period of 100 years.With this argument, the PGA given by UBCand mc for the city of Addis Ababa could beinterpreted to be 0.15g for a return period of100 years. Further, the return periodconsidered for the 6.5M 'scenario earthquake'used in the RADIUS project is not specifiedeven though it is feasible that it may haveconsidered a higher return period as large as200-475 years.

Ill. The tllird source of variation is the influence of

additional data obtained over the years fromdifferent sources including NEIS and ~consideration of other major earthquakes in the~region including the 1991 earthquake in Juba,southern Sudan. Reports from Kebede [4] havebenefited from these additional data as

compared to the original seismic zoningreported by Gouin [3].

Conservative Seismic Zoningfor Torsionally I"egular Buildings 17

Proposed Interim Solutions

While some of the differences in the reportedseismic zonings of the ,country could be accountedfor as explained in the previous section, in balance,however, even with the consideration of the aboverationale, there remain noticeable differencesamong the seismic zonings. For instance, theconsideration of a comparable return period for thedesign PGA for the Addis Ababa region for UBC1997 and the EBCS-8: 1995 still leaves adifference in PGA of 0.15g against 0.07g which isa factoI'.of two. On the other hand, Kebede [4]indicates that the sejsmic zoning adopted by himand other researchers - which are supposed to bemore complete than previous national seismiczoning attempts - are still preliminary and athorough investigation is necessary to have acomplete and reliable seismic zoning for thecountry. Again, with regard to the seismicity of theAddis Ababa region, the report by Asfaw [10]strongly suggests that the majority of the strongestground shakings recorded in the city were causedby earthquakes that occurred 200 kilometersnortheast of the city (i.e., in Zone 3 and possiblyZone 4 areas). This certainly brings questionsregarding the validity of assuming a PGA of 0.07gof Zone 2 for this region which seems· to beaffected quite significantly by the seismici~r ofZone 3 and 4 areas within the western escarpmentsof the Afar Triangle and MER.

Therefore, in the face of these ambiguities anduncertainties reported above, further investigationand collaboration between seismologists andearthquake engineers is required before thecontroversy could be settled. For current designpractice, until such information is widely available,

how~ver, it still remains difficult to the support thecurrent ESBC8-95 classifications that put themaximum peak ground acceleration in AddisAbaba as 0.07g. In the same token, it also lookslike the peak ground accelerations predicted by theRADIUS project are on the high-end.and are notsupported by strong data. As a temporary solution,until further data is available, it is proposed that amore conservative zoning than the EBCS-8:1995be adopted that increases the PGA to thosecomparable to at least Zone 3 (pGA of O.lg ­0.12g).

SEISMIC TORSIONAL PROVISION ANDACCIDENTAL TORSION

Seismic torsional provisions are mainly intended toaccount for the differences between actual andcomputed eccentricities of building structures,dynamic effects of torsional response, and thespatial variation of the ground motion. Theseprovisions are applicable for both symmetrical andtorsionally irregular buildings. Historically, the useof 2-dimensional frame analysis - prior to theadvent and wide-spread use of computer-basedtools - had necessitated the consideration ofadditional story shears for torsionally irregularbuildings due to torsional moments. Since these 2­dimensional analyses did not explicitly consideredthe effect of the shift of the center of rigidity of abuilding from the centers of mass, it was, therefore,a rational approach to account for additional storyshears. Subsequently, almost all building codes,like UBC-78, UBC-88, ESCP-l:1983 (the oldEthiopian building code), and ATC hadtraditionally defined the seismic torsionalprovisions to account for not only the shift in the

Table 1: Predicted PGA in Addis Ababa and Afar Depression as per different sources

Kebede and AsfawEBCS-8: 1995RAPIUS (1999)'UBC-97(1996)

(1995) (1997)Addis

PGAfor0.08g [Eq 3]0.07gNANAAbaba

O.OlpPGAfor

0.12g [Eq 3]NA0.13g-0.5g0.3g0.0050

(extrapolated)AflV'

PGAfor0.14g [Eq 3]0.16g0.25g - 0.75gNADepression

O.Olp0.53g [Eq 16] (extrapolated)

