Tim Courtney, Colin Taylor and Paul Greening report from Earthquake Engineering Field Investigation Team visits to Turkey and Taiwan, following the recent large earthquakes
Dams in Turkey and Taiwan have been tested in recent months by earthquakes of varying severity. In these locations, structures met with contrasting fortunes. This report provides feedback about the performance of dams in the regions that were subjected to earthquake loading.
On 17 August 1999 an earthquake of magnitude 7.4 struck the Koçaeli region of northwest Turkey. The epicentre was approximately 7km south of Izmit. The Koçaeli earthquake was the result of a rupture of the northernmost strand of the North Anatolian fault system, which has produced seven earthquakes with magnitudes greater than 7.0 since 1939.
Of 472 operating dams in Turkey, 48 lay within the area affected by the earthquake. The Turkish authorities were quick to announce via the website of the Turkish Commission on Large Dams that none had suffered any damage.
Two of the dams closest to areas where there had been damage to residential and commercial buildings were visited by the Earthquake Engineering Field Investigation Team (EEFIT). These were the recently completed Yuvaçik dam, located within approximately 7km of the earthquake’s epicentre, and the Gokçe dam, located around 55km to the southwest close to the town of Yalova. Gokçe is an embankment dam 50m high and is around 10 years old. Its reservoir capacity is 25.5hm3 but it was impounding only a fraction of this capacity at the time. The dam was found to have suffered no obvious damage dur-ing the earthquake. The intake structure is of a similar design to that on the larger Yuvaçik dam, and while it showed some evidence of cracking, it appeared generally to be very robust.
The fortunes of the dam were in contrast with those of the nearby town of Yalova, which had suffered large amounts of damage. It is currently believed that the level of damage in Yalova itself was due to poor ground conditions. Structures in villages on the hillsides close to the dam were observed to have suffered negligible damage.
The Yuvaçik dam is located close to the epicentre of the earthquake and around 7km from the fault rupture. Seismic design criteria were included at the design stage and the result has been that the dam incurred very little damage. A full programme of damage assessment and remedial works to the dam and its water distribution network is currently in hand.
The Yuvaçik dam forms part of the recently completed domestic and industrial water supply project for the Izmit area, and is one of the world’s largest recent water supply projects. The scheme was devised in response to the swelling population and increased industrialisation of the Izmit bay area. It was realised by a consortium of Thames Water International, Gama Industry and Guris Construction. Thames Water will operate the dam for 15 years, after which its operation will be transferred to Izmit Municipality.
The dam was conceived for water storage and subsequent treatment to provide a potable supply. A by-pass has been installed to allow emergency supplies of raw water to be drawn off the reservoir raw water feed to the treatment plant for industrial supply. The throughput of treated water is 142Mm3 per year.
The dam is a rock and gravel embankment dam with a clay core — this design was chosen due largely to the high seismicity of the area. It was first impounded in June 1998 and was brought into service in January 1999. The dam height is 108m with a live storage capacity of 55Mm3 and a total storage capacity of 66Mm3.
At the time of the earthquake, the retained depth of water was approximately 93m. It has been reported that a tidal wave was induced in the dam with a height of approximately 2.5m.
The dam was designed with a 1.5m freeboard to allow for wave action and settlement over its lifetime. Seismic design of the dam was carried out to the Turkish State Hydraulics Institute (DSI) standards; a horizontal peak ground acceleration of 0.15g was used for the dam, with a lower acceleration used for design of the associated structures.
While no strong motion sensors were present in the dam, a network of survey points on the dam and surrounding valley have allowed preliminary measurements of the dam’s response to be taken. Prior to the earthquake, monitoring of the earthfill embankment over the previous year had measured settlement of 25mm. Following the earthquake, a maximum settlement of the earthfill embankment was measured as approximately 130mm. The earthquake resulted in almost negligibly small horizontal movements of the dam — in the region of 30mm, with the dam recovering to half of this value within weeks of the quake.
Mechanical and electrical works at the dam were designed to DSI standards. A visual inspection after the earthquake indicated that civil and mechanical plant at the water treatment plant had incurred relatively minor damage. The output from the plant was disrupted for only a short time (around two hours) while a shutdown was implemented to carry out damage inspections. An onsite generator was brought into service for three days after the earthquake to supply electricity for the dam and water treatment plant until the mains supply was restored.
From Yuvaçik the water distribution pipeline runs along the north coast of the Sea of Mamara as far as Gebze, and along the southern shore as far as Golcuk. By chance this region coincides very well with the extent of structural damage to residential and industrial buildings to the west of the epicentre. The distribution network comprises approximately 103km of steel pipeline with diameters between 2.2m and 1.2m, and 45km of ductile iron pipeline with diameters between 900mm and 250mm. The whole of the pipeline is buried in excavated trenches in the top few metres of ground.
