Finding a replacement5 February 2010
The design and construction of a replacement dam on difficult foundations was recently completed at a salmon smolt facility in Scotland. RM Doake from AECOM explains how the design process was shaped by working in such an environmentally sensitive area
The new Loch Coire nan Arr dam is an embankment dam about 175m long with a maximum height of 6.5m, built at the outlet from a natural loch in Scotland. It has been developed by the owners, Lighthouse Caledonia, for augmenting the fresh water supply to their salmon smolt facility at Russel Burn, Kishorn. It will increase the reliability of production whilst allowing hydro power development in the future.
Loch Coire nan Arr is located in the west Highlands of Scotland, within the southern reaches of the Applecross mountains just north of Skye. The location of the loch is at an elevation of about 125m AOD within Coire nan Arr, a large glacial bowl (corrie) formed in Precambrian Torridonian Sandstones. This is generally a very tough sandstone and very resistant to weathering. The corrie is formed with high cliffs and very steep slopes rising over 600-700m above the loch and the floor is covered in glacial moraine and tills with extensive peat bogs in flatter areas. The area is designated as a Site of Special Scientific Interest (SSIS) under UK legislation and Special Area of Conservation (SAC) under European legislation, as well as being of outstanding landscape value.
During the mid and late 1990s, the natural Loch Coire nan Arr was raised by a progression of small structures to augment river flow for water supply to the smolt facility some 1.5km away near the shore of Loch Kishorn. Abstractions for supply are from an intake lower on the Russel Burn near the smolt facility, and water is released from the loch as required to maintain the required river flow at the intake.
The original dam developed into a concrete, steel and timber retaining wall structure extended on the left side by a low embankment formed of locally-won glacial moraine materials. The dam had a maximum height of 2.6m, and an overall length of about 80m. All the structures were founded at a shallow depth on the glacial deposits. In this form, the loch had a surface area of about 0.2km2 and a storage capacity above natural ground level of some 370,000m3.
The principal outlet was a 750mm diameter pipe set in the dam wall at the level of the outlet of the loch. Later, in around 2002, outlet flow capacity was augmented by a further 750mm diameter outlet pipe, constructed at a lower level through the right abutment. It was controlled by a sluice gate at the downstream end, some 125m downstream of the dam wall.
The reservoir had its first formal inspection under Section 8 of the Reservoirs Act in 2004, and some inadequacies were identified, notably a lack of flood handling capacity and a vulnerability to seepage. The total freeboard available was only 0.4m and the dam had been overtopped on a number of occasions, such that the embankment at the left side had been rebuilt several times.
Because of its form of construction, the dam did not lend itself to being modified easily for upgrading at the then storage level or being raised. Work ceased until 2006 owing to change of ownership of the facilities but the new owner Panfish Scotland (and the subsequent owner Lighthouse Caledonia) also confirmed a need for additional water storage for their operations and wanted to make provision for hydro power development. In addition, planners and environmental bodies, notably Scottish Natural Heritage (SNH), were keen for the appearance of the dam to be improved. The proposal was therefore put forward to:
• Construct a new dam downstream of the original structure.
• Incorporate a larger spillway.
• Raise the storage level by 1.5m.
• Allow the sensitivity of the area to be taken into account.
Initially, project management, outline design, planning and environmental issues and contract negotiation were by HGA Consultants, Inverness, with AECOM Edinburgh providing the All-Reservoirs Panel Engineer and technical support. During the course of the preparation of the detailed design in 2008, AECOM took on responsibility for all the above aspects.
The final dam is 6.5m high and 175m long, creating a reservoir with a storage capacity of some 690,000m3. The spillway is designed to pass the 1 in 1,000 year flood with full wave surcharge allowance.
The site is on glacial moraine and tills, with areas of peat bog, but no geotechnical information was initially available for the region of the new dam footprint other than from what could be observed on the surface. Geotechnical investigations were therefore developed in 2007 based on a provisional line for the new dam. Bedrock appeared likely to be at shallow depth in the stream bed and left bank region, but under thicker deposits on the right bank. The investigations were organised by HGA and Applied Geology, with the scope of the investigations specified by AECOM. Provision was made for 8 boreholes and 17 trial pits, with allowance for a spare borehole and trial pit.
