You are on page 1of 38

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/346004760

Seismic vulnerability of vernacular residential buildings in Bhutan

Preprint · November 2020


DOI: 10.13140/RG.2.2.22077.69607

CITATIONS READS
0 1,055

3 authors, including:

Dipendra Gautam Rajesh Rupakhety


Interdisciplinary Research Institute for Sustainability (IRIS) University of Iceland
104 PUBLICATIONS   1,137 CITATIONS    119 PUBLICATIONS   602 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A simplified approach for capacity assessment of three leaf stone masonry structures View project

Multi-hazard risk assessment along Bhotekoshi River Corridor (Sino-Nepal border region) View project

All content following this page was uploaded by Dipendra Gautam on 18 November 2020.

The user has requested enhancement of the downloaded file.


1 Seismic vulnerability of vernacular residential buildings in Bhutan

2 Nimesh Chettri, Dipendra Gautam*, Rajesh Rupakhety

3 Abstract

4 This study presents an overview of seismic hazard in Bhutan and vulnerability of

5 Bhutanese vernacular constructions derived from damage caused by two earthquakes that

6 occurred in 2009 and 2011. Interpretations of damage and seismically vulnerable features

7 of such constructions are provided from field observations. The taxonomy and

8 vulnerability of all existing Bhutanese buildings are presented using a newly developed

9 classification system. Empirical seismic fragility functions for vernacular stone masonry

10 buildings are constructed using the damage data of 2009 and 2011 earthquakes. The sum of

11 observations highlight that Bhutanese masonry buildings are highly vulnerable even in

12 moderate far-field earthquakes.

13 Keywords: seismic vulnerability; fragility function; vernacular construction; stone

14 masonry; rammed earth; Mongar earthquake; Sikkim-Nepal border earthquake; Bhutan.

15 1. Introduction

16 Bhutan is exposed to earthquakes originating in the eastern fringe of the Himalaya. The Himalayan

17 arc is one of the most active seismic regions in the world. The major shear zones of the Himalayan

18 arc are the Main Himalaya Thrust (MHT), the Main Boundary Thrust (MBT), the Main Central

19 Thrust (MCT), the South Tibetan Detachment System (STDS), and the Indus-Tsangpo Suture

20 Zone (ITSZ). These features span the entire Himalayan arc including the Bhutan Himalaya. An

21 out of sequence thrust called the Kakhtang Thrust Fault (KTF) located between the MCT and

1|Page
22 STDS is a peculiar feature in the Bhutan (Grujic et al., 2002). Drukpa et al. (2006) report

23 significant mid- and deep crustal strike-slip deformation in the Indian Plate in this region. This

24 deformation is responsible for mid-crustal to deep strike-slip earthquakes. Diehl et al. (2017)

25 indicate potential segmentation of MHT between eastern Nepal and western Bhutan by an

26 obliquely striking dextral Dhubri-Chungthang Fault (DCF). Bilham and England (2001) noted an

27 apparent seismic gap between the epicenters of the 1934 Bihar-Nepal Earthquake and the 1950

28 Assam Earthquake. The Bhutan Himalaya lies in this apparent gap where instrumentally recorded

29 seismicity is low compared to the rest of the Himalaya. However, recent studies using paleo

30 seismic (Berthert et al., 2014) and historical (Hetenyi et al., 2016) records provide evidence of at

31 least two large earthquakes in Bhutan. These earthquakes occurred around 1150 and 1713 and

32 were about 8 in magnitude. Diehl et al. (2017) conclude that Mw 7-8 earthquakes can occur in the

33 area between the DCF in the west and the MFT in the east of Bhutan.

34 Information about seismic hazard and vulnerability of constructions in Bhutan is very limited in

35 the existing scientific literature. Similar to many other countries in the Hindu Kush Himalaya,

36 vernacular constructions are common in Bhutan. Such constructions have been found to be very

37 vulnerable to seismic ground shaking. Seismic vulnerability of vernacular constructions in the

38 Himalaya has been reported in studies such as Ali et al. (2013), Ahmad et al. (2012a, b), Gautam

39 and Rodrigues (2018), Gautam et al. (2018), Gautam (2018), and Htwe Zaw et al. (2019). These

40 studies focus on vernacular buildings in the central and western part of the Himalaya. Seismic

41 vulnerability of vernacular buildings in Bhutan has not yet been reported in the scientific literature.

42 A preliminary study (GHI, 2012) that studied potential earthquake scenarios in Bhutan reported

43 that widespread destruction can be expected across the country if an earthquake larger than

44 moment magnitude 7 occurs in the country. Many residential buildings in Bhutan are either non-

2|Page
45 engineered or poorly engineered. In rural areas, most of the dwellings are vernacular constructions

46 made up of rammed earth, stone masonry, wattle and daub, bamboo and timber (Chettri and

47 Thinley, 2019). Chettri and Sharma (2019) report some observations on seismic vulnerability of

48 Bhutanese vernacular buildings. Fatalities and serious injury due to building collapse of

49 vernacular stone masonry and adobe buildings have been observed in past earthquakes (Sferrazza

50 Papa and Silva, 2018; Maqsood and Schwarz, 2010; Gocer, 2020; Gautam and Chaulagain, 2016;

51 Gautam et al.,, 2020; Varum et al., 2017; among others). Such constructions are common in

52 Bhutan.

53 Poor seismic performance of vernacular buildings in Bhutan became apparent after two recent

54 earthquakes: Mw 6.1 in 2009 (Mongar earthquake) and Mw 6.9 in 2011 (Sikkim-Nepal border

55 earthquake). Vernacular constructions were the most affected building type during these

56 earthquakes. The Bhutan Building Code does not cover design and strengthening of traditional

57 houses and international codes may not be compatible to the local construction systems and unique

58 features of Bhutanese buildings. In this context, a systematic study of seismic vulnerability of

59 vernacular residential buildings in Bhutan is important to understand and manage potential risk

60 due to impending earthquakes.

