The Northern Sumatra Earthquake of March 28, 2005

EERI Special Earthquake Report — August 2005

Learning from Earthquakes

Editor’s Note: Beginning with the killed around 1,000 people while they occurred. Right after the earth- March issue and in each one sub- were either sleeping or trying to run quake, electric power was cut off sequently, we presented six reports from the second and third fl oors. In and telecommunication disrupted. by the many teams that observed Simeulue, some house shops col- In Gunung Sitoli and Teluk Dalam, the effects of the December 26, lapsed, and about 100 people were Nias, the damage was concentrated 2004, earthquake and tsunami in killed. There was a rush to exit in certain areas close to the beach. countries around the Indian Ocean. houses and run to higher places be- In Simeulue the damage was also This report adds information on cause the people feared a tsunami concentrated in areas close to the structural damage, largely from similar to that of December 26, 2004. beach. Strong aftershocks (32 km shaking, on two islands west of There was a local tsunami, but not from Gunung Sitoli [M 6.0] and 41 Sumatra in a subsequent large of the size or ferocity of the previous km from Sinabang (M 5.5]) did not earthquake in roughly the same one. cause any further damage to either epicentral area as the December damaged or undamaged buildings. quake. Publication of this report is Nias Island is part of North Sumatra supported by funds from the Na- Province and Simeulue Island is part tional Science Foundation through of Aceh Province. The population of Tectonic Effects EERI’s Learning from Earthquakes Nias is 730,000 and that of Simeulue During the survey in Simeulue as Program under grant # CMS- is 80,000. Most of the people on Nias well as in Nias Island, we observed 0131895. and Simeulue were already asleep, some subsidence in certain very or about to be, when the earthquake limited areas. There were also “pop- Within a few days of this earthquake, a team set out from Jakarta to study the earthquake damage on the islands of Nias and Simeulue. The team leader and author of this report was Teddy Boen, consulting engineer in Jakarta and senior advi- sor, WSSI. Survey team members on Nias were Teddy Boen, G. Hen- dra, P. Idrus, K. John, James Wong, M. Agus, and B. Bernard. Survey team members on Simeulue were Figure 1. Teddy Boen, G. Hendra, and Akil R. Main shock and Karsono. aftershocks of the Nias/ Introduction Simeulue earthquake, On Monday night March 28, 2005 March 28, 2005 at 11:09 pm local time, the Nias and (USGS). Simeulue islands were shaken by a M8.7 earthquake with a depth of 30 km. The epicenter of the earthquake was on the Great Sumatran fault, between the two islands at 2.074°N, 97.013°E (see Figure 1). This is 115 km from Gunung Sitoli, the district capital of Nias, and 84 km from Sinabang on Simeulue. In Gunung Sitoli and Teluk Dalam, many house shops collapsed and

1 EERI Special Earthquake Report — August 2005

Figure 2. “Pop up” beach at Matanurung, East Simeulue. Figure 3. Local tsunami height approximately 1.80 m.

up” beaches in Nias and Simeulue, by the tsunami, and some buildings engineered buildings except for caused by uplift (see Figure 2). Staff collapsed. The local tsunami went district head offi ces, hospitals, and members of NGOs who worked in inland up to 400 m in Hiliamaetani- some other government buildings. several places across Simeulue re- ha. From the damage, it appeared Engineered buildings consist mostly ported that they also observed simi- that the fl ow force was not too of reinforced concrete combined lar pop-up beaches all over the is- strong. In Simuelue, traces of local with masonry walls. Almost 95% of land. tsunami could be seen in Labuan the buildings are nonengineered, Bajau primary school # 2 (see Fig- made of confi ned masonry half- After the earthquake, a local tsunami ure 4). The tsunami height was ap- burnt brick, concrete block, and hit limited areas. In Nias, along the proximately 0.5 m and no evidence timber. road entering Sorake and in Sorake of fl ow force could be seen. itself, the tsunami height was ap- Engineered Buildings: Some en- proximately 1.80 m, as can be seen Damage to Buildings gineered buildings in Nias and Sim- on the wall of house in Hiliamaetani- eulue were damaged during the ha, approximately 1 km from Sorake Two types of buildings were affect- earthquake, and some collapsed. (see Figure 3). In Sorake, several ed: engineered and nonengineered. There are two main reasons for this. building foundations were scoured In Nias and Simeulue, there are few The reinforced concrete buildings

Figure 4. Damage by local tsunami in Labuan Bajau school, Simeulue. Figure 5. District head offi ce in Simeulue.

