Physcial hydraulic model investigation of critical submergence for raised pump intakes

(1)

Thesis presented in partial fulfilment of the requirements for the degree of Master of Engineering (Civil) at Stellenbosch University

PHYSICAL HYDRAULIC

MODEL INVESTIGATION OF

CRITICAL SUBMERGENCE

FOR RAISED PUMP INTAKES

by

SH KLEYNHANS

FEBRUARY 2012

DEPARTMENT OF CIVIL ENGINEERING

STUDY LEADER: Prof. GR BASSON

(2)

DECLARATION

Herewith I declare that I know the meaning of plagiarism and that all the text, calculations, results, drawings and graphs in this thesis are primarily my own work. All other work has been cited as appropriate in accordance with the prescribed referencing method.

Name:

St. No.:

Signed:

(3)

SINOPSIS

Oor die afgelope vier dekades is verskeie ontwerpriglyne vir die berekening van minimum watervlakke, om werwelvorming by pompinlate te voorkom, gepubliseer. Hierdie ontwerpriglyne vereis dat die klokmond van die pompinlaat nie hoër as 0.5 keer die deursnee van die klokmond (D) bokant die kanaalvloer geleë moet wees nie.

Sandvang kanale vorm ‘n integrale deel van groot riveronttrekkingswerke, met pompinlate wat aan die einde van hierdie kanale geleë is. Die kanale word aan die stroomaf kant van die pompinlaat voorsien met sluise sodat die kanale gespoel kan word. Hierdie sluise is tipies 1.5 m hoog. Dit is derhalwe nodig om die hoogte onder die klokmond dieselfde te maak as die hoogte van die sluis sodat die klokmond die spoelwerking nie beïnvloed nie. Die vraag is egter – wat is die impak op die minimum vereiste watervlakke indien die klokmond op ‘n hoër vlak installeer word?

‘n Fisiese hidrouliese model met ‘n 1:10 skaal is gebruik om die minimum watervlakke te bepaal waar tipes 2, 5 en 6 werwels aangetref word vir prototipe inlaatsnelhede van 0.9 m/s tot 2.4 m/s en klokmond hoogtes van 0.5D, 1.0D en 1.5D bokant die kanaalvloer. Vier klokmond konfigurasies is getoets. Die minimum vereiste watervlakke was die laagste vir die tradisionele plat klokmond met ‘n lang radius buigstuk en was dus die voorkeur klokmond.

Die eksperimenttoetsresultate vir die voorkeur klokmond is met die gepubliseerde ontwerpriglyne vergelyk om te bepaal watter van die ontwerpsriglyne van toepassing sal wees vir verhoogde klokmond installasies.

Uit die eksperimenttoetsresultate is dit duidelik dat die vereiste watervlakke skielik verhoog sodra die klokmond installasie ‘n seker hoogte bokant die kanaal vloer oorskry. Daar is bevind dat hierdie verskynsel by al vier klokmond konfigurasies voorkom sodra die verhouding tussen die prototipe klokmond inlaatsnelheid teenoor die snelheid in die kanaal hoër as 6.0 is.

Daar word aanbeveel dat die minimum vereiste watervlak vir pompinlate met dieselfde geometrie as die fisiese model, met Knauss (1987) se vergelyking bereken word, naamlik S = D(0.5 + 2.0Fr), waar die snelheidsverhouding tussen die klokmond en kanaal 6.0 nie oorskry nie, en dat die vergelyking gepubliseer deur die Hydraulic Institute (1998), S = D(1 + 2.3Fr), gebruik word waar die snelheidsverhouding 6.0, so bereken met Knauss (1987) ser vergelyking, wel oorskry. Die prototipe klokmond inlaatsnelheid moet ook beperk word tot 1.5 m/s.

(4)

SYNOPSIS

Various design guidelines have been published over the past four decades to calculate the minimum submergence required at pump intakes to prevent vortex formation. These design guidelines also require the suction bell to be located not higher than 0.5 times the suction bell diameter (D) above the floor.

Sand trap canals are an integral part of large river abstraction works, with the pump intakes located at the end of the sand trap canals. The canals need to be flushed by opening a gate, typically 1.5 m high, that is located downstream of the pump intake. This requires the suction bell be raised to not interfere with the flushing operation, which leads to the question – what impact does the raising of the suction bell have on the minimum required submergence?

A physical hydraulic model constructed at 1:10 scale was used to determine the submergence required to prevent types 2, 5 and 6 vortices for prototype suction bell inlet velocities ranging from 0.9 m/s to 2.4 m/s, and for suction bells located at 0.5D, 1.0D and 1.5D above the floor. The tests were undertaken for four suction bell configurations with a conventional flat bottom suction bell, fitted with a long radius bend, being the preferred suction bell configuration in terms of the lowest required submergence levels.

The experimental test results of the preferred suction bell configuration were compared against the published design guidelines to determine which published formula best represents the experimental test results for raised pump intakes.

It became evident from the experimental test results that the required submergence increased markedly when the suction bell was raised higher than a certain level above the floor. It was concluded that this “discontinuity” in the required submergence occurred for all the suction bell configuration types when the ratio between the prototype bell inlet velocity and the approach canal velocity was approximately 6.0 or higher.

It is recommended that, for pump intakes with a similar geometry to that tested with the physical hydraulic model, critical submergence is calculated using the equation published by Knauss (1987), i.e. S = D(0.5 + 2.0Fr), if the prototype bell inlet velocity/approach canal velocity ratio is less than 6.0, and that the equation published by the Hydraulic Institute (1998), i.e. S = D(1 + 2.3Fr), can be used where the ratio, as determined with Knauss’ (1987) equation, exceeds 6.0. It is also recommended that prototype bell inlet velocities be limited to 1.5 m/s.

(5)

ACKNOWLEDGEMENTS

The author would like to acknowledge the contributions of the following individuals and funding institution:

 Prof GR Basson, for the support and guidance he provided as study leader;

 The laboratory personnel of Stellenbosch University, especially Christiaan Visser, for arranging the construction of the physical model and assisting with modifications during the test phase;

 TCTA, for funding the construction of the physical hydraulic model;

 Charlie Espost, for providing the pumpset and variable speed drive;

 My colleague, Schalk van der Merwe, for encouraging me to enrol for a Master’s degree and for his continuous support at work while I completed my studies on a part-time basis; and

 My wife, Maré – this thesis would not have been possible without your support and encouragement over the past two years.