61 The main objective of this study is to shed some light on seismically vulnerable features of

62 residential constructions in Bhutan and to present empirical fragility model for the most vulnerable

63 construction type. The empirical model calibrated with data of damage caused by historical

64 earthquakes; an approach that has been widely used for seismic fragility modelling (see, for

65 example, Elnashai and Di Sarno, 2008; Rossetto and Elnashai, 2003; Bessason et al., 2020;

66 Gautam, Fabbrocino et al., 2018; Porter et al., 2007; Elnashai et al., 2004;.

3|Page
67 2. Seismic hazard in Bhutan

68 Seismic hazard in Bhutan is mainly due to the seismic activities concentrated along the MFT and

69 the DCF. Due to lack of strong motion data and ground motion models, quantification of seismic

70 hazard in Bhutan is challenging. The Indian Standard Codes of Practice (Bureau of Indian

71 Standards, 2002), which is used since 1997 for the design of reinforced concrete buildings in

72 Bhutan, assigns seismic zones of IV and V to Bhutan. According to the GSHAP (Giardini, 1992)

73 seismic hazard map, the 475-year return period peak horizontal acceleration in Bhutan ranges from

74 0.16 to 0.4 g. Results of a preliminary probabilistic seismic hazard assessment presented in Goda

75 et al. (2017) indicate 475-year return period peak horizontal acceleration in the range 0.3 to 0.4 g

76 across Bhutan. No other rigorous study of seismic hazard in Bhutan can be found in the existing

77 literature. Bhutan was affected by many large earthquakes occurring in the Hind-Kush Himalaya.

78 The 1713 Arunachal Pradesh earthquake (Mw 7.0), the great Shillong earthquake (Mw 8.3), the

79 1905 Kangra earthquake (Mw 8.0), the great Bihar-Nepal earthquake (Mw 8.4), the 1947 Assam

80 earthquake (Mw 7.8), the great 1950 Assam earthquake (Mw 8.7), the 1988 Eastern Nepal

81 earthquake (Mw 6.8), and the 2011 Sikkim-Nepal border earthquake (Mw 6.9) shook the Himalayan

82 region including Bhutan and caused substantial damage to lives, properties, structures, and

83 infrastructures (Gautam et al., 2016). Historical records of damage caused by earthquakes in

84 Bhutan are rare. Damage assessment for post-disaster needs assessment (PDNA) have been

85 conducted for two recent earthquakes, some details of which is presented in the following sections.

86 2.1. Mongar earthquake (2009)

87 This earthquake occurred at 2:53 pm local time on September 21, 2009. The United States

88 Geological Survey estimated its moment magnitude to be 6.1. in the Richter scale. The epicenter

89 was situated 180 km east of the capital Thimpu, in Mongar District and the estimated depth was

4|Page
90 14 km. Damage assessment after the earthquake is reported in RGoB-UN-WB (2009). The

91 earthquake caused 12 deaths and 47 injuries. Most of the damaged buildings were rubble stone

92 masonry constructions in mud mortar (Joint RGoB-UN-WB, 2009). Seven aftershocks were

93 recorded within half an hour of the mainshock (Rodgers et al., 2014). The Indian state of Assam,

94 West Bengal, Arunachal Pradesh and a few north eastern Indian states were also affected by this

95 earthquake. The effects were also felt in parts of China, Myanmar, and Bangladesh. The United

96 States Geological Survey (USGS) estimated shaking intensity of MMI between VI and VII in the

97 epicentral region. The intensity contours of the Mongar earthquake-based on the USGS Shakemap

98 are shown in Fig. 1. The earthquake affected 4950 households, 25 basic health units, 91 school

99 buildings, more than 800 cultural/religious heritage structures, 22 renewable natural resource

100 offices, and 27 government and public office buildings. The statistics of damage to vernacular

101 stone masonry residential buildings in the nine affected districts is presented in Table 1. In total,

102 4950 buildings were affected; among them 462 collapsed or were severely damaged (beyond

103 repair), 884 sustained major damage, and 1335 sustained partial damage, and 2269 suffered minor

104 damage. Major damage corresponds to very heavy structural and non-structural damage, serious

105 failure of walls, inner wall collapse, and partial structural failure of roofs/floors. Partial damage

106 corresponds to moderate structural damage, heavy non-structural damage, large and extensive

107 cracks in most walls, roof tiles detached, chimney fracture at the roof line, and failure of non-

108 structural elements such as partitions and gable walls. Minor damage corresponds to slight

109 structural damage, moderate non-structural damage, cracks in many walls, falling of large pieces

110 of plaster, and partial collapse of smoke chimneys on roofs.

