PAPER ON A PROPOSED INTAKE CONFIGURATION FOR BULK WATER PUMP

ABSTRACTION FROM HEAVY SILT LADEN RIVER FLOW

 

MYBURGH, N.J.

 

Chief Engineer, Hydraulic Studies, Department of Water Affairs and Forestry, Pretoria, South Africa

Private Bag x313, Pretoria, South Africa, 0001

Tel: 2712-3388483; Fax: 2712-3261780; E-mail: dca@dwaf.pwv.gov.za

 

 

Abstract

The details of a new proposed design for intake works where bulk water is abstracted by pumps from rivers with large sediment loads, is described. The key results of a study tour, a literature survey and two scale models employed to develop the proposed layout, are presented.

 

Keywords: sediment; intake; pumping station; forebay; vortices; skimming weir; funnel; off-take; scale model; flow straightener; abstaction

 

INTRODUCTION

Recent failures of major bulk water, pump station river intakes, as a result of severe sedimentation, have prompted the Department of Water Affairs and Forestry to search for an intake configuration that would be relatively free from sedimentation problems.

The modus operandi that was adopted to find a reliable solution to this problem consisted of the following:

·        a study tour to a variety of intake works in the country in an attempt to learn from past failures and successes;

·        a literature survey with a view to establish sound engineering principles/techniques/practices on which to base a new design;

·        testing of the new design by means of two hydraulic models and finally

·        to build a prototype to verify the findings of these studies in practice.

 

At this stage the first three phases have been completed and a brief description thereof is presented in this paper.

 

MAIN RESULTS OF STUDY TOUR AND LITERATURE SURVEY

The study tour revealed that perhaps the most outstanding problem area is the fact that there exists a serious conflict between established design standards for pumping stations' forebays and the intrinsic behaviour of transported sediment, in particular the settling mechanisms of suspended sediment, i.e. most design manuals require uniform, tranquil, irrotational approach flow conditions directly in front of the pumps. This may be fine for clear water pumping but at river intakes with silt laden flow the mentioned criteria provide ideal conditions for the deposition of fine suspended sediment in pump stations' forebays.

 

The literature survey's contribution was all the engineering principles/practices as described hereunder under 'Proposed New Design'. These principles were combined in a rather unique manner to produce the proposed layout.

 

OBJECTIVES OF THE STUDY

The main objective was to find a configuration that would exclude virtually all bed load and to minimise the deposition of suspended sediment in the pumping forebay, whilst having approach flow conditions inside the forebay that would not have any detrimental effects, e.g. cavitation, on the pumps. The works had to function properly under heavy silt laden conditions.

 

PROPOSED NEW DESIGN

Figure 13, p 25 depicts the proposed design for the Middledrift intake works of the Mhlatuze River Augmentation Scheme that was used as a trail case for purposes of this study. The design capacity of the works was 8,1 m³/s which had to be delivered by six pumps with equal capacity plus two standby pumps. The proposed design consists of the following components:

·        an oblique head wall on the up stream side, placed at an angle of 45 degrees with the direction of flow. The idea of this wall was to create a scouring mode directly in front of the main intake gates(which are contained in the wall)and to induce a scour hole through diving and swirling flow similar to what occurs at a bridge pier;

·        as there was no bend available on the river where the abstraction works would be located, an artificial bend was introduced. Abstraction would then take place on the outside of the bend as is the established practice.