(6)

LIST OF FIGURES

Figure 1.1: Dimension variables for pump intake (Hydraulic Institute, 1998) ... 2

Figure 1.2: Proposed layout of Lower Thukela abstraction works ... 3

Figure 1.3: Physical model of the proposed Lower Thukela abstraction works ... 4

Figure 1.4: Proposed raised suction bell installation in sand trap canal ... 5

Figure 2.1: Flat bell installation for flows ≤ 1 m3_{/s (Sulzer Brothers Limited, 1987) ...10}

Figure 2.2: Slanted bell installation for flows > 1 m3/s (Sulzer Brothers Limited, 1987) ...10

Figure 3.1: Rankine combined vortex (Prosser, 1977) ...14

Figure 3.2: Free surface and sub-surface vortex strength classification (Hydraulic Institute, 1998) ...16

Figure 3.3: Critical submergence versus bell inlet velocity ...34

Figure 3.4: Approach velocities versus bell inlet velocity ...34

Figure 4.1: Plan view of physical model ...36

Figure 4.2: Sectional view of physical model ...36

Figure 4.3: View of pump model ...38

Figure 4.4: View of stilling basin and flow straightener pipes ...38

Figure 4.5: Dimensions of suction bell configurations ...39

Figure 4.6: Rear view of suction bell configurations (from left to right: Type 1B, Type 1A, Type 2B, Type 2A) ...40

Figure 4.7: Side view of suction bell configurations (from left to right: Type 2A, Type 2B, Type 1A, Type 1B) ...40

Figure 4.8: ADV support frame ...42

Figure 4.9: Vectrino instrument ...42

Figure 4.10: Flow diagram of daily test procedure ...44

Figure 4.11: Path of dye injected upstream of suction bell ...47

Figure 4.12: Evidence of a coherent dye core behind the suction bell ...48

Figure 4.13: Positions of ADV measurements at inlet bell ...49

Figure 4.14: XYZ coordinates of ADV instrument (Nortek, 2004) ...50

Figure 4.15: Position of ADV measurements in sand trap canal ...51

(10)

Figure 5.2: Type 1A bellmouth, 1.0D above floor – prototype type submergence ...54

Figure 5.4: Type 1B bellmouth, 0.5D above floor – prototype type submergence ...56

Figure 5.13: Type 1B bellmouth, 0.5D above floor – prototype type submergence for Type 2 vortices ...65

Figure 5.19: Type 2 vortex (surface dimple) ...72

Figure 5.20: Type 5 vortex (pulling air bubbles to intake) ...72

Figure 5.21: Type 6 vortex (full air core to intake) ...73

Figure 5.22: Break-away caused when water level is at joint on segmented bend ...74

Figure 5.23: Type 1A suction bell – submergence required for Type 2 vortices at bell heights of 0.5D, 1.0D and 1.5D ...75

(11)

Figure 5.26: Type 1B suction bell – submergence required for Type 2 vortices at bell heights of 0.5D, 1.0D and 1.5D ...77 Figure 5.27: Type 1B suction bell – submergence required for Type 5 vortices at bell heights of 0.5D, 1.0D and 1.5D ...78 Figure 5.28: Type 1B suction bell – submergence required for Type 6 vortices at bell heights of 0.5D, 1.0D and 1.5D ...78 Figure 5.29: Type 2A suction bell – submergence required for Type 2 vortices at bell heights of 0.5D, 1.0D and 1.5D ...79 Figure 5.30: Type 2A suction bell – submergence required for Type 5 vortices at bell heights of 0.5D, 1.0D and 1.5D ...80 Figure 5.31: Type 2A suction bell – submergence required for Type 6 vortices at bell heights of 0.5D, 1.0D and 1.5D ...80 Figure 5.32: Type 2B suction bell – submergence required for Type 2 vortices at bell heights of 0.5D, 1.0D and 1.5D ...81 Figure 5.33: Type 2B suction bell – submergence required for Type 5 vortices at bell heights of 0.5D, 1.0D and 1.5D ...82 Figure 5.34: Type 2B suction bell – submergence required for Type 6 vortices at bell heights of 0.5D, 1.0D and 1.5D ...82 Figure 5.35: Type 1A suction bell – submergence against bell/approach velocity ratios for Type 2 vortices ...86 Figure 5.36: Type 1A suction bell – submergence against bell/approach velocity ratios for Type 5 vortices ...86 Figure 5.37: Type 1A suction bell – submergence against bell/approach velocity ratios for Type 6 vortices ...87 Figure 5.38: Type 1B suction bell – submergence against bell/approach velocity ratios for Type 2 vortices ...88 Figure 5.39: Type 1B suction bell – submergence against bell/approach velocity ratios for Type 5 vortices ...88 Figure 5.40: Type 1B suction bell – submergence against bell/approach velocity ratios for Type 6 vortices ...89 Figure 5.41: Type 2A suction bell – submergence against bell/approach velocity ratios for Type 2 vortices ...90 Figure 5.42: Type 2A suction bell – submergence against bell/approach velocity ratios for Type 5 vortices ...90 Figure 5.43: Type 2A suction bell – submergence against bell/approach velocity ratios for Type 6 vortices ...91

(12)

Figure 5.44: Type 2B suction bell – submergence against bell/approach velocity ratios for Type

2 vortices ...92

Figure 5.45: Type 2B suction bell – submergence against bell/approach velocity ratios for Type 5 vortices ...92

Figure 5.46: Type 2B suction bell – submergence against bell/approach velocity ratios for Type 6 vortices ...93

Figure 5.47: Comparison of submergence between Type 1A and 1B suction bell configurations for Type 2 vortices ...94

Figure 5.53: Comparison of submergence between Type 1B and 2B suction bell configurations for Type 2 vortices ...98

Figure 5.56: X, Y and Z velocities for 0.9 m/s bell inlet velocity at Position 2 ... 102

Figure 5.57: Plan view of resultant velocities at suction bell for prototype bell inlet velocities of 0.9 m/s to 2.4 m/s ... 106

Figure 5.58: Position 1C: X, Y and Z velocities for 0.5D and 1.2 m/s inlet bell velocity ... 111

Figure 5.59: Position 1C: X, Y and Z velocities for 0.5D and 1.2 m/s inlet bell velocity for sample numbers 195 to 295 ... 112

Figure 5.60: Average resultant velocities along centrelines 1-4-7, 2-5-8, and 3-6-9 ... 113