5|Page
111 2.2. Sikkim-Nepal border earthquake (2011)

112 The 2011 Sikkim-Nepal border earthquake occurred on September 18, 2011 in the area bordering

113 Nepal and Sikkim. The focal depth was estimated to be ~35 km by the United States Geological

114 Survey (USGS). The recorded peak ground acceleration at Gangtok and Kathmandu were 0.15g

115 and 0.05g respectively (EERI, 2012; Gautam, 2017), g representing acceleration due to gravity.

116 Geographical distribution of damage in Nepal, Sikkim, and Bhutan indicated that the rupture

117 propagated eastwards from the epicenter. Reports of damages and losses caused by the earthquake

118 in Nepal can be found in Gautam et al. (2018) and Chaulagain et al. (2018). Damage caused by

119 the earthquake in India is reported in EERI (2012). The earthquake damaged 7965 buildings in

120 Bhutan. This earthquake affected all the 20 districts of Bhutan, but western districts suffered more

121 damage. Haa, Paro, Samste, and Chukka were the most affected districts. This earthquake caused

122 one fatality, 14 injuries, and damage in residential as well as public structures and infrastructures

123 (RGoB-UN-WB , 2011)). Building damage caused by this earthquake was classified into three

124 categories (RGoN-UN-WB, 2011). Minor damage category corresponds to slight damage to the

125 structure which do not pose any structural threat and can be easily repaired. Some examples are

126 fall of the plaster, small cracks in walls, diagonal door and window cracks, hairline corner cracks,

127 and shifting of some roof elements. It also includes partial failure of some non-structural elements.

128 Major damage category includes significant damage including partial collapse, but repairable, of

129 some wall or other load bearing structural elements. Total collapse category includes those

130 buildings that have collapsed or are damaged beyond repair. Of the affected residential buildings,

131 345 collapsed or were damaged beyond repair, 1660 suffered major damage, and 5960 suffered

132 minor damage. Distribution of damage to residential buildings in different districts of Bhutan is

133 presented in Table 2.

6|Page
134 3. Earthquake damage to vernacular buildings in Bhutan and lessons

135 learned

136 Both the 2009 and the 2011 earthquakes have exposed seismic vulnerability of Bhutanese

137 vernacular constructions. Some of the frequent damage types sustained by such constructions are

138 described below.

139 Cracks

140 Many vernacular buildings sustained cracking of walls. Vertical cracks (sometimes corner

141 separation as shown in Fig. 2a) were formed at the corners of walls due to poor connection between

142 two orthogonal walls. Vertical cracks were also noticed in the mid sections of walls next to

143 openings. Lack of proper connection between adjacent walls caused extensive vertical cracking in

144 stone masonry walls (Fig. 2d). Shearing action caused by shaking parallel to the wall causes

145 extensive damage at the corner whereas out-of-plane shaking with high intensity motion can cause

146 collapse of poorly built masonry walls (Bayulke 1978, , Tomaževič and Lutman, 1996).

147 Horizontal cracks were observed in rammed earth walls (Fig. 3). Poor connection between

148 different lifts of rammed earth is one of the reasons behind these cracks. When the joints between

149 different lifts of rammed earth line up vertically, a weak plane is formed, which is susceptible to

150 vertical cracking. Another factor contributing to poor performance of masonry walls is the

151 presence of large openings. Lintel cracks, occurring vertically along the edge of windows and

152 doors were commonly observed in masonry buildings. Some examples of lintel cracks are shown

153 in Fig. 4. Cracks due to putlog holes were also observed in many buildings. Putlog holes are small

154 holes made in the walls to receive the ends of poles or beams used for scaffolding. As shown in

155 Fig. 4d, cracks tend to originate at or near these putlog holes. As shown in Fig. 4d, these cracks

156 can line up and join and cause severe loss of structural function. Examples of walls cracking near

7|Page
157 floor joints are shown in Figs. 4e and 4f. Two factors contribute to such cracks. The first factor is

158 the concentration of stress due to the load transfer at the joist wall interface. The other factor is

159 relative movement between the floor joists and the walls.

160 Wall collapse

161 Walls in many vernacular buildings collapsed during the 2009 and the 2011 earthquakes. Some

162 examples of collapsed walls are presented in Fig. 5. Such failures could be attributed to various

163 reasons such as quality of construction materials, very long spans of walls without cross walls,

164 large openings, etc. Stone walls in Bhutan are usually made of two wythes. Through stones which

165 can be used to integrate the two wythes were lacking in many buildings. Due to the lack of through

166 stones, the two wythes open and the wall bulges and cracks making it vulnerable to failure in shear

167 or out-of-plane collapse. Furthermore, lack of through stones is a problem common to masonry

168 constructions in many parts of the HKH (see, for example, Gautam et al., 2016). Out of plane

169 collapse was observed in long walls which lacked cross walls to reinforce them in shear. Material

170 quality is one of the important factors. Excessive sand content in mud mortar reduces its binding

171 capacity. Walls in some rammed earth buildings, which are common in Haa and Paro, collapsed.

172 Rammed earth walls are weak in shear and are also brittle. Rammed earth walls are laid in layers.

173 When orthogonal walls are constructed independent of each other, corner integrity is

174 compromised. This leads to separation of corners which makes the walls vulnerable to out of plane

175 collapse. For better connection of corners, the layers should be laid out in such a way that if one

176 of the walls in a layer extends to the building outer edge and the orthogonal wall starts from the

177 inner edge; in the next layer, the orthogonal wall should extend to the outer edge. This forms an

178 interlocking pattern between layers, providing structural integrity at the corners. This is,

179 unfortunately, commonly ignored to speed up construction.