Discussions with site investigation contractors highlighted the difficulties of drilling through the moraine materials, which contained very large boulders while it was clear that the finer materials contained very little clay. The only method considered practicable for obtaining reasonable sample recovery from boreholes was by sonic drilling.
The drilling and insitu testing was carried out in July 2007 by Boart Longyear. Boreholes were drilled 3m below drift deposits to prove bedrock. In-situ testing included SPTs, undisturbed sampling, and falling head permeability tests. Laboratory testing by specialist laboratories Enverity and Geolabs included PSD, Atterberg limits, moisture contents, dry density relationship tests and 300mm shear box strength tests.
A bedrock of siltstones and sandstones was encountered over the whole site generally at a depth of around 5m and shallower only locally near the river with 1m at the shallowest. The overburden is typically a layer of variable orange-brown sandy gravels and boulders about 0.5-1.5m thick overlying dense – very dense grey-brown silty-sandy gravels and boulders. Local peat pockets also occur, and boulders could reach sizes in terms of metres. All materials were derived from the local Torridonian Sandstones. Although the upper layer was described as ‘made ground’ it is considered that the layer is only locally associated with the construction of the original dam, and that it generally represents moraine of the last phase of glaciation. The finer and denser underlying layer would represent more highly worked glacial till or moraines heavily consolidated by the later glaciation.
The in-situ permeability tests in the glacial materials generally showed permeabilities generally in the order of 1x106 to 1x107m/sec. Assessments of permeability by Darcy’s equation, based on the D10 size, indicated permeabilities in the order of 1x 104m/sec, although it was considered that this method would not take adequate account of the denseness and amount of fines. The laboratory tests showed that the main overburden materials have grading envelopes generally very close to the critical curve for stability against internal erosion, given by the criterion of H/F=1 where F is the percent by weight passing a particular size D and H is the percentage between D and 4xD. However, some samples showed greater potential for internal erosion, with stability indices (minimum H/F ratios) down to 0.43. Characteristic tests showed the materials were non-plastic, and effective stress shear strengths were around c’ = 0, Ø’ = 37°.
A wide range of alternative forms of construction was considered to identify a practical and economic approach. The nature of the glacial materials is such that seepage control in the foundation is difficult and the material is not suitable for the watertight element of an embankment. No other more cohesive materials were available locally.
The potential for very large boulders within the foundations ruled out a sheet pile cut-off. A diaphragm wall approach might have been more practical as such boulders could be excavated, but it would have been extremely difficult to form the bentonite slurry panels in places particularly on the steep right abutment.
The alternative of an open cut excavation to rock was considered for construction of a concrete cut-off wall or installation of a geosynthetic polymer barrier (GBR-P) membrane. However, because of the depth, the water bearing and cohesionless nature of the materials and closeness to a live reservoir, the open cut approach was ultimately considered to be too problematic and risky.
The most viable option was to control seepage by limiting the hydraulic gradient across the foundation. The quantity of seepage is not critical in itself as any flow contributes to the required releases for compensation and river regulation. Seepage through the foundation materials could potentially cause internal erosion by suffusion (migration of fine particles through pores between the particles of the coarser fractions) and/or piping (progressive erosion by end movement within a tunnel or crack).
The usual approach to determining critical hydraulic gradients is to consider upward flow against gravity; however, gravity has little effect in the horizontal flow conditions applicable here. For suffusion, the critical gradient for movement of fines in sandy gravel material under horizontal flow has been reported as 0.17 for stability indices (H/F)min of 0.36-0.5. Critical gradients for piping are in the order of 0.14 for fine sand with a coefficient of uniformity of 3 and 0.37 for gravel. The foundation materials under consideration are broadly graded (coefficient of uniformity generally about 30) and a gradient of 0.14 was adopted as being adequately conservative to cover both mechanisms of internal erosion.