Figure 5.61: Average resultant velocities along Planes A, B and C ... 114

Figure 5.62: Path of particle approaching the inlet bell ... 115

Figure 6.1: Comparison of submergence between design guidelines and measured Type 2, 5 and 6 vortices for a bell height of 0.5D ... 116

(13)

Figure 6.2: Comparison of submergence between design guidelines and measured Type 2, 5 and 6 vortices for a bell height of 1.0D ... 117 Figure 6.3: Comparison of submergence between design guidelines and measured Type 2, 5 and 6 vortices for a bell height of 1.5D ... 118 Figure 6.4: Comparison of submergence between design guidelines and measured Type 2 vortices for bell heights of 0.5D, 1.0D and 1.5D ... 119

(14)

LIST OF TABLES

Table 1.1: Recommended dimensions for pump intakes ... 2

Table 1.2: Recommended dimensions versus proposed dimensions for Vlieëpoort and Lower Thukela sand trap canals ... 6

Table 2.1: Modelling scenarios ...11

Table 2.2: Repeat tests performed to verify reliability of results ...12

Table 2.3: ADV recording scenarios ...12

Table 3.1: Prototype information and scale effects ...29

Table 3.2: Critical submergence recommended by Gorman-Rupp ...32

Table 3.3: Critical submergence recommended by KSB ...32

Table 3.4: Critical submergence recommended by Metcalf and Eddy (1981) ...33

Table 4.1: Configuration set-up of ADV instrument ...41

Table 5.1: Test results – inlet bell Type 1A, 0.5D above floor ...52

Table 5.4: Test results – inlet bell Type 1B, 0.5D above floor ...55

Table 5.13: Repeatability test results – inlet bell Type 1B, 0.5D above floor, Type 2 vortices ...65

(15)

Table 5.15: Repeatability test results – inlet bell Type 1B, 0.5D above floor, Type 6 vortices ...67 Table 5.16: Repeatability test results – inlet bell Type 2B, 0.5D above floor, Type 2 vortices ...68 Table 5.17: Repeatability test results – inlet bell Type 2B, 0.5D above floor, Type 5 vortices ...69 Table 5.18: Repeatability test results – inlet bell Type 2B, 0.5D above floor, Type 6 vortices ...70 Table 5.19: Reynolds Numbers for approach velocities ...84 Table 5.20: Water levels for ADV tests at suction bell ... 100 Table 5.21: Summary of ADV measurements (model) at suction bell for prototype 0.9 m/s bell inlet velocity ... 102 Table 5.22: Summary of ADV measurements (model) at suction bell for prototype 1.2 m/s bell inlet velocity ... 103 Table 5.23: Summary of ADV measurements (model) at suction bell for prototype 1.5 m/s bell inlet velocity ... 104 Table 5.24: Summary of ADV measurements (model) at suction bell for prototype 1.8 m/s bell inlet velocity ... 104 Table 5.25: Summary of ADV measurements (model) at suction bell for prototype 2.1 m/s bell inlet velocity ... 105 Table 5.26: Summary of ADV measurements (model) at suction bell for prototype 2.4 m/s bell inlet velocity ... 105 Table 5.27: Water levels for ADV tests along approach canal ... 107 Table 5.28: Summary of ADV measurements (model) along approach canal ... 108 Table 6.1: Bell inlet velocity/approach canal velocity ratios for Type 1B suction bell located 0.6 m (prototype) above canal floor ... 121 Table 6.2: Bell inlet velocity/approach canal velocity ratios for Type 1B suction bell located 1.2 m (prototype) above canal floor ... 122 Table 6.3: Bell inlet velocity/approach canal velocity ratios for Type 1B suction bell located 1.8 m (prototype) above canal floor ... 124

(16)

LIST OF ABBREVIATIONS

a Acceleration

A Cross-sectional area of inlet bell ADV Acoustic Doppler velocimeter AWWA American Water Works Association

B Distance from back wall to pump inlet centreline C Distance from floor to underside of the inlet bell

Cd Discharge coefficient

CFD Computational fluid dynamics

d Suction pipe diameter

D Inlet bell diameter

E Euler Number

F Force

Fg Force due to gravity

Fp Force due to pressure

Fr Froude Number at suction bell

Fst Force due to surface tension

Fv Force due to viscosity

g Gravitational acceleration

HDPE High density polyethylene

Hz Hertz

(17)

L Length (linear dimension)

Lr Geometric scale

ℓ/s Litres per second

m Metre

MCC Mokolo Crocodile Consultants

MHz Megahertz

mm Millimetre

m/s Metres per second

m3/s Cubic metres per second

p Prototype

P Pressure

PVC-U Un-plasticised polyvinyl chloride

Q Flow rate

R Radius

Re Reynolds Number

rpm Revolutions per minute

S Submergence

Sc Critical submergence

T Time

TCTA Trans Caledon Tunnel Authority

V Velocity

(18)

Vt Tangential velocity Vx Velocity in x-direction Vy Velocity in y-direction Vz Velocity in z-direction

W Width of the inlet channel

We Weber Number

Scale factor

% Percentage

°C Degrees Celsius

σ Surface tension of liquid

ρ Mass density of liquid

ⱱ Dynamic viscosity of liquid

Γ Circulation

(19)

1. INTRODUCTION

1.1 Background to the research project

The design of pump intakes is often based on published design guidelines and empirical formulae that were derived from physical model or scaled prototype studies. In a few instances, designs are also based on existing pump intakes that are operating satisfactorily, provided that similar site conditions exist.

These design guidelines and empirical formulas have specifically been developed to deal with hydraulic phenomena that could cause problems at pump intakes, for example (Hydraulic Institute, 1998):

 Submerged vortices;

 Free-surface vortices;

 Uneven velocity distribution in approach channels;

 Excessive pre-swirl of flow entering the pump;

 Non-uniform spatial distribution of velocity at the impeller eye; and

 Entrained air of gas bubbles.

The presence, duration and magnitude of the above phenomena at pump intakes could impact negatively the operation of the pumping system and result in a reduction in pump performance (i.e. both flow rate and pressure), excessive vibrations, increased noise and higher power consumption.

