Ulley reservoir is located to the south of Rotherham town centre and was originally built to provide a water supply to the town. It was constructed between 1871 and 1875 and comprises an earth embankment with a puddle clay core. The embankment is 205m long and the maximum height is 16m retaining a reservoir of 580,000m3. It originally had two spillways which were located one on each mitre, discharging into the natural watercourse at the toe of the embankment.

The area was subject to coal mining and in 1943 a new spillway was built downstream of the left mitre. The arisings were placed on the crest of the embankment to raise it to counter the effect of mining subsidence. In 1969, it was realised that the 1943 works had not raised the clay core so works were undertaken to add ‘plastic concrete’ to the top of, and keyed into, the clay core. As a result of the various works, the operational top water level was progressively lowered.

The integrity of the dam was threatened during heavy rain on 25 June 2007 when storm flows in one of the spillways damaged the spillway walls. The spillway that failed was located on the left hand mitre and had a chute gradient of 1 in 5 followed by a level section which discharged into the natural watercourse. The invert comprised a series of steps made up of stone slabs with a small lip along the downstream edge and incorporating a low flow pipe within the lip. The spillway had stone masonry walls made up of tapered units to provide an appearance of ashlar masonry. The stability of the tapered units depended on a good packing of mortar behind the visible face. When originally constructed, the mortar would have been a lime mortar and over the years this washed out leaving an open structure which was less stable than when originally constructed.

During the incident, a single spillway took all of the flow from the reservoir. The water in the chute did not flow smoothly as it was disrupted at each lip. The effect was to increase turbulence and increase the air content of the flow ­causing the flow to bulk as it reached the bottom of the chute. Photographs taken before failure showed the water to be splashing outside the chute and well above the side walls. The capacity of the downstream level section was less than the volume being delivered down the chute. Water levels in the spillway rose at the base of the chute until the walls were overtopped.

Progressive failure

It is not clear what happened next. It is possible that the turbulent flow within the spillway was enough to destabilise the masonry walls, or that overtopping resulted in the ground outside the walls to be washed away and the walls being pushed over. It is known that the start of the problem was at the toe. A wall progressively failed working upstream along the chute and the unprotected embankment fill behind the walls was washed away by the water. The spillway was founded on natural bedrock which did not erode and this contributed to containment of the scouring process.

The supervising engineer called for a section 10 inspection to be carried out under the Reservoirs Act and the remedial works were designed to address the conclusions and recommendations of the section 10 report. In the interests of safety, areas of focus included work to the spillways, core, rip-rap and drawdown facility plus other more minor matters. The design process took each structural element in turn and determined what was present on site and compared it to what was needed to allow the reservoir to operate safely as a category A reservoir.

Spillway design

The first step was to determine the design flood and hence the design outflow to be accommodated by the spillway. A flood study was undertaken to assess design flows and also generate the level rise in the reservoir to inform design decisions on the top level of the waterproofing (the puddle clay core). There was no information about the clay core and a geotechnical investigation was undertaken to obtain the material parameters of not just the core but the shoulder material as well.

The site is within a country park and a decision was taken that the rehabilitated reservoir had to be able to operate without a wave wall. There had not previously been a wavewall and to introduce one would have detracted from the setting as a country park. The crest was also identified as being very narrow and it was agreed to increase the crest level by 0.5m to provide a safer working platform. The spillway width was determined to limit the rise in water level during a PMF event to below the reduced crest level taking into account wave run-up. This produced a weir length of 20m and maintained the final water level at the same level as existed prior to the June 2007 incident. The new spillway has to accommodate a PMF of 125m3/sec.

Several options were evaluated for the location of the new spillway. There were complex issues associated with putting it in either abutment. On the left, it would clash with the existing 1943 spillway which still provided a useful function for removing modest storm discharges. On the right, there was steep sidelong ground which would have presented major geotechnical challenges to insert a new spillway. Also on the right, the width of land available was limited which would have required a less efficient weir arrangement. The decision was made to line the new spillway up with the receiving watercourse. This avoided any bends, provided an efficient weir and removed the water quickly away from the toe of the embankment by being the shortest route from the water body to the receiving watercourse.