 

The inside of the bend was arranged such that it converged towards the down stream end to form a funnel channel. The principle being employed here was the induction of an acceleration of the flow to velocities high enough to sustain self-cleaning conditions in front of the skimming weir. On the up stream side of the funnel the mean flow velocity should be smaller than that of the body of water passing the main intake gates up.stream of the oblique wall, so that sediment particles would rather opt to pass the gates. However, velocities everywhere in the funnel should be high enough to keep the sediment entrained in the water body that would exit the works and go back to the river;

·        a slot was introduced at the top of the outside wall of the funnel channel to form a skimming weir^[1], allowing only the clearer part of the water to be drawn off over the top of the weir;

·        three main inlet gates were provided in the oblique head wall to have, apart from its main function, the facility to shut off the entire flow when sediment loads outside the works would become so heavy during high river flows, that it would no longer be possible to control the deposition thereof inside the works;

·        a control tail gate was provided at the downstream end of the funnel channel;

·        eight pump suction inlets were located on the bottom of a compact pumping forebay and provided with standard bellmouths. These were linked individually to suction pipes with a minimum length of 50 pipe diameters which would act to straighten the flow lines in order to have the desired aforementioned approach flow conditions. The principle here is the minimum length^[2] that a fluid has to travel in a pipe before fully rough turbulent, uniform flow conditions occur.

·        the intake structure formed the interface with a weir with a stepped crest across the river. The crest of the weir was stepped down towards the intake to ensure that low river flows would always pass through the intake. The weir was necessary to stabilise the abstraction water level in the river during low river flows in this particular case. If the weir can be avoided it should be, because its very existence usually leads to sediment deposition in the vicinity of the intake, thereby only aggravating the situation.

 

FLOOD HYDROLOGY

Expected floods at Middledrift in the Tugela river range from about 2 050 m³/s for the 2 years flood to 19 500 m³/s for the Regional Maximum Flood ^[3]. As abstraction conditions become totally uncontrollable under high flood conditions, it was decided to model conditions only up to the 20 years flood of about 5 200 m3/s.

 

PREVAILING SEDIMENT CONDITIONS

The only available records for high river flows for the site concerned indicated very fine entrained sediment( clay and silt with a D₅₀-value of about 0.005 mm). The highest measured concentration was 20 200 mg/l at a flow rate of about 2 600 m³/s. The average sediment yield of the catchment was about 450 t/km²/a.

 

EXPERIMENTAL SET-UP OF SCALE MODELS AND TESTING PROTOCOL

Two hydraulic models were employed to study the performance of the proposed design, the one had an undistorted scale of 1:65 and the other an undistorted scale of 1:25(distorted scale models would render velocities so high that sediment deposition would not be possible for the very high flow conditions that had to be modelled). The smaller scale model was used to study the general hydraulic conditions up- and down stream of, as well as inside the intake structure, whilst the large scale model was mainly used to focus on the intake structure itself. Model boundaries were chosen according to standard norms.

Both models had a movable bed with the basin behind the weir entirely filled with the movable material. The latter consisted of a sandy silt with a D50-value of about 0.05 mm and a dry density of 1360 kg/m³. Standard stilling pools were installed up stream and adjustable tailgates down stream of the models and flows were measured by accurate electro-magnetic flow meters. The perspex pipes that were used to simulate the pumps' suction were calibrated to deliver the correct flows which were measured by means of a calibrated v-notch. Model materials consisted of perspex and timber.

For each scenario sediment was introduced at average prototype concentrations at the upstream boundary of the models on a continuous basis until an equilibrium was reached. The modelling agent was the same sandy silt that was used for the movable bed.

 

DISCUSSION OF RESULTS

From numerous tests over the entire flow range it was evident that the oblique wall was very successful in inducing a scouring mode in front of the main gates. It was also clear from the particular formation and direction of ripples on the models' beds, which are indicative of the direction of bottom currents, that the bed load was mainly directed away from the intake.

 

The artificial bend and funnel channel combination performed particularly well. Discharges through the funnel had to be controlled by the funnel's control gate such that a water level was sustained in the pumping forebay, high enough to rule out the formation of vortices that could suck undesirable air into the pumps. Under these conditions velocities high enough to keep the sediment entrained were recorded for all test scenarios. For the average daily flow of 50 m³/s, velocities of 0,65 m/s and 1,45 m/s(prototype values) were recorded at the entrance and exit of the funnel, respectively. If the modified Liu-diagram] ^[7] is employed to look at the self-cleaning abilities of the funnel from a more theoretical point of view, the following is obtained:

 

The entrance velocity of 0,65 m/s yields a shear velocity(V*) = 0,031 m/s and if a prototype sediment particle size of 0,15 mm(the biggest recorded) is assumed , a fall velocity(Vss)of 17,6 mm/s is obtained. Then the ratio V*/Vss = 1,76. Calculation of the Particle Reynolds Number yields a value = 4,04. When these values are plotted, sediment movement is clearly indicated.