Figure 1.1 shows a typical single pump intake with the dimension variables to design the pump

intake. Table 1.1 provides a description of each of the dimension variables and the recommended dimensions based on the guidelines published in the American National

(20)

Figure 1.1: Dimension variables for pump intake (Hydraulic Institute, 1998) Table 1.1: Recommended dimensions for pump intakes

Dimension variable Description of dimension Recommended

dimension

B Distance from the back wall to the pump inlet centreline

B = 0.75D (1)

C Distance between the inlet bell and the floor C = 0.3D to 0.5D

D Inlet bell diameter See Note 2

Fr Froude number at suction bell -

H Minimum liquid depth H = S + C

S Minimum submergence at pump inlet bell S = 1D + 2.3Fr

W Entrance width of pump inlet bay W = 2D

X Length of pump inlet bay X = 5D minimum

(1) This recommended dimension is not achievable for suction bells fitted with bends inside the wet well. (2) The inlet bell diameter can be determined from the following recommended velocity ranges:

a. Inlet bell velocities could range from 0.6 m/s to 2.7 m/s for flows less than 315 ℓ/s per pump;

b. Inlet bell velocities could range from 0.9 m/s to 2.4 m/s for flows exceeding 315 ℓ/s but less than 1 260 ℓ/s per pump; and

(21)

The “minimum submergence at the pump inlet bell” is also referred to in the literature as “critical submergence”, which is defined as “submergence at which, after the flow reached steady conditions of depth and discharge, air from free-surface is ingested either continuously or intermittently through the agency of the vortex” (Denney, 1956; cited in Rajendran & Patel, 2000).

It should be noted that, in Table 1.1, all pump intake dimensions are expressed as a function of the inlet bell diameter, D. This ensures geometric similarity of the hydraulic boundaries and dynamic similarity of the flow patterns.

The question that arises is – what will the impact be on the minimum submergence when any of the recommended pump intake dimensions listed in Table 1.1 are changed?

1.2 Motivation for the study

Figure 1.2 shows the main components associated with the abstraction works proposed for the

Lower Thukela Bulk Water Supply Scheme, i.e. a weir, a boulder trap, a gravel trap, sand trap canals and a fishway (Basson, 2011). A trash rack is provided at the inlet to the sand traps, with the raw water pumps located at the end of the sand trap. Figure 1.3 shows the physical model constructed for the abstraction works of the Lower Thukela Bulk Water Supply Scheme.

Figure 1.2: Proposed layout of Lower Thukela abstraction works

Gravel trap Submerged slot Boulder trap Crump weir: low notch Fishway Sand trap

(22)

Figure 1.3: Physical model of the proposed Lower Thukela abstraction works

A similar abstraction works arrangement, with six sand traps, is proposed for the river abstraction works located at Vlieëpoort on the Crocodile River (Basson, 2010).

The sand trap canals are designed with velocities of less than 0.2 m/s at the minimum operating level to trap sediment larger than 0.3 mm in diameter. The sand trap canals also have bed slopes of 1:80 to allow flushing of sand and gravel under gravity by opening the downstream gate (see Figure 1.3). Furthermore, in order to effectively flush the sand trap, the raised downstream gate opening should be at least 1.5 m (Basson, 2010; Basson, 2011).

Dry well pump installations are proposed for both the Lower Thukela and Vlieëpoort abstraction works. This would require the installation of a suction bell inside the sand trap canal (see

Figure 1.3). The suction bell, however, needs to be raised to not interfere with the flushing

operation and to ensure an effective gate opening of 1.5 m, as shown in Figure 1.4. Upstream gate

Downstream gate

Suction bell Sand trap canal

(23)

Figure 1.4: Proposed raised suction bell installation in sand trap canal

Table 1.2 summarises the dimensions proposed for the Lower Thukela and Vlieëpoort sand trap

canals in comparison with the dimensions recommended in the American National Standard for

Pump Intake Design (Hydraulic Institute, 1998).

It is evident from Table 1.2 that, for the Vlieëpoort and Lower Thukela sand trap canals, dimension variables “B” and “C” are not in accordance with the dimensions recommended in the

American National Standard for Pump Intake Design (Hydraulic Institute, 1998). This leads to

the question – what impact does the raising of the suction bell have on the minimum submergence required to prevent air entrainment? This question is the motivation for this study.

Raising the suction bell to ensure the effective flushing of the sand traps would, however, be a deviation from the published pump design guidelines.

(24)

Table 1.2: Recommended dimensions versus proposed dimensions for Vlieëpoort and Lower Thukela sand trap canals

Dimension variable

Dimension description

Vlieëpoort sand trap canal (2 m3/s per pump)

Lower Thukela sand trap canal (0.467 m3/s per pump) Recommended dimension (1) (mm) Proposed dimension (2) (mm) Recommended dimension (1) (mm) Proposed dimension (mm) D Inlet bell diameter 1 500 1 200 700 700 W Entrance width of pump inlet bay 3 000 2 400 1400 2 000 B Distance from the back wall to

the pump inlet centreline

1 125 2 600 525 1 500

C Distance

between the inlet bell and the

floor 750 1 500 350 1 500 (3) S Minimum pump inlet bell submergence 2 200 2 180 1 450 1 200

(1) Recommended dimensions refer to the dimensions recommended by the Hydraulic Institute (1998).

(2) The sand trap canals were initially designed for a flow of 2.5 m3/s per pump, an inlet bell diameter of 1.4 m and a pump inlet bay entrance width of 3.0 m. The dimensions were reduced as the design flow reduced to 2 m3/s per pump.

(3) The initial consideration was to install the suction bell at a height of 350 mm (as reported by Basson, 2011) and to use a guiderail system to raise the suction bell during times of flushing. This system has the risk that air could be entrained on the suction pipework should the guiderail system malfunction. It was later decided to rather install the suction bell at a fixed height of 1 500 mm above the floor.

(25)

1.3 Objective of study

The objective of the study was to determine, by means of a physical hydraulic model, the minimum submergence levels required to prevent air entrainment for suction bell inlets located at different heights above the canal floor. The measured submergence levels were compared against the design guidelines that are available to calculate minimum submergence, after which recommendations were formulated for the design criteria to be applied for raised pump intake installations, which are similar in geometry to the physical model.

The following variables were investigated in the physical model study:

 Different bell inlet heights above the floor level;

 Different bell inlet velocities; and

 Different bell configurations.

The physical model study did not consider aspects such as the widening or narrowing of the sand trap canal, or velocity distributions at the impeller eye. The testing and modelling of changes in geometry, and the determination of velocity distributions at different locations along the pump intake, can be performed more effectively (both from a time and cost perspective) with three-dimensional computational fluid dynamics (CFD). Such a CFD model would first have to be calibrated against the results obtained in the physical model study. Therefore, a further objective of the physical model study was to obtain sufficient information on velocity distributions along the sand trap canal to assist with defining the boundary conditions, and the calibration, of the CFD model. The calibration and application of the CFD model forms part of another study (Hundley, 2012) and is not covered in this thesis.