8|Page
180 Corner failure

181 Proper structural connections between orthogonal walls is essential to generate box action which

182 is desirable for seismic resistance. Box action ensures smooth and regular load path between

183 different structural elements. When shear force is being transferred from a wall moving in its strong

184 direction to an orthogonal wall, integrity of the connection between them is critical. Lack of proper

185 connection leads to buckling and failure of the corner (see, for example, Bayulke, 1978). Some

186 examples of corner failures are shown in Figs. 6a. 6b, and 6c.

187 Delamination of wythes

188 Stone masonry walls with two wythes were commonly observed to suffer delamination.

189 Delamination is the separation of the two wythes, resulting in falling of one or the both. When the

190 space between the two wythes of a wall are filled with smaller stones rather than being connected

191 by a through stone, the two wythes behave independently and separate from each other. An

192 example of delamination of outer wythe in a stone masonry building is shown in Fig. 6f.

193 Gable failure

194 Gable walls are highly vulnerable to seismic actions especially when they are not connected to the

195 main structural system. The failure is typically out of plane collapse of the gable wall, which can

196 also damage the wall below the gable. An example of gable wall collapse is shown in Fig. 6g.

197 Displacement of Rabsey

198 Rabseys are decorative windows used in Bhutanese architecture. They are usually large and rest

199 their weights on walls. They are often not anchored well to the wall to prevent out of plane

200 movement, but rather held in place by their weight. This lack of anchorage and binding caused

9|Page
201 many Rabseys to move out of plane and get displaced from the wall. An example of a displaced

202 Rabsey is shown in Fig. 7.

203 Dislocation of connection in roofing members

204 Roof trusses in vernacular buildings in Bhutan are structurally weak. Connections between

205 structural elements such as purlins, rafters, and ties are weak and not suitable for lateral load

206 transfer. Some examples of poor connections in the roofing system and the consequent failures

207 are shown in Fig. 8.

208 4. Taxonomy and vulnerability of Bhutanese buildings

209 The number of buildings in the different districts of Bhutan as per the 2017 census is mapped in

210 Fig. 9. The density of buildings is the highest in Thimpu, Samtse, Chukka, and Sarpang. The 2017

211 census data classifies Bhutanese buildings into 13 categories based on construction. The

212 typological distribution based on the 2017 census data is presented in Fig. 10. Reinforced concrete

213 (RCC) buildings constitute 33.4% of the total building population in Bhutan. Stone masonry in

214 mud mortar constructions comprise 21.4% of buildings.

215 Based on construction material, morphology, and anatomy of the existing construction systems in

216 Bhutan, a taxonomical classification as shown in Fig. 11 is proposed. The classification is based

217 on review of construction systems and observed damage. Three main categories are defined, which

218 are further divided into sub-categories. To the best of the authors’ knowledge, vulnerability

219 classification of the different types of buildings in Bhutan has not been reported in the literature.

220 We present a preliminary heuristic classification based on the performance of the buildings in past

221 earthquakes as well as experience gathered from damages sustained by similar constructions in the

222 HKH. Using the EMS-98 scale (Grunthal, 1998), each building typology is assigned most likely,

10 | P a g e
223 probable, and rare vulnerability class. The proposed classification might not be optimal as it lacks

224 sufficient data from Bhutan and relies from information from similar constructions elsewhere, but

225 it can serve as a starting point for more detailed and accurate classification of seismic vulnerability

226 of Bhutanese buildings. According to our estimate, more than 40% of buildings in Bhutan can be

227 classified under vulnerability class A and B, which indicates that even moderate shaking can cause

228 significant damage. This was the case during the 2009 and the 2011 earthquakes. The 2009 Mongar

229 earthquake damaged 22.6% of all buildings in Trashigang district, which lies ~12 km from the

230 epicenter. Similar observations were made during the Sikkim-Nepal border earthquake. In Samtse

231 district, located at more than 135 km from the epicenter, ~30% of stone masonry buildings were

232 damaged. As the current Bhutan Building Regulation does not incorporate seismic provisions, all

233 the existing RC building stocks are placed under vulnerability class C (most likely). The other

234 building class falls mostly under vulnerability class C-D, which signifies that the other

235 construction systems such as the wooden construction, buildings constructed with timber planks,

236 bamboo, wattle and daub, among others may have better seismic performance if their connections

237 and foundation system are properly constructed. Furthermore, danger to life safety is lower in these

238 buildings due to their light weight and lower susceptibility to out of plane collapse.

239 5. Fragility functions for Bhutanese vernacular stone masonry buildings

240 We present seismic fragility functions for Bhutanese vernacular stone masonry buildings based on

241 the damage data collected after the 2009 and the 2011 earthquakes. The data was collected by the

242 Royal Bhutanese Government and other agencies as part of the post-disaster need assessment

243 (PDNA) initiatives. The damage data from the 2009 earthquake was classified into four categories

244 as described in section 2.2. These four categories along with the estimated loss (replacement/repair

245 cost) are shown in Table 1. The data from 2011 earthquake was classified into three categories as

11 | P a g e
246 shown in Table 2. To combine the data from both earthquakes in fragility analysis, the data from

247 the first two categories of the 2009 earthquake, both of which correspond to 100% loss, were

248 combined. In this way, three damage states were defined: DS-1, DS-2, and DS-3. The DS-3