The method adopted for controlling the seepage gradient is by laying a geosynthetic clay barrier (GBR-C) membrane directly on the prepared foundation surface below the embankment. The GBR-C membrane was anchored in a trench at the upstream end and covered generally by a 0.5m thick layer of sandy gravel protective layer, with a 0.2m thick sand filter extending over the downstream end and the adjoining foundation surface. The sand filter was designed to be effective against finer material possibly transported by suffusion, while the membrane protective layer was also designed to act as a filter/drain against the bulk foundation materials where the layer extended further below the rest of the downstream shoulder of the embankment.
The downstream end was terminated at a location set back from the toe of the downstream shoulder of the embankment. This would ensure that there would be adequate overburden pressure to resist uplift if migration of fines led to blockage of the end filter and a build-up of pore pressures in this region. Because of the length of seepage path required, this layout meant that the upstream end of the membrane extended upstream of the embankment construction particularly in the river bed region. The dam alignment had to be developed to keep the membrane at a practical distance from the original dam.
The glacial materials were too sensitive to moisture content to use for fill, and in any case it would have been difficult to obtain permission for borrow pits because of the environmental sensitivity of the area. However, a new quarry had recently been opened nearby to supply armour stone and rockfill for construction of a new ferry terminal on the island of Raasay. The quarry is in Torridonian Sandstone and the rock is highly suitable for such purposes.
The embankment layout adopted is a rockfill embankment formed over the impermeable foundation membrane, sealed with a double-sided textured geosynthetic polymer barrier (GBR-P) membrane on the upstream face, and with a landscaped downstream shoulder zone of spoil material from the excavations and surface stripping. The GBR-P membrane is protected on both sides by thick non-woven geosynthetic textiles (GTX) and the upstream surface protected by stone cover layers and rip-rap.
The GBR-P face membrane is joined to the GBR-C foundation membrane at the toe of the upstream slope by sandwiching the end of the GBR-P between the foundation membrane and a second overlapping layer of GBR-C. The GBR-P membrane is anchored at the crest of the embankment by a concrete beam. The slope of the upstream face is 2.5H:1V, and that of the downstream face of the rockfill is 1.5H:1V. The downstream face of the embankment has been landscaped to a flat slope between 2.5 and about 4H:1V with mixed local peaty material and spoil from the foundation excavations.
The original alignment of the left wing of the dam had been straight, but a deep pocket of peat had been found here during the site investigation. Detailed investigation of peat depths in the region of the left abutment at the start of construction showed the peat was more extensive and deep than had been anticipated. A ridge upstream appeared to consist substantially of coarse moraine material and was very exposed to wave action, and had already been rejected as a suitable alignment. A revised alignment was therefore adopted to terminate on a low ridge downstream, after trial pits confirmed suitable foundation material occurred at a shallow depth
Foundation preparation consisted of stripping off surface peaty deposits and loose orange-brown coarse granular moraine deposits too close to the top level of the dense grey-brown finer glacial till. Some areas of more uniform silty sand occurred in pockets towards the river bed region, and these were removed to expose broadly-graded till. The surface of the till was then excavated down to a sound surface and the surface given a final preparation immediately before laying of the GBR-C. A sound, smooth surface was required for forming a good interface with the GBR-C, and trials were carried out at the start of laying to establish the best method of achieving this, given the inevitable disturbance caused by mechanical cleaning and plant trafficking and the occasional need to make-up around protruding boulders. It was found that a light rolling with a self-propelled vibrating roller reworked loose surface material satisfactorily and produced a uniform smooth surface.
Control of the reservoir level during construction was by the bottom outlet, supplemented as necessary by a new 600mm diameter scour pipe. This had been connected early to the original higher outlet pipe intake so that the scour could be available throughout the main construction period. Constraints on the draw-down to store flood inflows were created by Lighthouse Caledonia’s requirements for a reliable supply to their smolt facility. The loch overflowed only once during construction, at a stage when there were no safety considerations and contingency measures and pre-planning ensured that the part-complete works were not significantly damaged.