PGAfor

0.19g [Eq 3]NA 0.3g0.005p

1.05g [Eq 16]

Journal of EEA, VoL 20, 2003

18 Samuel Kinde

ei = 0.05 LiAx (la)

and Li is the depth (d) or width (b) of the buildingdepending on the direction of load Considered.

where,

Ax = ( °max )2 ~ 3.0 for 81ft<U > 8 ••••• (1b.)1.20••••

2

eoo.x

ex = 1.7eoo.x - -b - ±O.lb (New Zealand Code, 1992)2 (2c)

eoo,Ye = 1.7eOOY ---±O.ld, . d

(Eqs 4.1.9.2-28ofNBCC-95)(2b)

ex = 1.5eca.x ± O.1b .

ey = 1.5eca,r ± O.ld

In comparison to other international codes forcountries with similar level' of constructionpractice, it may be argued that the EBCS~8:1995 isnot conservative enough if Ax is less than 2 forseismic torsional provisions. Despite the fact thatthe code is built OIP a rational approach, adopting a5% ratio much like UBC 1997, however, may notbe adequately conservative. It can further be arguedthat the adoption of 5% eccentricity by the UBC1997 code that is commonly used in the US may beacceptable to US practices given the availability ofadequate level of construction quatity control.Therefore, a recommendation is proposed 'here ~the accidental eccentricity provisions in theEthiopian building code be revised to adopt a larger

The general form of the design'eccentricity adoptedby most building codes can be expressed as:

ex=aecG.x±pb (13)

The same expressi.on as used by some of theinternational codes is given below,

Most buildings codes with similar seismic risklevels such as Ethiopia like Mexico [13,14] andNew Zealand [15] still have maintained the old

,provisions where the actual eccentricity is taken asthe sum of a given percentage of the building depthor width and the distance between the center of

rigidity and center of mass of the buildings (seeTable 2). The same procedure is used in, theCanadian code NBCC-95 [16] which takes the totalaccidental eccentricity as the sum of 10% of thebuilding weight or depth and 1.5 times the distanceof the center of rigidity from the center of mass at agiven floor level.

Accidental Torsion in Equivalent Static LoadProcedure

The accidental torsion for ESL (equivalent staticload) procedure as per EBCS-8:1995 is oftheform:

centers of mass and the spatial variation of theground-motion (i:e., accidental torsion), but alsosome magnification of the actual eccentricitybetween the center of rigidity and the center ofmass (Le., dynamic eccentricity). However with thewidespread availability of three-dimensionalanalysis tools that explicitly distribute the storyshears to each of the frames according to theirrelative stiffness, the scope of seismic torsionalprovisions - particularly in the US - have then bemodified to account only (or what is calledaccidental torsion, Le., torsional moments due tothe shift in the centers of mass and the spatialvariation of the ground motion. This explains whythe modern revisions of the Uniform Building Code(1994, 1997) define accidental torsion as apercentage of only the linear geometry (depth orwidth depending on the geometry). With regard tothe Ethiopian building code, EBCS-8:1995, Section2.3.2.1 of the code says:

"In addition to the actual eccentricity, in order tocover uncertainties in the location of masses and inthe spatial variation of the seismic motion, thecalculated center of mass at each floor 'i' shall beconsidered displaced from its nominal location ineach direction by an additional eccentricity."

It can be argued that the wording of the code withregard to the actual eccentricity considered is vagueand subjected to misinterpretation. For analysis ofbuildings based on three-dimensional models, onecan argue that the actual eccentricity between thecenters of rigidity and mass has been explicitlyaccounted for by the model and hence the onlyeccentricity to be considered is the shift of thecenter of mass. This is what is commonly practicedin tile interpretation of codes like UBC 1997 and itseems that EBCS-8:1995 had followed suit andadopted seismic torsional provisions for accidentaltorsion without calculated eccentricity. However, itcan also be argued that the code expects the storyforces to be applied at a location which is the sumof the actual dynamic eccentricity and theaccidental eccentricity. In this interpretation, thefactor used in multiplying the dynamic eccentricity(Le., a.) is then unity as opposed to 1.5 which wasused in the previous code, ESCP-l:1983 as shownin Eq. (2a).

