HYDRAULIC MODEL STUDY OF BRENT RESERVOIR SIPHON SPILLWAY

¹Research Assistant, Department of Civil Engineering, University of Glasgow, Glasgow, G12 8LT, UK. E-mail: K.B.Koopaei@civil.gla.ac.uk

²Senior Lecturer in Hydraulic Engineering, Department of Civil Engineering, University of Newcastle upon Tyne, Newcastle upon Tyne, NE1 7RU, UK. E-mail: Eric.Valentine@newcastle.ac.uk

³Professor of Water Engineering, Department of Civil Engineering, University of Glasgow, Glasgow, G12 8LT, UK. E-mail: D.A.Ervine@civil.gla.ac.uk

Abstract: A physical hydraulic model study has been conducted to investigate the hydraulic characteristics of the siphons on the Brent Reservoir spillway in order to establish reliable stage discharge relationships. The existing bellmouth siphon system is unsuitable causing the siphons to prime suddenly at discharges around 3m³/s. This is due to the sudden removal of an air pocket from the siphon crown. The model tests were carried out in two stages. In the first stage, the existing geometry is examined. The reservoir levels for priming of each of the existing siphon configurations were established. Based on the results from stage 1 of the experiments, it was concluded that the air inlet requires redesign and various options to improve the air regulation should be considered. In the second stage, various options to regulate air inlet and establish a stable siphon perfomance over the entire range of discharges were considered. It was found that the most stable conditions are provided by a slot being cut in the spillway hood at an appropriate level. This geometry provides excellent air-regulated stability, unimpaired spillway capacity, and is insensitive to tail water level and wave conditions in the reservoir.

Keywords: air-regulated flow, siphon spillway, reservoir, airation, two phase flow, air entrainment

1 INTRODUCTION

The Brent Reservoir is located in Northwest London, close to the M1 motorway, the A406 (North Circular Road) and the A5 (Edgware Road). An aerial photo of the reservoir is shown in Fig. 1.

The dam was constructed in the mid 1830s to form the Brent reservoir (or Welsh Harp) and designed to supply the Grand Union Canal. It is no longer required for this purpose. The reservoir’s main use is for leisure activities.

The siphons were completed in 1936 to protect the dam from overtopping in the event of an extreme flood. In 1939 two of the siphons primed simultaneously and this is the only known priming. Recently, the reservoir level rose to within a few inches of the lower siphon crest level and there was concern that the siphon would prime again and cause flooding downstream.

The aim of this paper is to report on the results of the physical model study on the Brent Reservoir siphon spillway. The results of the physical model study including the experimental set-up are explained in the following. A cross-section through one of the siphons is shown in Fig. 2.

2 PHYSICAL MODEL STUDY

The physical hydraulic model study was carried out in the hydraulics laboratory of the University of Newcastle upon Tyne on behalf of consultants Babtie group and client British Waterways Board. The study has been carried out employing standard physical modelling techniques. A model of a single siphon with adjustable air intake bellmouths is mounted between two tanks representing the reservoir and downstream conditions. Water level control is provided in each tank. The physical model arrangement is shown in Fig. 2.

The siphon is unobstructed on both sides to permit observation and measurement. Water supply is provided by the laboratory constant head system. Flow and water level measurement are achieved by calibrated laboratory apparatus. Flows to and from the model are measured by orifice plates and a v-notch weir.

The model has been constructed to a geometric scale of 1:10. This results in dimensions which satisfy the Froude scaling rules and takes advantage of standard material component sizes. It is also a suitable scale to minimise aeration side effects.

The siphon is constructed in clear perspex and hardwood, in sections with flanged watertight and smooth internal joints to allow dismantling for modification. The siphon model was built by Precision Scale Models Ltd and has been installed between two tanks constructed in high quality marine plywood.

The upstream tank was designed with sufficient area and volume to provide flow conditions which do not artificially influence the behaviour of the siphon. The upstream tank surface area is approximately 5 square metres (2m×2.5m. including flow inlet baffles). This tank has a volume of approximately 5 cubic metres with an allowance for freeboard. These dimensions have been sufficient to guarantee that no local flow boundary influences the siphon flows. The intention was to create a model reservoir which does not surge when the siphon suddenly runs full.

An overflow control weir provided at the upstream edge of he tank enables the reservoir level to be controlled independently of the siphon flow behaviour. It is positioned to isolate the overflow from the siphon model.

3 EXPERIMENTAL PROGRAMME

Stage-discharge measurements have been carried out on the 1:10 scale model which can be configured to represent all of the five prototype siphons. The siphons are identified in terms of both air intake bellmouth geometry and the relative level of the bellmouth to the siphon crest level. This is shown in Fig. 4 with details listed in Table 1.

Siphon No.	Bellmouth length (m)	Bellmouth above crest level (m)	Crest level(m)
1	0.616m	+0.203	38.71
2(&4)	0.718m	+0.102	38.71
3	0.972m	0.00	38.56
5	0.514m	+0.305	38.71

The differences between the siphons involves the geometry of the air regulation intake bellmouths, and in the case of siphon No.3, the crest level. As two of the crest and bellmouth geometries are identical the modelling of four different bellmouths has enabled all of the prototypes to be rated.

In addition to the above existing geometries the case where the bellmouth was removed was also examined. This is equivalent to siphon No.3 with a bellmouth flange level of 38.61m just above the siphon crown level of 39.47m. Fig. 3 shows the four different bellmouths and the 90degree elbow also tested.

The results of the initial stage of the experiments are explained in details in the next section. These showed that the existing bellmouth air intake system is not adequate to prevent the siphons from priming suddenly.