1.4 Layout of the thesis

The report is structured as follows:

 The methodology to achieve the study objective is described in Section 2;

 Section 3 covers the literature review, which focuses on the theory and causes of vortex formation, the possible scaling effects associated with physical hydraulic models, and the available design standards to calculate the required submergence;

 The experimental set-up, instrumentation and test procedures are described in Section 4;

(26)

 Section 6 compares the available design standards with the results obtained from the physical model testing; and

(27)

2. METHODOLOGY

2.1 Physical model study

The Hydraulic Institute (1998) recommends that physical hydraulic model studies be performed when the sump and piping geometry deviate from published norms, as is the case with the suction bell configurations at the Vlieëpoort and Lower Thukela abstraction works. This recommendation is supported by Prosser (1977), who stated that the main factor influencing the need for model testing is whether the proposed scheme varies radically from existing satisfactory designs.

In light of the deviations from published pump intake design norms, the Trans Caledon Tunnel Authority (TCTA) instructed Mokolo Crocodile Consultants (MCC) to undertake a physical hydraulic model study, as well as a CFD model study, of the proposed Vlieëpoort sand trap canals with the raised suction bells, to determine the minimum submergence required to prevent air entrainment. MCC appointed Stellenbosch University to construct the physical model, but the TCTA decided to postpone the laboratory testing, and subsequent design of the Vlieëpoort abstraction scheme, due to uncertainties related to the water demands of end users, which could vary from 1 m3/s to 2.5 m3/s per pump. The TCTA did, however, agree that the physical model could be used by Stellenbosch University for other research projects, such as this one, which focussed primarily on the Vlieëpoort scheme.

2.2 Modelling scenarios

The minimum submergence levels required in the sand trap canals will be the maximum water level required to:

 Prevent air entrainment due to vortex formation; or

 Prevent priming problems, i.e. the minimum water level should be above the pump volute.

It should, however, be noted that, in practice, the height of the weir must not only be designed to ensure that the minimum submergence level in the sand trap canals is achieved, but also needs to be designed to ensure a sufficient hydraulic head across the abstraction works to flush the boulder trap, gravel trap and sand trap canals. The latter requirement was not dealt with in this study.

(28)

The following variables were considered in the design of the overall study methodology and determined the scenarios to be tested:

 The height of the suction bell above the floor level to be altered to test various heights;

 The flow rate to be varied from 1.0 m3/s to 2.5 m3/s per pump, or alternatively the bell inlet velocities to be varied from 0.9 m/s to 2.4 m/s;

 The preferred suction bell configuration to be determined, as Sulzer Brothers Limited (1987) recommend the use of a slanted bell for flows exceeding 1.0 m3/s (see Figures 2.1 and 2.2 for details of different bell configurations); and

 The radius of the suction pipe bend to be determined, as it determines the height of the pump volute. Sulzer Brothers Limited (1987) recommends that the suction pipe radius needs to be equal to or greater than 1.5 times the diameter of the suction bell.

Figure 2.1: Flat bell installation for flows ≤ 1 m3_{/s (Sulzer Brothers Limited, 1987)}

(29)

Table 2.1 summarises the modelling scenarios analysed, based on the variables listed above. Table 2.1: Modelling scenarios

Suction bell type

Radius of suction bend (1)

Bell inlet velocities (m/s)

Height above canal floor

Bell type reference (2)

Flat bell 1 x diameter of suction pipe 0.9; 1.2; 1.5; 1.8; 2.1; 2.4 0.5D; 1.0D; 1.5D 1A

Flat bell 2 x diameter of suction pipe 0.9; 1.2; 1.5; 1.8; 2.1; 2.4 0.5D; 1.0D; 1.5D 1B Slanted bell 1 x diameter of suction pipe 0.9; 1.2; 1.5; 1.8; 2.1; 2.4 0.5D; 1.0D; 1.5D 2A Slanted bell 2 x diameter of suction pipe 0.9; 1.2; 1.5; 1.8; 2.1; 2.4 0.5D; 1.0D; 1.5D 2B

(1) The radii of the bends were based on the suction pipe diameter and not the bell diameter as proposed by Sulzer Brothers Limited (1987), as the bends are manufactured in accordance with AWWA C208 where the radius is given as a function of the suction pipe diameter, e.g. 1d, 1.5d, 2.0d, etc., where “d” is the diameter of the suction pipe.

(2) The bell type reference shown in the last column is used in the report to refer to the different bell configurations.

The water level at which air was entrained was established in the physical model for each of the 72 scenarios listed in Table 2.1.

The formation of vortices is unsteady and unstable, i.e. vortices form intermittently at different locations near the pump intake and vary in strength over time. Therefore, the water levels at which the different types of vortices form, required subjective interpretation through observation of the predominant type of vortex present at a given water level. As these subjective interpretations might impact on the reliability of the test results, and in order to verify the reliability and repeatability of the tests, the tests for the two long radius (i.e. two times the diameter of the suction pipe) inlet bells, situated 0.5D above the floor, were repeated three (3) times for all six (6) bell inlet velocities. The decision to repeat the tests only for suction bells located 0.5D above the floor was based on the fact that almost all conventional pump intakes are located at this level. Table 2.2 summarises the additional tests that were performed to validate the reliability of the test results.

(30)

Table 2.2: Repeat tests performed to verify reliability of results Suction

bell type

Radius of suction bend Bell inlet velocities (m/s) Height above canal floor

Flat bell 2 x diameter of suction pipe 0.9; 1.2; 1.5; 1.8; 2.1; 2.4 0.5D Slanted bell 2 x diameter of suction pipe 0.9; 1.2; 1.5; 1.8; 2.1; 2.4 0.5D

As the objective of the study was to also take sufficient velocity readings that could be used for calibrating the CFD model, an acoustic Doppler velocimeter (ADV) was used to take three-dimensional velocity readings at different positions and heights along the sand trap canal. The ADV readings were only taken for the preferred inlet bell configuration at the velocities and heights indicated in Table 2.3. The decision on the bell inlet velocities, for which ADV readings were taken was based on the fact that the majority of pump intakes are designed for bell inlet velocities ranging from 1.0 m/s up to 1.5 m/s. ADV readings were also taken for a bell inlet velocity of 1.8 m/s for calibrating the CFD model for higher bell inlet velocities.