249 damage state corresponds to collapse or damage beyond repair, i.e., 100% loss. The DS-2 damage

250 state represents major damage corresponding to 25-30% loss, and the DS-1 represents minor

251 damage corresponding to 7.5-10% loss. The damage statistics shown in Tables 1 and 2 include all

252 residential buildings, but we use only the stone masonry buildings in fragility analysis as most of

253 the data comes from this building type. We used the damage data from Trongsa, Samtse,

254 Phuentsholing, Gasa, Trashiyangtse, Lhuentse, Mongar, Tsirang, Trasigang, Pemagatshel,

255 Zhemgang, Sarpang, and Punakha districts (see Fig. 9) where stone masonry is dominant.

256 A challenging aspect of seismic fragility modelling based on the observed damage data is lack of

257 reliable information about ground shaking. Ground motion amplitudes or intensity measures (IM)

258 of ground shaking caused by the earthquakes were not recorded in Bhutan. In this work, empirical

259 ground motion prediction equation (GMPE) is used to estimate IMs experienced by the affected

260 buildings. Since there is lack of recorded strong ground motion and therefore a GMPE suitable for

261 Bhutan, selection of a proper ground motion model is challenging. In this work, the GMPE

262 proposed by Ramkrishnan et al. (2019) is used because it is based on data from the northern and

263 central Himalayan region. The GMPE is given as:

264 log Y = −2.135 + 0.437 M − 1.099 log(X + e−0.08M ) ± 0.549 … (1)

265 Where, Y is the peak ground acceleration (g), X is the hypocentral distance (km), and M is the

266 moment magnitude of the earthquake.

12 | P a g e
267 Since the damage statistics are available at district level, PGA at the centroids of the districts were

268 estimated from Eq. (1) and used as an IM for fragility modelling. This model does not incorporate

269 effects of local soil and topography in ground shaking, which can be important in some areas of

270 Bhutan. For example, Phuentsholing, and Gasa regions experienced relatively higher damage than

271 other areas at the same epicentral distance. Phuentsholing Thromde lies on top of Phuentsholing

272 Formation belonging to the Baxa group of rocks. It consists of phyllite and quartzite overlain by

273 about 2 m of reddish-brown topsoil. Site effects in this area have played a role in amplifying ground

274 shaking. Such effects are not modelled in the present study due to lack of data.

275 As noted by several authors (e.g. Shinozuka et al., 2002; Porter et al., 2007; Gautam et al., 2018;

276 Gautam, 2018; Gautam and Rodrigues, 2018); Bessason and Bjarnason, 2016; among others),

277 empirical seismic fragility functions generally follow the lognormal distribution. The fragility

278 function is therefore modelled by a lognormal distribution given by:

ln(𝑃𝐺𝐴) − ln (λ)
279 F𝑑𝑠 (PGA) = Ф [ ] … (2)
η

280 Where F𝑑𝑠 (PGA) is the probability that a building reaches or exceeds a damage state ds when

281 exposed to peak ground acceleration PGA, Ф[·] represents the standard Gaussian cumulative

282 distribution function, λ is the median value of the distribution and η represents the logarithmic

283 standard deviation. The model parameters were calibrated separately for each of the damage states

284 by converting Equation 2 to a linear regression problem by taking the inverse Gaussian cumulative

285 distribution function of each side. The estimated parameters are presented in Table 3. The

286 parameters were estimated using data from 29640 buildings. Among them, 69, 539, and 2094

287 buildings sustained damage states DS-3, DS-2, and DS-1, respectively.

13 | P a g e
288 The corresponding fragility functions are shown in Fig. 12. The functions are calibrated primarily

289 with data from relatively low ground shaking areas. The fragility functions indicate that damage

290 initiation occurs at PGA as low as 0.1g. At a PGA of 0.1g, there is about 8% probability of

291 experiencing some damage. At a PGA of 0.5g, about 15% of buildings would collapse, and about

292 22% would sustain major damage. The low damage scenario as depicted by the fragility functions

293 is related to relatively low height of the stone masonry buildings in Bhutan. Furthermore, biases

294 in the estimation of PGA due to lack of instrumented records is another likely reason. We estimated

295 the PGA for each damage location using a regional GMPE that does not exactly represent the

296 Bhutanese territory. Thus, there is large uncertainty in the estimated PGA values, which translates

297 into uncertainties in the fragility functions. It is worth noting that most of the data comes from

298 areas exposed to low PGA, and therefore fragility at PGA higher than about 0.25g might have been

299 under-estimated.

300 6. Conclusions

301 Observations of damage caused by two strong earthquakes is used to study seismic vulnerability

302 of residential buildings in Bhutan. Observed damage are analyzed to understand vulnerable

303 features of vernacular constructions prevalent in Bhutan. Based on these observations, several

304 construction deficiencies in stone masonry and rammed earth buildings are explained. A taxonomy

305 of Bhutanese residential buildings along with a heuristic vulnerability classification based on the

306 EMS-98 scale is presented. Utilizing the damage statistics from the two earthquakes, seismic

307 fragility functions for stone masonry buildings in Bhutan are presented. Some limitations of the

308 presented model need to be noted. The damage data is limited in the sense that most of it comes

309 from buildings exposed to low ground shaking. Furthermore, the IM estimated from the GMPE is

310 associated with large uncertainties as it does not account for potential effects of local site and

14 | P a g e
311 topography. Lack of strong motion data is therefore a big challenge for seismic hazard and risk

312 assessment in Bhutan. Given these uncertainties, the preliminary findings reported here need to be

313 improved in the future with more accurate modelling of ground shaking, use of damage data from

314 similar constructions in the surrounding countries and by supplementing empirical damage data

315 with analytical models and numerical simulations. Furthermore, collection of damage data in

316 greater detail, for example at higher geographical resolution, will be useful for future studies.

