The Taishir hydroelectric project which entered into service in November 2008 is, by most conventional standards, a very small scheme. What sets it apart from other hydro projects is the uniqueness of its location in Mongolia, in particular: Its isolation from any important centres of population, construction materials and resources; and the extreme environmental and climatic conditions to which it is exposed.
Location and access
The Taishir hydroelectric project is an 11MW scheme located on the Zabkhan River at the border between Gobi Altai and Zabkhan provinces in western Mongolia. [See Figure 1]. At the site, approximately 10km upstream of the nearest village, Taishir Soum Centre, the river has cut a 130m deep gorge (Ulaan Boom or Red Gorge) through a massive granitic formation. The regional terrain is essentially treeless and is sparsely populated by nomadic herdsmen. The elevation of the project area is around 1700m.
The site is approximately 1100km west of Ulaan Baatar (UB), the national capital and only significant population centre in the country. Land access to the project from UB is about 30% over paved roads, with the rest being across semi-desert countryside where new trails are made by each passing vehicle. Access by air is from UB via Altai (capital of Gobi Altai province) approximately 55km from the site.
Nearly all of the plant, equipment and materials required for the project were imported through the XinJiang (China)-Bulgastai (Mongolia) border crossing some 400km to the south, via similar trail conditions to those from UB to the site.
The Taishir project was designed to maximise winter generation, with riparian releases through a micro-turbine in summer when the reservoir is filling. The most distinguishing feature of the project is the 55m high roller compacted concrete gravity dam with crest elevation of 1708m, crest length of 190m and a volume of 250,000m3.
The dam creates a reservoir storing nearly 1Bm3 of water at full supply level (FSL) of 1704m. The live storage between FSL and normal minimum operating level (MOL) of El 1696m is 300Mm3.
The 250m long, 5m maximum height earthfill saddle dam across the lower elevations of the northern plateau, was required mainly to direct flood flows to the spillway and avoid future scour in this area.
Permanent access to the dam and powerhouse for operation and maintenance is via a 300m long access tunnel leading from the saddle area to the excavated platform at dam crest level on the right abutment. [See Figure 2]. Powerhouse access is via an electrically-powered platform on the downstream side of the dam.
The spillway, designed to pass the routed PMF outflow of 1450m3/sec, consists of a 75m wide ungated ogee crest discharging via a stepped chute on the downstream face of the dam into a stilling basin at the toe. Located at the upstream toe of the dam, the intake delivers a maximum flow of 30m3/sec to the turbines via a 2.7m diameter steel penstock embedded in a trench under the dam. This low intake setting is possible because of minimal sediment loads in the river and the very large dead storage volume.
At the toe of the dam's right abutment, the powerhouse has 3x3.45MW units plus a fourth of 650kW designed to operate when the reservoir is below MOL for the main units, ie during the prolonged reservoir filling period. [See Figures 4 and 8]. Under average hydrological conditions, annual average energy production will be 37GWh.
The substation is located on the right abutment platform, above dam crest level.
The project was adopted with the objective of interconnecting the above-noted provincial capitals, as well as some smaller communities, with an independent local grid to supplement the limited available diesel generation capacity.
At the time of project commitment, the available power capacity at the two capitals was only 5MW out of a nominal 20MW of old diesel capacity. Furthermore, less that 25% of the total coal-fired district heating capacity was operational.
Some emergency short-term measures were adopted at this time on the understanding that hydro energy from Taishir would form the future backbone of the regional power supply, noting that unless additional supply could be brought online, electricity and heat availability during the winter months would continue to be strictly limited. Obviously, power and heat demand is very low during the summer months and peaks during winter.
Climate and hydrology
Climatic conditions at the site can be summarised as follows:
- Annual precipitation about 200mm.
- Annual temperature range -50°C to +37°C.
- Annual average temperature 0°C.
Due to the fact that the ambient temperatures average well below zero from October through to April, the construction season for RCC was limited to just over five months per year.