Table 2.3: ADV recording scenarios Suction bell type Radius of suction

bend

Bell inlet velocities (m/s)

Height above canal floor

Preferred option Preferred option 1.2 0.5D, 1.0D, 1.5D

Preferred option Preferred option 1.8 0.5D

2.3 Criteria to decide on minimum submergence in proto-type

The minimum submergence levels required to prevent air entrainment due to vortex formation, were determined in the physical model study for each of the four suction bell configurations (refer to Table 2.1). A comparison was made between the results of the four suction bell configurations to select the configuration with the lowest required submergence levels for all velocities and heights above the canal floor.

The results of the preferred suction bell configuration were then compared against published design guidelines that are generally used to calculate minimum submergence levels. The published formula that best represents the experimental test results for raised pump intakes was identified, after which recommendations were formulated on the design criteria to be applied for the prototype design of the Vlieëpoort and Lower Thukela sand trap canals and pump intake configurations.

(31)

3. LITERATURE REVIEW

The purpose of the literature review is to (a) review the causes of vortices, (b) investigate the likely scale impacts on the formation of vortices in a physical hydraulic model, and (c) obtain design guidelines for the calculation of critical submergence to avoid the entrainment of air. This section of the report will therefore address:

 The fundamentals and theory of vortex formation (Section 3.1);

 The impact of scale effects in physical models (Section 3.2); and

 The design guidelines available to calculate critical submergence (Section 3.3).

3.1 Fundamentals and theory of vortex formation

3.1.1 Introduction

Vortex formation at pump intake structures has been studied for more than 50 years. The guidelines for the “Hydraulic Design of Pump Sumps and Intakes” were developed by the British Hydromechanics Research Association (Prosser, 1977) and referenced research work performed in the 1950s, e.g.:

 “Studies of submergence requirements of high specific-speed pumps” (Iversen, 1953);

 “Hydraulic problems encountered in intake structures of vertical wet-pit pumps and methods leading to their solution” (Fraser and Harrison, 1953); and

 “The prevention of vortices and intakes” (Denny and Young, 1957).

The three fundamental causes of vortices were defined by Durgin and Hecker (1978; cited in Knauss, 1987) as (a) non-uniform approach flow to the pump intake due to geometric orientation, (b) the existence of high velocity gradients, and (c) obstructions near the pump intake.

It was further recommended by Prosser (1977) that flow approaching a pump intake that changes from free surface to a closed conduit should be:

 Uniform flow – the velocity, in magnitude and direction, of fluid particles should be the same at all points across the section considered;

 Steady flow – the velocity, in magnitude and direction, should not change with time; and

(32)

The equations used to describe the motion of vortex flow are complex, as velocities could vary spatially (i.e. in terms of the depth and width of the approach canal) and local geometric features could influence flow fields, which all impact on the calculation of the resulting circulation. A simplified analytical approach was proposed by Prosser (1977) to calculate circulation. Figure 3.1 shows a schematic representation of vortex flow that consists of a relatively small central portion of fluid that is rotating as a highly viscous solid body (also referred to as a “forced” vortex), combined with a non-viscous “free” vortex region extending radially outward from this central portion.

Figure 3.1: Rankine combined vortex (Knauss, 1987)

In the central area of the vortex, the fluid is assumed to rotate so that the tangential velocity, vt, varies linearly with the radius, r, so that

(33)

where

vt = tangential velocity (m/s)

ω = angular velocity in radians per unit time

r = radius (m).

In the free vortex area, the tangential velocity varies inversely with the radius and directly with the circulation. For a circular curve of radius r, the circulation, Γ (m2_{/s), is expressed as,}

Γ = 2π rvt (3.2)

Figure 3.1 also shows the variation in tangential velocities between the forced and free vortex

areas, i.e. r1 is the radius at which the transition from forced to free vortex occurs.

The velocity head associated with the circulation will reduce the local hydrostatic pressure, which will result in a localised lowering of the water surface. This drop in water surface will vary from a surface dimple to a full air core vortex based on the strength of the circulation. The classification of vortices is the subject of the next paragraph.

3.1.2 Vortex classification and strengths

Vortices are generally classified as (a) free surface vortices, starting from the free water surface, or (b) sub-surface vortices, starting from the floor, side or back wall of the intake structure. The free surface vortices, which might result in air being drawn in from the surface, could result in unbalanced forces on the pump impeller, and vibrations that have a negative impact on the pump performance.

The sub-surface vortices could introduce excessive swirls at the pump inlet, which could also result in unbalanced forces on the pump impeller eye.

The Alden Research Laboratory (Hydraulic Institute, 1998) developed a visual classification system to classify the strength of vortices. Figure 3.2 shows the strength classifications for free surface and sub-surface vortices.

The focus of this study will be on free surface vortices, as the objective is to determine the minimum water level required to prevent air entrainment.

(34)

Figure 3.2: Free surface and sub-surface vortex strength classification (Hydraulic Institute, 1998)

3.1.3 Acceptance criteria for pump intakes

The acceptance criteria for physical hydraulic model studies of pump intakes, as per the Hydraulic Institute (1998), are:

 Free surface vortices entering the pump must be less severe than Type 3, on condition that they occur less than 10% of the time or only for infrequent pump operating conditions;

 Sub-surface vortices entering the pump must be less severe than Type 2;

 The average swirl angle must be less than five (5) degrees. Swirl angles of up to seven (7) degrees will be accepted, but only if they occur less than 10% of the time; and

 The time-averaged velocities at points in the throat of the bell inlet must be within 10% of the average axial velocity in the throat of the bell inlet.

(35)

The above acceptance criteria do not explicitly address the air-by-water volumes that could be tolerated by pumps, if any. The following information pertaining to air-by-water volumes was obtained from various authors:

 3% free air showed a drop in hydraulic efficiency of 15% for centrifugal pumps (Prosser, 1977);

 3% air-by-water volume in a suction line can considerably reduce the head-discharge curves of centrifugal pumps (Padmanabhan & Hecker, 1984);

 3 to 4% free air may give rise to a small, but continuous, decrease in pump efficiency (Knauss, 1987);

 7 to 20% free air is required before pump operations are interrupted (Knauss, 1987);

 The formation of air entraining vortices upstream of the pump intake must be prevented (Sulzer Brothers Limited, 1987); and

 3 to 5% air in the suction pipe can lower the pump efficiency (Karassik, Messina, Cooper, & Heald, 2001).