317 The findings show that Bhutanese vernacular constructions, like others in the Hindu Kush region

318 are very vulnerable to ground shaking. Lack of proper connections between orthogonal walls as

319 well as through stones connecting the two wythes of stone masonry walls makes them very

320 vulnerable to cracking, corner failure, and out of plane collapse. At a PGA of around 0.5g, which

321 can be expected to exceed if a large earthquake occurs near Bhutan, about 15% of stone masonry

322 buildings are expected to collapse. This poses great threat of not only economic loss, but also risk

323 of injury and death to the building occupants. Since these buildings are constructed with thick and

324 heavy stone walls, even a partial collapse can be life threatening. This points to the need for

325 improving seismic performance of Bhutanese stone masonry buildings.

326 Improvement in construction methods can be useful in this regard. For example, provision of

327 through stones, timber bands, and proper connections between orthogonal walls can provide box

328 action and improve seismic performance of masonry buildings. Other measures such as proper

329 detailing of joints between orthogonal walls of rammed earth buildings, plugging of putlogs after

330 construction is complete, and proper anchorage of Rabseys to the walls are also recommended.

331 Other retrofitting measures such as containment with wire mesh and mortar; strengthening with

332 cane, timber and bamboo; corner reinforcement with timber and bamboo; reinforcement of

333 openings with timber planks; reinforcement of Rabseys with timber bands, etc. are being

15 | P a g e
334 investigated, and some preliminary results are presented in the contribution by Chettri et al.

335 (2021).

336 Despite the uncertainties in the presented models, the results show that some Bhutanese stone

337 masonry buildings can sustain high damage at relatively low ground shaking levels. Given the

338 threat of very large earthquakes in Bhutan and surrounding areas in the Himalaya, improving

339 seismic resistance of vernacular buildings is of paramount importance in mitigating seismic risk

340 in Bhutan.

341 Data availability statement

342 Some or all data, models, or code that support the findings of this study are available from the

343 corresponding author upon reasonable request.

344 References

345 Bayulke, N. (1978). Behavior of brick masonry buildings during earthquakes. In Seminar on

346 Constructions in Seismic Zones. Rome: International Association for Bridge and Structural

347 Engineering.

348 Bessason, B., & Bjarnason, J. Ö. (2016). Seismic vulnerability of low-rise residential buildings

349 based on damage data from three earthquakes (Mw6.5, 6.5 and 6.3). Engineering

350 Structures, 111(June 2000), 64–79. https://doi.org/10.1016/j.engstruct.2015.12.008

351 Bessason, B., Bjarnason, J. Ö., & Rupakhety, R. (2020). Statistical modelling of seismic

352 vulnerability of RC, timber and masonry buildings from complete empirical loss data.

353 Engineering Structures. https://doi.org/10.1016/j.engstruct.2019.109969

354 Chaulagain, H., Gautam, D., & Rodrigues, H. (2018). Revisiting Major Historical Earthquakes in

16 | P a g e
355 Nepal: Overview of 1833, 1934, 1980, 1988, 2011, and 2015 Seismic Events. In H. Gautam,

356 Dipendra, Rodrigues (Ed.), Impacts and Insights of Gorkha Earthquake (Vol. I, pp. 1–17).

357 Boston, Ma: Elsevier. https://doi.org/10.1016/B978-0-12-812808-4.00001-8

358 Chettri, N., Gautam, D., & Rupakhety, R. (2021). From Tship Chim to Pa Chim: Seismic

359 vulnerability and strengthening of Bhutanese vernacular buildings. In R. Rupakhety & D.

360 Gautam (Eds.), Masonry Construction in Active Seismic Regions (1st ed.). Elsevier.

361 Chettri, N., & Sharma, D. (2019). Damage analysis and interpretation of failure mechanism of

362 Bhutanese vernacular structure post 2009 and 2011 earthquake.

363 Chettri, N., & Thinley, J. (2019). The Comparative Study on Vernacular Dwellings in Bhutan.

364 EERI. (2012). Learning from earthquakes: The Mw 6.9 Sikkim-Nepal border earthquake of

365 September 18, 2011. California. Retrieved from https://www.eeri.org/wp-

366 content/uploads/Sikkim-EQ-report-FINAL_03-08.pdf

367 Elnashai, A. S., Borzi, B., & Vlachos, S. (2004). Deformation-based vulnerability functions for

368 RC bridges. Structural Engineering and Mechanics.

369 https://doi.org/10.12989/sem.2004.17.2.215

370 Elnashai, Amr S., & Di Sarno, L. (2008). Fundamentals of Earthquake Engineering.

371 Fundamentals of Earthquake Engineering. https://doi.org/10.1002/9780470024867

372 Gautam, D. (2018). Observational fragility functions for residential stone masonry buildings in

373 Nepal. Bulletin of Earthquake Engineering. https://doi.org/10.1007/s10518-018-0372-2

374 Gautam, D., Adhikari, R., Rupakhety, R., & Koirala, P. (2020). An empirical method for seismic

375 vulnerability assessment of Nepali school buildings. Bulletin of Earthquake Engineering.