The system has two major pumping stations and four minor pumping stations. Water supply via the two major pump stations on the pipeline was temporarily interrupted due to low suction pressure at the upstream pump station. Supplies were restored following extensive checks on the control and instrumentation systems.
The main damage to the water distribution network occurred to pipeline buried in the vicinity of the fault. As a result of the large-scale disruption caused by the earthquake to the ground in the vicinity of the surface rupture, these pipelines were subject to large relative displacements. Deflection of the 2.2m diameter treated water main caused a leak at a location where it crossed the fault line. Damage also occurred to a number of ductile iron branches, resulting in leakage. Despite these disruptions to the pipeline, water supplies were maintained.
The pictures on p18 show a length of pipeline that crosses the surface rupture. It is clear that the pipe has suffered significant shear deformation, a close-up view being shown in the second picture. It is currently postulated by the site engineers that concentrated shear at the point of ground rupture caused movement of the pipe within its backfilled trench, resulting in axial extension, strain hardening and axial compression. The pipeline generally withstood these gross deformations, but in the most severe cases, leaks resulted.
The magnitude 7.4 Ji-Ji earthquake that hit Taiwan on 21 September 1999 affected a number of dams, although only one major dam was seriously affected.
The principal feature of the earthquake was the major outcropping of the causative thrust fault along an 80km length. The fault break ran north-south through the centre of the island to the east of the city of Taichung, at the foot of the main mountain range that dominates the eastern side of the island. At the northern end of the fault break, the ground displacements measured around 10m horizontally and 7m vertically.
The Shihkhang dam, essentially a gated barrage located about 20km north east of Taichung, was clipped at its northern end by the fault break, with spectacular consequences. The southern end of the dam rose about 9m relative to the northern end, while the valley width shortened by about 6-7m. The northern gravity dam section was driven into the valley side by about 1.5m. The net effect was akin to an axial buckling failure of the dam.
Despite being ripped apart by the fault movements, the dam failure did not lead to a catastrophic flood downstream. There was no sign of overtopping of the failed section, for the relatively flimsy fencing along the crest was still intact and had no debris on it. A possible explanation is that the reservoir may have in fact drained upstream, since the reservoir floor rose about 10m adjacent to the dam, making its elevation higher than a substantial part of river upstream. Also, the dam was not completely breached. The failed section was still able to retain water to some degree, and this may have led to a relatively slow overall discharge.
Evidence all along the fault break indicated that the ground shaking close to the fault (ie within a few hundred metres) was, on the whole, not very intense. It appears that the large fault displacements occurred relatively slowly over a period of 30-50sec. Further away from the fault, for example near the epicentral region several kilometres from the surface rupture, intense ground shaking was clearly evident from the many building collapses. It is likely, therefore, that the Shihkhang dam did not experience particularly strong shaking, the damage patterns of the dam being primarily due to the large differential displacements.
Apart from the ruptured section, most of the steel radial gates forming the dam were operable. When power supplies were restored, all but three of the relatively undamaged gates were opened, to ensure that the expected flood from an impending typhoon could be passed. The winding mechanisms of those gates that could not be opened were generally damaged by buckling of the drive shaft connecting the gearing on either side of the gates as a result of the cross-valley shortening. The steel gates themselves appeared to have suffered no significant damage.
There was some leakage apparent through an open joint in the gravity dam section and from an outlet pipe. Although the base of the dam had clearly slid about 1.5m into the valley side, there did not seem to be any leakage through the dam-foundation interface.
Sun Moon lake dams
The EEFIT team visited a number of dams forming Sun Moon lake, about 30km east of the epicentre. The two main embankment dams were about 30m high; one had a central concrete core wall. They suffered only superficial settlements, of up to 150mm. The various intake structures suffered no obvious damage. There were reports of rock slides causing operational problems at the power generation plant associated with the reservoir, but the EEFIT team did not visit this site.
Adjacent to the main lake, a small dam, approximately 6m high, failed. This dam impounded what appeared to be a small recreational lake next to a hotel. The dam comprised a concrete gravity section on the downstream side, with an earth slope on the upstream side covered by a 150mm thick concrete slab. The earth slope failed, probably as a result of liquefaction. Significant leakage was apparent from the low level outlet at the centre of the dam, probably as a result of differential movement around it. It was clear the dam could no longer retain its reservoir. This did not constitute a significant hazard, since the main Sun Moon lake was immediately downstream of the dam.