By placing the spillway in the embankment, the existing crest footpath was severed and a new timber clad footbridge was provided to maintain public access. The spillway is of reinforced concrete construction with joints to allow a limited amount of rotation to take account of any future long term embankment movements. The view was taken, based on monitoring information, that only nominal consolidation was likely to occur given the embankment was built in 1875 and load was being removed to construct the spillway. The joints incorporate waterbars with dedicated underdrains at each joint to enable any seepage to be monitored and prevent seepage permeating into the earth fill.

The clay core had been raised in 1969 using a ‘plastic concrete’. The geotechnical investigation showed that this concrete had hydrated in places and was therefore brittle and unable to act with the underlying clay core as an effective waterproofing barrier. Also, by hydrating, it developed the ability to bridge over the core; four small seepages were observed in the scour hole during the incident and the ‘plastic concrete’ in the corresponding point in the core had hydrated. In addition, the quality of construction of the ‘plastic concrete’ was not perfect because trench sheets had been left in place which had corroded.

The geotechnical investigation comprising boreholes and lab testing found that the original clay core was in better condition than expected and that the shoulders provided a good filter. It also found that the embankment is founded on broken rock which acts as a large drain under the embankment and reduces the hydraulic gradient across the core. The designers considered the risk of hydraulic fracture and piping failures.

Definitive answers were difficult to find using the laboratory test results for particle size distribution and permeability but, based on published sources for other embankments, it was possible to identify a probability of failure using an event tree. This was found to be low. In light of this, a decision was taken not to touch the clay core but to just replace the ‘plastic concrete’, using a bentonite slurry. The ‘plastic concrete’ was excavated from the core and replaced by the bentonite. To ensure an efficient watertight barrier, the bentonite was keyed into the clay. Also, it was important to ensure that the interface between the bentonite and the new spillway would remain watertight. This was achieved by sloping the outer face of the spillway structure and incorporating a waterbar in the outer spillway face.

The drawdown system had used the original pipe work. There was corrosion to the iron pipes and an analysis of the system showed that the drawdown capacity was limited. The new design requirement was to be able to achieve a drawdown rate of 1m per day during Q10 inflow conditions. This means that for 90% of the time, a 1m per day rate could be achieved.

Various options were considered. Since the reservoir was not used for water supply, then there was only an operational requirement for very limited drawdown of the reservoir. However should there be an emergency, then the reservoir would need to be drawn down significantly. There was also the problem that the reservoir is now home to many fish and there are environmental constraints on emptying the reservoir. The design team wanted to limit any effect on the waterproofing system from the construction of a new draw-off pipe and this led to the decision to retain the existing pipe route through a downstream tunnel and avoid new penetrations through the waterproofing. The downstream portal of the tunnel was appropriately located to allow the drawdown pipe to discharge into the new stilling basin.

On the upstream side of the crest, the decision was made that for the normal operational range of water levels, flows could be controlled using a gravity system but for the rare occasions when a greater drawdown might be required, this could be achieved by using a siphon system. The designed drawdown was one pipe which can be operated under either gravity or siphon conditions, depending on reservoir water levels.

The original rip-rap was limited in extent and new provision was based on the flood study output. Wave calculations were used to identify a stone size for the armour stone which in turn was bedded on a graduated layer of filter stone. The stone used for the rip-rap is a dense Pennine sandstone obtained from near Huddersfield.

The rehabilitation scheme started in September 2009 and was completed in June 2010. The detailed investigations and design work, including the building of a working model on the new spillway, was carried out by consultants Ove Arup. Ringway contractors carried out the spillway work.

The authors are Dave Crook, Supervising Engineer and Jon Horrocks, Associate Director, Arup, 13 Fitzroy Street, London W1T 4BQ, England.