JournalofEEA, VoL 20,2003

ConservativE! Seismic Zoningfor Torsionally Irregular Buildings 19

(i.e., 10%) percentage of the width or depth toreflect the Ethiopian construction industrypractices.

Accidental Torsion in Response SpectraAnalysis

The EBCS-8:1995, much like most building codes,specifies accidental torsion for response spectra­based dynamic analysis. The accidental torsion fordynamic cases is, however, limited to generating anadditional static equivalent torsional moment. Forengineers, this additional step of combiningdynamic story shears (or member forces) withadditional static story ,forces (or member forces)may be unnecessary additional step, particularlywhen modem structural engineering software cansimply consider a given shift in the centers of massand extract periods and mode-shapes in preparationfor response spectra analysis that accounts foraccidental torsion. Therefore, it is proposed that thecode also specifically address accidental torsion indynamic analysis where at least the automaticdynamic analysis that accounts for accidentaltorsion is given as an alternative. It is believed thatthis option will encourage. design engineers toaccount for accidental torsion in dynamic analysisthrough a direct and rationale approach.

Table 2 summarizes the common practices adoptedby some of the international building codes and theEthiopian building code along with the proposedremedial approach.

CONCLUSIONS

The seismic zoning of Ethiopia as reported bydifferent researchers over the past 30 years or so ismarked by distinct and signjficant variations inPGA for the most economically important parts ofthe country such as the Addis Ababa region and theAfar Triangle. This study investigates thesevariations and suggests that newer data, adifference in the earthquake return period and theconsideration of local site conditions could accountfor some of the differences. However, it issuggested that more detailed and well-documentedresearch in effect of local site conditions is requiredbefore a significant amount of confidence could bebuilt in the seismic zoning adopted by the currentseismic building code. As an interim solution,increased PGA is suggested for the Addis Ababaregion.

The study also investigates seismic coderequirements on accidental torsion for dynamicanalysis practiced in the country at the presenttime. Based on the argument that the qualitycontrol in construction industry is on a developingstage, the study suggests the implementation of amore stringent torsional requirements in the currentEthiopian seismic building code in par with othercountries of comparable construction practice andsimilar earthquake hazard levels.

Table 2: Comparison of accidental torsiop provisions by different codes

UBC-1988UBC-1997ESCP-I: 1983EBCS:8-1995Proposed Change

for EBCS:8-1995Statice,,=1.5ecG,X±0.05be,,=±0.05be,,=1.5ecG,x±0.1be,,=±0.05be,,=±O.lb

analysisey=l.5eca,y±o.05dey=±0.05dey=1.5ecG,y±o:Idey=±0.05dey=±O.ld

DynamicMi =eiFimove CMs by ±e",Mi =eiFiMi =eiFimove CMsby

analysis

x x x±ey and do RSA or

x x xx x x±e,,~±ey and do

Mi =eiFi Mi=eiFiMi =eiFiMi iFiRSAory y y

y y yy = ey yx x x Mi =eiFiMi iFi Y = ey. y

x x x

Mi =ei'Fiy y y

CM = center of mass. 'i' - story number. RSA - Response spectra analysis. Fx and Fy are story shears.Mx and My are the torsional moments.

Journal of EEA, VoL 20, 2003

20 Samuel Kinde

lai .

-V'qII}·· .J1 ~; ;;.~ .

-·_JrDNtA.~..

..4SA',•• y~ ••.• ~~-~~~--:. ::

~,~;;;~I.•. ~:,,~~.

.-----.--.-.-.-,----.------.-.---.-- .. .. ...ScJJ.·.':in '·XUlJ9'.

Figure 1 Seismic zoning of Ethiopia as per Gouin (1976)

E.TfU:Q}"A$EISUIC 'TtISK MAPti)fl.r~!4 ~$ ·'!W'TENSITY

'''''H$OMAL IA~~"'~

sUDAN

Figure 2 Seismic zoning of Ethiopia as per ESCP-l:1983

Journal of EEA, Vol 20, 2003

ColIMwtive Seismic Zoning/or ToniolUl/ly In'egular Buildings

Ii

Figure 3 Seismic Zoning of Ethiopia as per EBCS-8:1995

21

~;;""?:'::"i 0 13g- 0 15g~;:~!..:...!:-: ..