Although the contributing basin area is more than 14,000km2, the long term annual average flow in the Zabkhan River at the damsite is only 13.4m3/sec. Even under average flow conditions, and without taking seepage, evaporation and ecological flows into consideration, the reservoir would have taken 18 months to fill to MOL. In fact, some three years were eventually required before the three main generating units could start operating.
Flows are predominantly derived from snow melt: they peak in the summer months and are negligible in winter. This is the opposite of regional demand patterns and thus, any hydro development without substantial inter-seasonal storage and regulation would be of limited benefit. Micro-size hydropower projects are unable to operate during the winter due to freezing of the rivers.
The spillway was designed to pass the probable maximum flood (PMF). The PMF inflow is estimated to be 2200m3/sec and the routed outflow is 1450m3/sec. To minimise the potential for overtopping of the unfinished RCC surface by very cold water, the diversion works were designed to pass the routed 1:10 year flood of 160m3/sec.
In general terms, the site is located in the Altai fold belt in a zone of granitic intrusives. Specifically, it is located in the Ulaan Boom gorge which is about 3.5km in length and up to 130m in depth. The Zabkhan River has cut its channel through the gorge to a bed level of around 1660m at the entrance and 1550m at the exit.
The valley sides have slopes of between 30o and near-vertical with steep cliff sections up to 30m in height. The eroded surface results in irregular rocky knolls, chasms and talus slopes.
The river channel is between 70 and 150m in width and has a relatively flat and shallow bed with sand and gravel bars exposed on the insides of bends. The channel sides are vertical banks to 1.5m in height indicating continuing down-cutting. There are also terraces of alluvial gravels and sands up to 50m width. Upstream of the gorge, the valley opens out to a long broad alluvial plain with a meandering channel extending for about 20km. This topographic feature is well suited for a large volume reservoir.
The main rock type at the site is part of large granite intrusion of Devonian Age, the highly fractured Ulaan Boom granite. This granite body forms a characteristic elevated and rocky terrain extending about 3km north of the Zabkhan River and about 11-12 km to the south. Away from the gorge, the rock mass forms a relatively level plateau at about 1700-1800m elevation.
The Devonian age granite intrudes a series of Pre-Cambrian and Cambrian sedimentary rocks. These form the prominent outcrops along the Zabkhan River between Ulaan Boom and Taishir village. Units of dolomite and limestone within these formations were exploited to produce the inert fines used in the RCC mix to improve workability, compactability and density.
The Taishir project site is located in an area of relatively low seismicity (Table 2). Lugeon values determined from water pressure packer tests ranged from 14 to 29, indicating that the bedrock in the dam foundation is of moderate to high permeability. Both consolidation and deep curtain grouting, were accordingly specified. Consolidation grout holes were generally 5m in depth and the two-line grout curtain was extended to a depth equal to at least 60% of the water head.
Extreme construction conditions
The Taishir RCC dam is, to the knowledge of the authors, subject to the most extreme temperature range for any concrete dam in the world. Because of the extreme climate, the concrete construction season was limited to less than 6 months per year. The original programme was for dam construction to be completed in a single working season, however a late start and slow early placement rates meant that placement had to extend into a second season.
The RCC mix was a low cementitious content (75kg/m3) with an impervious upstream membrane [See Figure 8]. This approach was adopted largely in response to climatic conditions but also because of the economic unavailability of supplementary cementitious materials (SCMs).
Limited funding for the design phase meant that pre-construction foundation investigations were not as extensive as would normally be expected for a dam of this size. The foundation when exposed proved to be more highly fractured, faulted and irregular than anticipated from drilling results [See Figure 5], leading to increased foundation preparation time and increased quantities for grouting and levelling concrete.
Equally, there was insufficient funding to permit RCC trial mixes to be carried out prior to awarding the construction contract The final mix, developed on the basis of laboratory trials, followed by several test fills of about 100m3 each, performed well during construction, including exposure to -45°C during the unplanned winter break.
Obviously, the main construction problems were climatic and logistic and the main challenge was construction of the dam. The specification philosophy for dam construction was based on the expectation that the selected contractor would have limited RCC experience and on the topographic requirement that all RCC production facilities be located in the saddle area. Key specification requirements included:
- Construction of a 300m long access tunnel from the saddle area to dam crest level (also required for permanent access).