2 EERI Special Earthquake Report — August 2005

Figure 6. Damage to district head offi ce in Nias. Figure 7. Damage to district head offi ce in Simeulue. were not designed for the seismicity dance with requirements for earth- cement mortar; and 3) timber build- of the islands. We also found that quake-resistant construction (see ings. In some areas, sea sand is poor quality of concrete and detail- Figures 6 and 7). used for concrete mix and, of ing contributed to the collapses. course, results in poor quality Even though Indonesia has a seis- Nonengineered Buildings: Most of concrete. mic code, it was not enforced in the buildings were nonengineered either island. Additionally, the build- that collapsed in Gunung Sitoli in The one or two-story buildings were ing permit system was not yet in Nias, Sinabang City in Simeulue, mostly houses, schools, and com- place. and villages along the road from munity health centers. The roofs of Sinabang to Labuan Bajau and from the houses mostly consist of gal- For example, both district head’s Gunung Sitoli to Teluk Dalam in Nias. vanized iron sheets or thatch made offi ces were damaged (see Figure There are three main categories of of coconut tree leaves; other build- 5). We surmise that in each case, nonengineered buildings: 1) one or ings use galvanized iron sheets. the structure was analyzed as 2-D, two-story houses of burnt brick or These masonry buildings survived with the torsional effect overlooked. concrete block masonry with sand/ the shaking without collapse, From the damage, it was evident cement mortar; 2) two to four-story though some had cracks in the that the concrete quality was low shop houses of burnt brick or con- walls (see Figure 8). and the detailing was not in accor- crete block masonry with sand/

Figures 8a and b. Nonengineered masonry houses damaged and collapsed in Nias.

3 EERI Special Earthquake Report — August 2005

Figure 9. Timber houses collapsed in Nias due to lack of Figure 10. Traditional indigenous house in Nias, not maintenance. damaged.

Figures 11a and b. Traditional indigenous houses damaged due to lack of maintenance in Bawogosali Village, Teluk Dalam, Nias. The two to four-story shop house buildings also use burnt bricks or concrete block masonry with sand and Portland cement mortar. The roofs of these shop houses are Figure 12. usually RC slabs. All the buildings Indigenous house use RC “practical” columns and built during Dutch beams as confi nement. The term occupation, slight- “practical” is used because the ly damaged due to specs are not based on any calcu- lack of mainten- lations and are adopted as a stan- ance in Nias. dard practice. The shop houses are built according to prevailing practice by contractors and foremen, without structural analysis or drawings. In neither Nias or Simeulue was the building permit system enforced;

4 EERI Special Earthquake Report — August 2005

Figure 13. “Pancake” type of collapse of shop houses Figure 14. Another pancake collapse in Gunung that killed many people in Teluk Dalam, Nias. Sitoli, Nias. therefore no structural check was tenance (Figure 11). These types of the fi rst story is completed. Only af- performed to catch poor-quality buildings are not built anymore be- ter several months or years, when materials or poor workmanship. cause of the lack of expertise in this more funds are available, do the generation, and because they are too second-story columns and walls get Timber houses consist of a timber costly to build. The other buildings, built; the process of constructing the frame and timber plank walls, and built by the Dutch some 150 years fi rst story is essentially repeated. galvanized iron sheet or thatch ago (Figure 12), are also earthquake- Some people continue the process roofs. In spite of the closeness to resistant when properly maintained. for the third and fourth stories. Col- the epicenter of the earthquake, the umn reinforcements and stirrups are damage was not signifi cant in rea- Causes of collapse: The main not suffi cient if the structure works sonably built timber structures. Very causes of collapse in the two to four- as an open frame, particularly in the few one-story timber houses col- story shop houses were poor mate- longitudinal direction. lapsed during the earthquakes. The rials and poor workmanship, exacer- damage that occurred resulted from bated by the prevailing procedures In the damaged structures we deterioration of the poorly main- of building in stages, depending on looked at, the length of the column tained columns (see Figure 9). In the availability of funds. The long splices for the upper fl oors con- addition, some timber columns intervals between the construction struction was generally insuffi cient, were not anchored to the founda- of various elements result in weak and the concreting at the splices tion. There were no tie beams to joints. was very poor. This resulted in the tie the columns together, so the shear failure of the columns, partic- columns moved in different direc- First the foundation is built with the ularly at the joints; this was the tions during the earthquake. Some column-reinforcing bars already in main reason for the “pancake” or timber houses were leaning after place. If additional funds are avail- soft-story collapse of shop houses, the earthquake because the con- able, the columns are poured. The particularly in Nias (Gunung Sitoli nections were not appropriately size of the columns average 30 x 30 and Teluk Dalam), that resulted in done. (40) cm with eight reinforcing bars most of the deaths (see Figures 13, of diameter 12 mm and stirrups 14 and 15). Two types of indigenous nonengi- (usually not properly bent) of 6 mm neered buildings still exist on Nias. at an average distance of 20 cm. For one-story confi ned masonry First, the traditional houses built Subsequently, the walls are built. houses, the construction is also some 200 years ago are earth- Construction of some buildings stops done in stages. First the foundation quake-resistant and withstood the after the walls are completed. The is built with column reinforcement shaking (see Figure 10). When next stage is constructing the beams embedded. The column size is 12 x these traditional buildings were and fl oor slabs for the next story with 12 cm with four reinforcing bars of damaged, the main cause was splices for the upper-story columns. diameter 8 mm and stirrups of dia- deterioration due to lack of main- Most buildings are in use as soon as meter 6 mm at an average distance

5 EERI Special Earthquake Report — August 2005

of 20 cm. If enough funds are available, the columns are poured. At a later date, the walls and the ring beams are built and, fi nally, the roof is constructed.