Based on the above statements by Prosser and Sulzer Brothers Limited, as well as the Hydraulic Institute’s acceptance criteria to not permit free surface vortices exceeding Type 3 in strength, it is recommended that the critical submergence levels be determined in the physical model as the level at which no air entrainment will take place.

3.1.4 Dimensionless parameters

In order to apply the results of the physical model to the prototype, dimensionless parameters are used to define the complex interaction between the geometry of the intake structure, the flow velocity and the liquid properties which all influence the critical submergence. The following dimensionless parameters were proposed by different authors to describe the motion of vortex flow:

Jain, Raju, & Garde (1978) proposed the following functional relationship for critical submergence, which is defined as the water level required above the inlet of the bell to maintain the circulation within acceptable limits:

Scr = f(W, D, Q, Γ, g, ρ, σ, ν) (3.3)

(36)

W = width of the inlet channel or diameter of the vortex tank (m) D = diameter of the inlet bell (m)

Q = flow rate (m3/s)

Γ = circulation of flow (2πrVt) (m2/s) g = gravitational acceleration (m/s2) ρ = mass density of liquid (kg/m3₎ σ = surface tension of liquid (N/m) ⱱ = dynamic viscosity of liquid (m2_/s).

A typical single pump intake configuration is shown in Figure 1.1 and includes the critical parameters.

Equation 3.3 was re-written, by means of dimensional analysis and the re-grouping of parameters, as:

√ (3.4)

With Q = πd2_{/4, it is possible to re-write Equation 3.4 as follows:}

√

(3.5)

The first, third and fourth terms of Equation 3.5 represent the Reynolds, Froude and Weber Numbers.

Anwar, Weller & Amphlett (1978) proposed the following relationship between the six independent non-dimensional parameters that could be used to describe the formation of vortices at horizontal pump intakes:

(

√

(37)

with C = distance from the floor to the underside of the inlet bell (m) A = cross-sectional area of the inlet bell (m2)

S = submergence depth above intake (m). Equation 3.6 was re-written in the following form:

(

)

(3.7)

with Cd = discharge coefficient

Re =

We = √

Knauss (1987) stated that critical submergence could be expressed by:

Scr = f(V, d, D, R, C, Γ, g, ρ, σ, ν) (3.8)

with V = average velocity at the inlet to the suction bell (m/s) d = suction pipe diameter (m)

R = radius of rounding of the bellmouth entry (m).

Equation 3.8 was re-written, by means of dimensional analysis, as:

√

(3.9)

The three last terms of Equation 3.9 represent the Froude, Weber and Reynolds Numbers. The Hydraulic Institute (1998) stated that, based on previous research, it can be concluded that the flow conditions at a pump intake are represented by the following dimensionless parameters:

(38)

√

(3.10)

It is evident that Froude, Reynolds and Weber Numbers need to be considered in physical model studies of pump intakes. The approach flow pattern in the vicinity of the sump governs the circulation, Γ. It can be accepted that circulation would be similar between the prototype and scaled model provided that the approach flow patterns are similar. The relevance and influence of these dimensionless parameters in physical model studies are considered in

Section 3.2.

3.2 Impacts of scale effects in physical models

3.2.1 Physical similarity

It is usually not financially viable to construct physical models of a pump intake at the full-scale of the system. Physical models, constructed at a smaller scale than the full-scale system, are therefore used to examine critical aspects that could influence the performance of the system. The laws of similitude make it possible to relate the results obtained with the model to the performance of the prototype. It is not possible, however, to satisfy all the laws of similitude simultaneously, which results in discrepancies between the performance of the model and that of the prototype. This is known as the “scale effect” (Webber, 1979).

“Physical similarity” is a generic term used to describe different types of similarity, e.g. geometric, kinematic and dynamic similarity. These three types of similarity are discussed in more detail below.

Geometric similarity

Geometric similarity is similarity of shape, which means that the model and prototype are identical in shape but differ in size. The ratio of any two dimensions in the model therefore corresponds to the ratio in the prototype, as shown in Equation 3.11.

(3.11)

with L = a linear dimension m = model

(39)

p = prototype.

The “scale factor” is generally defined as the ratio between a linear dimension in the prototype and the corresponding dimension in the model. If the linear scale of the model is 1:x, the scalar relationship for area is 1:x2 and for volume it is 1:x3 (Webber, 1979).

It is not always possible to achieve complete geometric similarity, especially in respect of surface roughness. The hydraulic behaviour arising from the boundary conditions is, however, the more important factor and dissimilar geometry is therefore often acceptable (Massey, 1989). Kinematic similarity

Kinematic similarity is similarity of motion, which implies similarity in geometry, time intervals, velocity and acceleration (Massey, 1989), meaning that the following ratios will apply between the model and the prototype (Webber, 1979):

and (3.12) with v = velocity a = acceleration.

Fluid motions that are kinematically similar form streamlines that are geometrically similar at corresponding times.

Dynamic similarity

Dynamic similarity is similarity of forces. If two systems are dynamically similar, the magnitude of forces at similarly located points in the model and prototype systems must be in the same ratio and act in the same direction. The following ratios apply between dynamically similar models and prototypes:

(3.13)

(40)

The forces in a system involving fluids could be due to pressure, gravity, viscosity or surface tension. The conditions for dynamic similarity can be expressed as (Webber, 1979):

(3.14)

with Fp = force due to pressure Fg = force due to gravity Fν = force due to viscosity

Fst = force due to surface tension.

Perfect dynamic similarity implies that all the ratios between all the forces remain fixed. This is, however, not the case, as the presence and significance of the various forces differ for different hydraulic structures. It therefore is important to ensure dynamic similarity of the dominant forces present in the model and prototype under consideration.

3.2.2 Similarity laws

The similarity laws that could be applicable to hydraulic model studies are discussed below. Euler Law

The Euler number describes the relationship between pressure and velocity and can be expressed as:

√

(3.15)

with E = Euler Number V = velocity

∆p = change in pressure ρ = density of liquid.

(41)

The Euler Law is relevant to enclosed fluid systems where the forces of gravity and surface tension are absent.

Froude Law

The Froude Law is applicable to systems where gravity is the significant force that influences the fluid motion. Systems with free surfaces, e.g. weirs, open channels, spillways and rivers, are typical examples where gravity is the dominant force.

The Froude number is expressed as:

√ (3.16)

with Fr = Froude Number V = velocity

g = gravitational acceleration

L = characteristic length, e.g. pipe diameter in the case of flow in a pipe.