The depth to bedrock in the river region was too great for easy foundation of a concrete spillway here and a location further from the river would cause excessive permanent disruption of the environmentally sensitive area. A concrete spillway was in any case regarded as undesirable visually, although the impact might have been reduced by some form of masonry facing. Given the ready availability of large rock, the spillway has therefore been formed by a rock-armoured chute over the central section of the dam.
The rock armour design is similar in concept to the major spillways constructed for the Khasab dams in Oman as presented in an icold paper by EH Taylor with which the author was involved briefly at the time of construction. The design followed the method of Hartung and Scheuerlein as simplified by J Knauss, presented in an ICOLD paper and also quoted in the latest edition of the Rock Manual, CIRIA C683. For ease of construction and to provide a margin for variations in the achieved standard, the layout was developed to allow the rock to be placed with ‘natural packing’ (or dumped embankment) according to the design method, although placing was specified to be in accordance with ‘dense’ criteria of current marine revetment specifications in order to be conservative. Again for ease of specification and control, the rock size was selected to be an industry standard 300-1000kg “heavy” grading as set out in the Rock Manual. These criteria lead to the slope of the spillway chute being 4H:1V. Such a flat slope was in any case considered desirable for ease of construction and tolerance of any damage that might occur from local effects.
The spillweir size was determined for the 1 in 1000 year event, requiring a 27m crest length and 1.2m flood surcharge. With a nominal 0.4m wave surcharge allowance, the total freeboard allowance is 1.6m. The rock armour of the chute is sized for the limiting event of the reservoir just overtopping the embankment, i.e a head of 1.6m on the spillweir (a unit discharge of 3.2m3/sec).
The water flowing in the chute infiltrates into the underlying rockfill embankment and discharges generally into the stilling basin apron area. This water could affect the stability of the rockfill if the induced water table in the rockfill becomes too high. A thick non-woven geosynthetic (GTX) textile is placed between the rock armour and the rockfill as a separation layer, and this layer also acts to limit the rate of infiltration of the water. A window in the GTX layer has been provided at the base of the chute to allow free drainage of the water from the rockfill. This window is covered by a geosynthetic grid as a high-permeability separation layer.
The principal environmental considerations when planning the works have been to minimise disturbance of the area during construction, to provide a shape and form that is as unobtrusive as possible, and to reinstate the site, particularly the drainage systems and peat bog areas, as close as possible to the original.
The original dam has been partly demolished and buried in an upstream platform adjacent to the upstream face of the new dam. Normally, the downstream face of a dam is made regular to allow easy visual detection of any settlement or movement occurring, but in this case a monitoring system on the concrete crest beam provides an adequate indication of the behaviour of the structural part of the embankment. The downstream face of the dam has therefore been landscaped to an irregular profile with boulders to replicate the moraines found elsewhere in the corrie. Vegetation is being re-established by recycling the original topsoil and heather and grass vegetation, supplemented by sowing of carefully selected grass seed. Temporary drainage ditches formed around parts of the site have been reinstated using the material removed.
The contract was negotiated with RJ McLeods Ltd, Dingwall, Highland, and awarded under a NEC3 contract. Work commenced on site in December 2008 and the dam was completed in May 2009.
The reservoir was first filled above the storage level of the original dam in May 2009. Seepage discharges initially showed some fines and colour, but such bedding in soon ceased and all discharges are now clear. Electrical piezometers in an array of the foundation show consistent seepage conditions, with a significant part of the total head drop from the reservoir occurring upstream of the foundation membrane. Total seepage flows have been moderate.
The spillway first operated in August 2009, and has performed well on several occasions with heads of up to 0.3m.
Richard M Doake, Consultant, AECOM, 1 Tanfield, Edinburgh EH3 5DA, Scotland
H Dalgety, Freshwater Manager Lighthouse Caledonia
P Nichol, Site Agent RJ McLeod