We observed several shop house buildings that were built and supervised by the owners, and did not suffer damage. We learned that they had been built contin- uously, not in stages with long intervals caused by Figures 15a and b. Nonengineered masonry shortages of funds. The shop houses damaged and collapsed due to workmanship and mate- weak columns in Lagundri, Gunung Sitoli, Nias. rials used were also of better quality.

Prospects for Change quake forces identifi ed in the Indo- Infrastructure Damage nesian seismic hazard map. This new Most of the damage to infrastruc- In all areas in Indonesia, a good culture is associated with social sta- earthquake-resistant design feature ture occurred to bridges, and there tus: as their economic condition im- were fi ssures along the roads pro- is available, namely reinforced con- proves, people tend to build masonry crete framing consisting of practical duced by subsidence and lateral houses at least partly because they spreading. Many bridges in both columns and beams. However, in are a sign of status. many places in both Nias and Sim- Simeulue and Nias were displaced from their supports from 30 cm eulue, the columns and beams The reconstruction period is a good were not constructed in accordance to 100 cm. Many bridge abutments opportunity to introduce the building settled during the earthquake, with prevailing practice on Java. In permit system and to start enforcing spite of that, most did not collapse thereby damaging bridge ap- the seismic code. Offi cials must start proaches. The differential settle- in the M8.7 earthquake shaking. making quality control part of the cul- Such masonry structures have be- ment was jointly caused by lurch- ture in order to improve the perfor- ing, liquefaction, and uneven com- come a new culture in Indonesia, mance of buildings in earthquakes. and from past earthquakes it has paction, particularly close to the become evident that they are ap- abutments (see Figures 16 and 17). propriately resistant to the earth- The main electric power generating plant in Gunung Sitoli in Nias was damaged by the shaking (Figure 18). Transformers were slightly overturned (Figure 19) and some gen- erator foundations settled. But the Figure 16. cable network in the city as well as Cracks on along the road from Gunung Sitoli road in Lasara, to Teluk Dalam was not affected, Gunung Sitoli, and after the power plant was re- Nias, caused paired, electric power resumed. In by lateral Teluk Dalam, the cable network spreading. was practically intact. The power- generating plant in Simeulue was not damaged, but some cable network to Lasikin was damaged and took a few days to restore.

6 EERI Special Earthquake Report — August 2005

Figure 17a. Movement of end support, bridge in Kelapa Figure 17b. Settlement of bridge approach in Teupah Street, Gunung Sitoli, Nias. Barat, Simeulue.

Figures 18a and b. Power generating plant in Gunung Sitoli, Nias.

Figures 19a and b. Transformers overturned at power generating plant in Gunung Sitoli, Nias.

7 EERI Special Earthquake Report — August 2005

Figure 20. road. After the earthquake, water Damage to drink- purifi cation was provided by a ing water pipes German NGO, and water was by local tsunami delivered by tank trucks. in Lagundri, Hiliamaetaniha, One part of the newly constructed Nias. jetty at Gunung Sitoli, Nias, sank into the sea, most probably because the piles were not driven down to the hard layer. Apart from that portion, the remaining jetty was still intact, which was fortunate in itself because hollow piles had been used. These should not, in fact, be used for undersea structures because the thickness of the concrete cover is not suffi cient to protect the Figure 21a. reinforcing bars from corrosion. In Deformed jetty Simeulue, many of the same type due to shear of piles supporting the jetty were failure of column sheared at the connection between joints, Simeulue. the pile head and the piles, because the upper parts of the columns were already deteriorated and weakened (see Figure 21).

Airport runways at Gunung Sitoli in Nias and Lasikin in Sinabang were not damaged; airplanes were able to land and take off right after the earthquake. At Lasikin Airport, some buildings collapsed, namely the terminal building (Figure 22), a storage building, and some airport Figure 21b. staff residential houses. Shear failure of the column joint, Simeulue.

Figure 22. Collapsed Lasikin Airport Terminal Building, Simeulue.

The water purifi cation plant in Gu- Simeulue, the water purifi - nung Sitoli, Nias, was not dam- cation plant was not acti- aged, but the piping network was vated prior to the earth- disrupted by the tsunami (see quake because of a change Figure 20). Distribution was done in the piping system due to by water tank trucks. In Sinabang, the widening of the main