17 | P a g e
376 https://doi.org/10.1007/s10518-020-00922-z

377 Gautam, D., & Chaulagain, H. (2016). Structural performance and associated lessons to be

378 learned from world earthquakes in Nepal after 25 April 2015 (M<inf>W</inf> 7.8) Gorkha

379 earthquake. Engineering Failure Analysis, 68.

380 https://doi.org/10.1016/j.engfailanal.2016.06.002

381 Gautam, D., Fabbrocino, G., & Santucci de Magistris, F. (2018). Derive empirical fragility

382 functions for Nepali residential buildings. Engineering Structures, 171, 617–628.

383 https://doi.org/10.1016/j.engstruct.2018.06.018

384 Gautam, D., Forte, G., & Rodrigues, H. (2016). Site effects and associated structural damage

385 analysis in Kathmandu Valley, Nepal. Earthquake and Structures, 10(5).

386 https://doi.org/10.12989/eas.2016.10.5.1013

387 Gautam, D., & Rodrigues, H. (2018). Seismic Vulnerability of Urban Vernacular Buildings in

388 Nepal: Case of Newari Construction. Journal of Earthquake Engineering.

389 https://doi.org/10.1080/13632469.2018.1498411

390 Gautam, Dipendra. (2017). Seismic Performance of World Heritage Sites in Kathmandu Valley

391 during Gorkha Seismic Sequence of April–May 2015. Journal of Performance of

392 Constructed Facilities, 31(5), 06017003. https://doi.org/10.1061/(ASCE)CF.1943-

393 5509.0001040

394 Gocer, C. (2020). Field Investigation of the Performance of Unreinforced Masonry Building

395 Structures during the June 12, 2017, Lesvos Earthquake in the Aegean Sea. Journal of

396 Performance of Constructed Facilities, 34(5), 04020095.

18 | P a g e
397 Grunthal, G. (1998). European macroseismic scale 1998. Luxemburg.

398 Maqsood, S. T., & Schwarz, J. (2010). Comparison of seismic vulnerability of buildings before

399 and after 2005 Kashmir earthquake. Seismological Research Letters.

400 https://doi.org/10.1785/gssrl.81.1.85

401 Porter, K., Hamburger, R., & Kennedy, R. (2007). Practical Development and Application of

402 Fragility Functions. Structural Engineering Research Frontiers, 1–16.

403 https://doi.org/10.1061/40944(249)23

404 Porter, K., Kennedy, R., & Bachman, R. (2007). Creating fragility functions for performance-

405 based earthquake engineering. Earthquake Spectra, 23(2), 471–489.

406 https://doi.org/10.1193/1.2720892

407 Rodgers, J. E., Lizundia, B. J., Yangdhen, S., Lotay, Y., Hortaçsu, A., & Tshering, K. D. (2014).

408 Post-earthquake Safety Evaluation Procedures for Bhutan Buildings.

409 Rossetto, T., & Elnashai, A. (2003). Derivation of vulnerability functions for European-type RC

410 structures based on observational data. Engineering Structures, 25(10), 1241–1263.

411 https://doi.org/10.1016/S0141-0296(03)00060-9

412 Sferrazza Papa, G., & Silva, B. (2018). Assessment of post-earthquake damage: St. salvatore

413 church in acquapagana, central Italy. Buildings, 8(3).

414 https://doi.org/10.3390/buildings8030045

415 Shinozuka, M., Feng, M. Q., Lee, J., & Naganuma, T. (2002). Statistical Analysis of Fragility

416 Curves. Journal of Engineering Mechanics. https://doi.org/10.1061/(asce)0733-

417 9399(2000)126:12(1224)

19 | P a g e
418 Tomaževič, M., & Lutman, M. (1996). Seismic behavior of masonry walls: Modeling of

419 hysteretic rules. Journal of Structural Engineering. https://doi.org/10.1061/(asce)0733-

420 9445(1996)122:9(1048)

421 Varum, H., Dumaru, R., Furtado, A., Barbosa, A. R., Gautam, D., & Rodrigues, H. (2017).

422 Seismic Performance of Buildings in Nepal After the Gorkha Earthquake. Impacts and

423 Insights of Gorkha Earthquake, 47–63. https://doi.org/10.1016/B978-0-12-812808-4.00003-

424 1

425

426

427

428

429

430

431

432

433

434

435

436

20 | P a g e
437 Tables
438
439
440
441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

21 | P a g e
457 Table 1. Damage to residential buildings caused by the 2009 Mongar earthquake