 

Scour on the outside of the bend was never really dramatically demonstrated during the tests, because there never occurred significant amounts of deposited sediment on the funnel's floor, because of the excellent self-cleaning abilities thereof.

 

The effectiveness of the skimming weir was not tested as it is based on a sound, established hydraulic principle.

 

The proposed layout of the pumping forebay is the culmination of numerous tests on a variety of layouts. The main finding of these tests was that the forebay should be kept as compact as possible to enhance velocities and thereby minimising the deposition of suspended sediment(a minimum velocity of 0.65 m/s is recommended). Also much less space for sediment deposition will then be available. The final layout was mainly determined by the length of the skimming weir and the diameter of the bellmouths of the suction pipes.

 

The control tail gate of the funnel channel has a vital role to play, as it is not only employed to control scour in the funnel, but it also serves to ensure a water level in the forebay that will rule out the formation of undesirable vortices. As safety grids are normally a requirement in front of suction pipes, it could be used beneficially to lower the required water level in the forebay, which means that less water has to enter the main inlet gates and therefore more water is available to enhance the scouring effect in front of the main inlet gates. The grids also had a remarkable stilling influence on the turbulence in the forebay. The abovemention measures will rule out most, if not all, of the bedload, but settlement of suspended sediment in the forebay is inevitable during times when pumping will not take place. A possible solution to this problem is a dedicated pump to circulate the stationary water during these times and thereby keeping sediment in suspension. Alternatively desilting equipment for the forebay could be provided. Furthermore the main inlet gates should be closed during times of high river flows to prevent excessive amounts of suspended sediment from entering the forebay.

 

Pump suction pipes with a minimum length of 50 pipe diameters are essential to straighten the flow lines in order to have uniform, irrotational approach flow conditions in front of the pumps' suction ends These conditions were clearly demonstrated during the tests by using standard dye and flow indicators. To rule out the settlement of fine suspended sediment in these pipes, minimum velocities of 0,7 m/s are recommended.

 

CONCLUSION

The study concluded that a practical compromise between conflicting natural hydraulic conditions and man-made requirements is indeed possible. Lastly, it is intended to test the hydraulic performance of the proposed layout also by using a 3D numerical model^[6] that became available recently.

 

ACKNOWLEDGEMENTS

The author wishes to express his gratitude to the Management of the Department of Water Affairs and Forestry for their permission to publish and present this paper.

 

REFERENCES

1.      Avery, P (1981): " Sediment exclusion at intakes - a review ", British Hydromechanics Research Association, Project RP 13327.

 

2.      Doherty, R.L.(1989-edition): " Fluid Mechanics with Engineering Applications " , p 217.

 

3.      WLPU-report 1596/10(1992): "Mhlatuze Basin Augmentation-Feasibility Study ", p 7.5.

 

4.      Prosser, M.J. (1977): "The Hydraulic Design of Pump Sumps and Intakes".

 

5.      Snell, E.F.A. (1994): "The Design of River Intakes to Minimise Abstraction of Sediment" , 50 Years of Water Engineering in South Africa, Wits University, South Africa.

 

6.      Atkinson, E.(1995): " A Numerical Model for predicting Sediment Exclusion at Intakes", HR Wallingford Report OD 130.

 

7.      Rooseboom, A (1992): " Sediment Transport in Rivers and Reservoirs - a Southern African Perspective ", Figure 3.7, p 3.20, WRC Report No. 297/1/92.