It was therefore decided to employ three different set of geometries in order to allow much larger volumes of air to be entrained to establish partial priming with stable air regulation. This was tested with:

(1) The installation of an additional air vent-pipe connecting the bellmouths to the siphon crest. This was operated alongside the existing vent pipe with each of the bellmouths as shown in Fig. 4. Further variations without the bellmouths (and therefore air intake level at 970mm above crest) and with a 90degree elbow with the same internal diameter as the air pipe replacing the bellmouth were tested.

(2) The bellmouth replaced by an elbow which shown in Fig. 3. This was tested with the intake in a vertical position with the invert at crest level, +200mm, and +400mm above crest level.

(3) Air slots were cut in the spillway hood with the invert at crest level, See Fig. 4. Three vertical slot heights were examined, 200mm, 300mm and 500mm. The 500mm high slot was modified by the insertion of a grid to simulate the requirement that steel reinforcement in the hood not be cut.

All of the above variations were tested with the siphon exit both drowned and free (i.e. high and low tailwater levels) to examine the effects of tailwater level on stability. The results are explained as follows.

4 EXPERIMENTAL RESULTS

Stage-discharge measurements have been carried out on the scale model and stage-discharge curves are obtained for the five different siphon geometries described in Table 1. Two different tailwater levels were examined i.e (a) submerged or drowned siphon outlet. (b) free siphon outlet where the tailwater was below the outlet soffit.

The results for the siphon No 3 only, are shown here in Fig. 5. The character of the results is similar across the range of geometries. All four existing configurations exhibit similar behaviour with some minor modifications. With rising reservoir level (RWL) gravity (weir) flow is initiated as RWL rises above crest level. In all cases the siphon primes when RWL is above the bellmouth entry and blackwater flow (full flow) is established immediately after priming. This applies to both tailwater levels and it is clear that levels between these two extremes where the siphon is still vented downstream have an insignificant effect on the rating. This effect is observable only in the blackwater region and is due to the small overall head difference in this “pipe flow” condition.

Some transient two-phase flow (air + water) is evident at, and just above, priming levels, particularly for the case of the smallest bellmouth (siphon No.5) but this does not persist where RWL is maintained at priming level and is probably suppressed sub-atmospheric flow.

For all other configurations rising RWL establishes weir flow which is followed immediately by blackwater flow on priming. For the drowned outlet, siphons No.1, 2(&4) and 5 prime at levels just above 400mm over the crest. Siphon No.3 would have primed on rising RWL between 50mm and 35mm over crest level (when related to crests Nos.1,2,4,and 5).

For the free outlet, all siphons prime at levels of approximately 630mm over their respective crest levels. No.3 would prime at about 470mm over crest level related to Nos.1,2,4 and 5. Both these sets of results indicate that simultaneous priming of 2 or more siphons could occur.

When RWL is reduced while full blackwater flow is established the rating curve retraces this region but diverges from the rising curve, and exhibits a hysteresis effect as it drops below the priming level. With a continuing reduction in RWL an air regulated region is established which for each configuration is very stable. This is a region of two-phase flow which regulates the flow within a narrow band of RWL until depriming occurs at a level below that of rising RWL priming for each case. Although this behaviour is most extreme for the longest bellmouth (siphon No.3) it is evident for all configurations. It is least apparent where the bellmouth was removed but it does still persist.

The siphon suddenly primes at discharges between 2.5 and 3 m³/s with the sudden sweeping out of the air pocket. When the siphon starts to prime there is no air-regulation as it goes straight to blackwater flow at around 13-14 m³/s. This is unacceptable, and means the air vent pipes are not operating as required. This is simply because the bell mouths are already submerged (in the case of No.1 by 200mm) and hence air cannot get into the vent pipe.

The results so far indicate that the siphon spillways do not exhibit the well-behaved type of rating curve which is required of modern siphon spillway design. Air regulation is not established at the onset of priming. Weir flow is followed immediately by blackwater flow, following the sudden sweep-out of the air pocket, thus creating a very sudden increase in flow rate from approximately 3 m³/s to 13 m³/s. Further, the hysteresis type of rating curve means that the priming point and depriming point are separated.

In order to improve the air regulation inside the siphons steps i) to iii) as explained in experimental programme were taken. The results for step iii) are only shown here in Fig. 6. It is clear that the most stable conditions are provided by a slot being cut in the spillway hood, See Fig. 4. Refinement of this proposal has resulted in the conclusion that a series of smaller slots (420mm wide by 90mm high) which are cut horizontally and centred at 380mm above the crest level provide an excellent solution. It is noted that a departure from the rectangular shape of the smaller slots would not be important but it is the elevation and gross area which are relevant.

5 CONCLUSIONS

The existing bellmouth siphon spillway system in Brent Reservoir is inadequate to prevent the siphons from sudden priming and causing flood downstream. A scaled model (1:10) of the siphon spillway was constructed and studied. Various options to improve the air regulation were tested using the physical model. The most stable stage-discharge conditions are found to be provided by a slot being cut into the spillway hood. This provided the best air-regulated stability, and full spillway capacity.

The authors acknowledge the contribution of Bill Davidson of the Babtie group and various members of British Waterways to the final design.

Valentine E. M (1994) "Brent reservoir siphon spillway hydraulic model study" Final Report prepared for Babtie Shaw & Morton Consulting Engineers, and British Waterways, Civil Eng. Dept., University of Newcastle upon Tyne, UK.

Chadwick A., and Morfett J. (1993) "Hydraulics in Civil and Environmental Engineering" published by E & FN SPON, an imprint of Chapman & Hall, London, UK.

Fig. 2 A general view of the experimental set-up in present physical model study.