Major Partial Minor

Collapse damage damage damage

(100% (100% (25-30% (7.5-10%

replacemen replaceme replacemen replaceme Total no. of

District t) nt cost) t cost) nt cost) affected buildings

Chhukha 0 0 0 22 22

Gasa 0 0 23 0 23

Sarpang 0 0 0 2 2

Samdrupjongkh

ar 7 0 10 11 28

Trashiyangtse 21 70 400 106 597

Lhuentse 0 6 24 87 117

Mongar 260 331 413 1253 2257

Punakha 0 5 0 0 5

Tsirang 1 3 2 28 34

Trashigang 147 445 439 572 1603

Pemagatshel 25 24 24 187 260

Zhemgang 1 0 0 1 2

Total 462 884 1335 2269 4950

458
459

22 | P a g e
460

461 Table 2. Damage caused by the 2011 Sikkim-Nepal border earthquake to residential buildings in

462 Bhutan.

Total Collapse Partial damage

(100% (25-30% Minor damage

replacement replacement (7.5-10% Total houses

Districts cost) cost) replacement cost) affected

Haa 170 250 805 1225

Paro 90 706 1238 2034

Wangdue

phodrang 2 52 0 54

Samtse 34 263 590 887

Chhukha 9 83 712 804

Phuentsholing

Thromde 0 16 69 85

Gasa 19 38 171 228

Trongsa 9 17 35 61

Bumthang 0 0 62 62

Sarpang 5 3 48 56

Samdrup

Jongkhar 0 0 18 18

Trashiyangtse 7 52 324 383

23 | P a g e
Lhuentse 0 1 59 60

Mongar 0 26 434 460

Thimphu 0 0 131 131

Thimphu

Thromde 0 10 44 54

Punakha 0 17 691 708

Tsirang 0 23 225 248

Dagana 0 0 117 117

Trashigang 0 100 115 215

Pemagatshel 0 0 32 32

Zhemgang 0 3 40 43

Total 345 1660 5960 7965

463
464

465

466

467

468

469

470

471

24 | P a g e
472 Table 3. Estimated parameters of the fragility functions

Damage grade Lognormal parameter

λ (g) η

DS-1 1.14 1.89

DS-2 1.34 1.28

DS-3 1.40 0.95

473

474
475

476

477

478

479

480

481

482

483

484

485

486

25 | P a g e
487 Figures
488

489

490 Fig. 1 Intensity contours of the Mongar earthquake in Bhutan and neighboring areas (modified

491 from United States Geological Survey)

26 | P a g e
492

493 Fig. 2 a) Vertical crack at the corner of two walls (image by: Royal Government of Bhutan,

494 2009), b) vertical crack near a door opening (image by: Royal Government of Bhutan, 2009), c)

495 vertical corner crack due to rotation of stone masonry wall (image by: Royal Government of

496 Bhutan, 2009), d) vertical crack due to lack of integrity (image by: Royal Government of Bhutan,

497 2009)
27 | P a g e
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513 Fig. 3 Horizontal cracks, vertical cracks and dislocated rammed earth blocks due to horizontal

514 cracks in a rammed earth house (modified from: Royal Government of Bhutan, 2009)

515
516
517

28 | P a g e
518
519 Fig. 4 a) and b) Vertical lintel crack nearby the window in stone masonry (image by: Royal

520 Government of Bhutan, 2009), c) lintel crack nearby the door in stone masonry (image by: Royal

521 Government of Bhutan, 2009), d) vertical cracks in rammed earth house due to putlog holes

522 (image by: Royal Government of Bhutan, 2009), e) cracks originated near the floor joints of a

523 rammed earth house (image by: Royal Government of Bhutan, 2009), f) cracks originating near

524 the floor joints of a stone masonry building (image by: Royal Government of Bhutan, 2009)

525
526
527
528
529
530

29 | P a g e
531
532 Fig. 5 Examples of partial and total collapse of stone masonry walls (images by: Royal

533 Government of Bhutan, 2009)

534
535

30 | P a g e
536
537 Fig. 6 a), b), c) Examples of corner failures (image by: Royal Government of Bhutan, 2009), d),

538 e) plaster failure in rammed earth walls (image by: Royal Government of Bhutan, 2009), f)

539 delamination of stone masonry wythes (image by: Royal Government of Bhutan, 2009), g) gable

540 collapse in stone masonry (image by: Royal Government of Bhutan, 2009)

541

31 | P a g e
542
543
544 Fig. 7 a) Rabsey-Bhutanese decorative window, b) a Rasbey displaced from the wall.

545
546

32 | P a g e
547
548 Fig. 8 a) Weak member connection, b) small prop size and simply supported truss, c), d) collapse

549 due to old and weak truss connection

33 | P a g e
550

551 Fig. 9 Districtwide distribution of building population in Bhutan

552

34 | P a g e
others 4832
mud blocks 4148
rammed earth 6952
wood planks 15155
Building material

cement blocks 8834


bricks 9495
stone with lime/cement mortar 12313
RCC 65142
cardboard 647
plywood 6358
stone in mud 41844
Bamboo with mud mortar 13879
cane/palm/trunk/bamboo 5629

0 10000 20000 30000 40000 50000 60000 70000


#

553

554 Fig. 10 Typological distribution of Bhutanese buildings

555

556

557

558

559

560

561

562

563

564

565

566

35 | P a g e
Building type Indication Description Vulnerability classes
A B C D E F
Others O1 Cane/palm/trunk/bamboo
O2 Wattle and daub
O3 Plywood
O4 Carboard
O5 Wood planks
Masonry M1 Stone in mud
M2 Bricks
M3 Stone in lime/cement
M4 Cement blocks
M5 Rammed earth
M6 Mud blocks
Reinforced RC1 without earthquake
concrete resistant design
RC2 with earthquake resistant
design
Vulnerability: Most likely Probable Rare
567

568 Fig. 11 Taxonomy and vulnerability of Bhutanese buildings

569

570

36 | P a g e
571

572 Fig. 12 Fragility functions for Bhutanese vernacular stone masonry buildings

37 | P a g e

View publication stats

You might also like