I:··:,::~..,::I 0.16g-0.18g

Ia 0.19g-021g

.~ ... ,-•.'. 0.22g-0.24gy •• ~ r••• ~ ••. l- .;

_ 0.25g - 0.29g

~....__m"'~l 0 <¥\n - 034g••••• hn_ ••• ~ •

•• ,~O;35g but SO.50g

Figure 4 Seismic zoning Of Addis Ababa as per.RADIUS Project (1999)

Journal 0/EEA, VoL 20, 2003

22 Samuel Kinde

REFERENCES

[1] EBCS-8:1995, Code of Standards forSeismic Loads, Ministry of Works andUrban Development, Addis Ababa, Ethiopia,1995.

[2] ESCP-1:l983, Ministry of Works andHousing, Addis Ababa, Ethiopia, 1983.

[3] Gouin, P., Earthquake History of Ethiopiaand the Horn of Africa, InternationalDevelopment Research Center, Ottawa,Canada, IDRC- U8e, 259p, 1976.

[4] Fekadu Kebede, "Seismic HazardAssessment for the Horn of Africa",Proceedings of the Ethiopian Association ofSeismology and Earthquake Engineering(EASEE) on Seismic Hazard Assessmentand Design of Structures for EarthquakeResistance, Addis Ababa, Ethiopia, pp 25­38, February 21, 1996, Addis Ababa,Ethiopia.

[5]. Fekadu Kebede, "Hazard Maps of SpectralResponse Acceleration' for Ethiopia",Proceedings of the Second Symposium ofthe Ethiopian Associati9n of Seismologyand Earthquake Engineering (EASEE),Addis Ababa, Ethiopia, April 4, 1997.

[6] Panza, G.F., Vaccari, F., Costa, G.,Suhadolc, P., Fall, D., 1996. "Seismic inputmodelling' for zoning and microzoning",Earthquake Spectra 12, 529-566.

[7] Fekadu Kebede and Laike Mariam Asfaw,"Seismic Hazard Assessment for Ethiopiaand the Neighboring CountrieS', Sinet ­Ethiopian Journal of Science, 19(1), pp 15­50, 1996, Addis Ababa, Ethiopia.

[8] Laike Mariam Asfaw, "Seismicity andEarthquake Risk in the Addis AbebaRegion", Sinet - Ethiopian Journal ofScience, 13(1), pp 15-35, 1990, AddisAbaba, Ethiopia..

JournalofEEA, Vol 20,2003

((

[9] IDNDR RADIUS Project, Addis AbabaCase Study, Final Report. Prepared by AddisAbaba RADIUS Group, September 1999.

[10] Laike Mariam Asfaw, "Site Amplification atthe Western Escarpment of ihe East AfricanRift System near Addis Abeba", Letters tothe Editor, Bulletin of the SeismologicalSociety of America, Vol. 72, No.1, pp. 327­329, February 1982.

[11] Bekele Mekonnen, "EBCS-8: 1995 - Basisof Design", Proceedings of the EthiopianAssociation of Seismology and EarthquakeEngineering (EASEE) on Seismic HazardAssessment and Design of Structures forEarthquake Resistance, Addis Ababa,Ethiopia, pp 25-38, February 21, 1996,Addis Ababa, Ethiopia.

[12] International Conference of BuildingOfficials (ICBO), Uniform Building Code,1997Edition, Whittier, CA.

[13] "Reglamento de Construcciones para elDistrito Federal [Construction Regulationsfor the Federal District (1993)]", MexicoCity, Mexico.

[14] "Normas Tecnicas Complementarias paraDiseno por Sismo [Auxiliary Technical

, Notes for Seismic Design (1987)]", MexicoCity, Mexico.

[15] Standards Association of New Zealand,General Structural Design and DesignLoadings for Buildings NZS4203: 1992.

[16] NBCC-1995, The National Building Code ofCanada, Canadian Commission on Buildingand Fire Codes. '