- RCC delivery from the mixing plant in the saddle area was to be via conveyor through the access tunnel and then by vacuum chute into trucks on the dam surface.
- Trucks on the dam were required to stay on the dam.
Foundation preparation delays meant that the dam could not be completed, as planned, in a single construction season. Extensive protection measures were thus required to insulate the unfinished dam surface during the winter months. In the following spring, it was determined that the horizontal RCC surface had been effectively protected from freezing.
The excellent cooperation between supervisor and contractor helped to ensure that a potentially problematic project advanced in accordance with the programme (except as noted above) with very little discord.
The final contract price for the dam project was within 10% of the pre-tender estimate. The main reasons for the overrun were:
- The unexpected foundation preparation difficulties, which led to both time and cost penalties. For example, over 3000m3 of levelling concrete was eventually required compared to the budgeted 300m3.
- As an indirect result of the foundation preparation difficulties, it was necessary to extensively insulate the exposed dam surface to withstand thermal damage during the unplanned winter hiatus.
The Taking-Over Certificate for the dam was issued in November 2007 when the civil works were substantially completed. Certification was conditional upon Sinohydro completing remaining works and required remedial measures. In July 2011, the Mongolian State Inspection Committee confirmed that the Final Certificate could be issued conditional upon successful implementation of left abutment (LA) seepage remedial works.
The LA seepage is the only significant problem area revealed to date [See Figure 9]. At the time of the July 2011 inspection, seepage rates at the foundation-abutment interface suddenly increased to 5l/sec when the reservoir reached El 1690m. Although the observed seepage would generally be considered acceptable, with the extreme Mongolian winter conditions, it is desirable that measures be undertaken to ensure that this seepage is controlled/discharged such that rock surface joints do not deteriorate over the long-term
Sinohydro demobilised from the site at the end of 2009 when the reservoir had only reached El 1685m and the rate of filling was being severely impacted by drought conditions prevailing since the start of impoundment. Although the civil works contractual warranty period expired two years following issuance of the Taking-Over Certificate, Sinohydro committed to remobilise for any required remedial works until the reservoir had reached FSL. At the time of writing, the highest reservoir level attained is El 1701.7m (September 2011).
Despite the LA concern, total seepage within the drainage gallery has remained relatively constant at generally less than 2l/sec. Furthermore, foundation piezometric levels have been relatively constant to date.
In consideration of the above, it is probable that the source of LA surface seepage is through the abutment rock mass above the drainage gallery level. The remedial work concept is to intercept and collect rock joint seepage and discharge flow downstream of the dam.
Sinohydro is planning to undertake remedial pressure relief work in the summer of 2012. Supplemental pressure grouting of the LA is not presently envisaged.
A wealth of experience has been accumulated from working on the Taishir hydro project. Generally, design considerations for a dam project, regardless of size, are very similar and there should never be any 'short-cuts' in the process.
The project has shown that RCC dams can be successfully designed and constructed in extreme environmental conditions. Some useful lessons include:
- Inadequate foundation investigation programmes inevitable lead to unforeseen construction difficulties and frequently to performance problems.
- Trial mix programs should be started as early as possible.
- The low cementitious approach to RCC mix design is appropriate where SCMs are not economically available and alternative dam types are not feasible.
- It is questionable whether the adopted concept would have been successful without the installation of a watertight membrane of proven quality on the upstream face of the dam.
- Even with low cementitious mixes, retarders can be very helpful.
- It is very difficult to get grout enrichment to work efficiently with low cementitious mixes. The use of face mix is preferred.
- High-level checks on night shift production are essential.
- When a project involves a large reservoir on a small river, the "hydrology risk" to the owner, designer and contractor should be carefully considered prior to project commitment, noting that extended reservoir filling duration due to drought means deferred revenue to the owner and extended commitment of resourced by the other parties.
However, given the applicable budget restraints, there is not much that the authors would do differently if staring the project over today.