For compliance with the Froude Law, the corresponding velocities must satisfy Equation 3.17.

(3.17)

with = scale factor.

Reynolds Law

The Reynolds Law is applicable to systems where only the forces of viscosity and inertia are present. An example is a submarine submerged deep enough not to create any waves on the surface.

The Reynolds Number is expressed as:

(42)

with Re = Reynolds Number V = velocity

L = length

= viscosity of liquid.

For compliance with the Reynolds Law, the corresponding velocities must satisfy Equation 3.19:

_(3.19)

Weber Law

The Weber Law describes the relationship between surface tension and velocity. Surface tension is very seldom a significant force, but could become a factor where there is an air-water interface in a structure with small dimensions.

The Weber number is expressed as:

√

(3.20)

with We = Weber Number V = velocity

σ = surface tension of liquid ρ = density of liquid

L = length.

(43)

(3.21)

Based on Equation 3.21 it can be derived that for the same fluid in the model and prototype, the model velocity should be x0.5 times that in the prototype.

Mach Law

The Mach Law describes the relationship between elastic forces and velocity, and is applicable to systems where the compressibility of the fluid is of importance (Massey, 1989).

The Mach number can be expressed as:

(3.22)

with

= Mach Number

V = velocity

C = acoustic velocity in liquid medium

The velocity, flow and time scales for a model based on Froude scaling, can be calculated as follows (Hydraulic Institute, 1998):

(3.23)

(3.24)

The model for the study of the Vlieëpoort pump intake is an open channel with a free surface. Gravitational forces are the dominant forces and the Froude Law will be the criterion to be satisfied. It will, however, be important to ensure sufficiently high Reynolds and Weber Numbers to mitigate the potential scale effects due to viscosity and surface tension. The Euler and Mach Laws are not relevant when performing physical model studies on pump intake channels.

(44)

(3.25)

with Lr = geometric scale m = model p = prototype V = velocity Q = flow T = time. 3.2.3 Modelling criteria

Minimum Reynolds and Weber Numbers

The following is a summary of minimum Reynolds and Weber Numbers, recommended by various authors, to minimise the viscous and surface tension effects for physical models based on Froude similarity:

 Jain et al. (1978) indicated that Reynolds and Weber Numbers of 5 x 104 and 120 would be required as minimum values;

 Daggett and Keulegan (1974; cited in Padmanabhan & Hecker, 1984) found that the viscous effects on vortex formation would be negligible if the Reynolds Number was greater than 3 x 104;

 Zielinksi and Villemonte (1968; cited in Padmanabhan & Hecker, 1984) concluded on the basis of experiments that viscous effects would be negligible for Reynolds Numbers greater than 1 x 104;

 Padmanabhan and Hecker (1984) concluded from their model testing that full-scale inlet losses were predicted accurately by the reduced scale models for Reynolds Numbers of 1 x 105 or greater;

 The Hydraulic Institute (1998) recommended that model studies be performed with Reynolds and Weber Numbers of 6 x 104 and 240, respectively, which include a 2.0 factor of safety on the Reynolds Number recommended by Daggett and Keulegan (1974; cited in Padmanabhan & Hecker, 1984) and the Weber Number recommended by Jain (1978);

(45)

 Jones, Sanks, Tchobanoglous and Bosserman (2008) recommend minimum Reynolds and Weber Numbers of 3 x 104 and 120 respectively; and

 Ahmad, Jain and Mittal (2011) selected the model geometric scale to ensure that minimum Reynolds and Weber Numbers of 6 x 104 and 240 are achieved at the bell entrance. The above Reynolds and Weber Numbers refer to those at the bell entrance.

Based on the above information, it is recommended that the minimum Reynolds and Weber Numbers should be 3 x 104 and 120, respectively, but that where feasible, these Numbers should be increased to 6 x 104 and 240.

Scales of other physical model studies

The following is a summary of scales used in other physical model studies or recommended for model studies:

 Anwar, Weller and Amphlett (1978; cited in Knauss, 1987) suggested a model scale of not less than 1:20 to reproduce vortex formation;

 Prosser (1977) recommended scales varying from 1:4 to 1:25 for sump models;

 A study by Dhillon (1979; cited in Hecker, 1981) indicated good model-prototype agreement of vortex strength for a 1:20 scale model that was operated on Froude similarity;

 A comparison between model-prototype vortex formation was undertaken by Hecker (1981) on 22 projects with model scales varying from 1:12 to 1:120. It was found that, in 16 projects, the model and prototype vortices were essentially equal, whereas five (5) projects indicated that the prototype vortices were stronger than that in the model, and only one (1) study indicated prototype vortices that were weaker than in the model;

 A model study by Arboleda and El-Fadel (1996) to evaluate the impact of approach flow conditions on pump sump design was undertaken at a scale of 1:18;

 Ansar and Nakato (2001) undertook a model study at a scale of 1:10 to investigate the impacts of cross flow on pump intakes; and

 Ahmad et al. (2011) used a 1:10 scale to model a pump sump with five cooling water pumps and three auxiliary cooling water pumps.

It is evident that the above authors recommend scales not exceeding 1:20 for physical model studies of pump intakes that are based on Froude similarity.

(46)

3.2.4 Scale of proposed physical model

The prototype dimensions of the Vlieëpoort pump intake are shown in Table 1.2. It was stated by the TCTA that the design flow per sand trap canal might be as little 1.0 m3/s and as high as 2.5 m3/s, which will result in prototype bell inlet velocities of 0.88 m/s and 2.21 m/s respectively for an inlet bell with a diameter of 1 200 mm.

It follows that the scale that has to be selected to model prototype bell inlet velocities ranging from 0.9 m/s up to 2.2 m/s still has to meet the minimum Reynolds and Weber Numbers for Froude based models. Table 3.1 summarises the prototype information and scale effects based on a model with a scale of 1:10. The velocities and flows were scaled in accordance with Equations 3.23 and 3.24.

It is evident from Table 3.1 that the minimum scaled Reynolds and Weber Numbers will be 3.4 x 104 and 133 respectively for a prototype bell inlet velocity of 0.9 m/s. The Reynolds and Weber Numbers will increase to 9.1 x 104 and 947, respectively, for a prototype bell inlet velocity of 2.4 m/s.

A scale of 1:10 therefore will satisfy the minimum recommended Reynolds and Weber Numbers of 3 x 104 and 120 applicable to physical model studies based on Froude similarity.

(47)

Page | 29

Table 3.1: Prototype information and scale effects

Description Prototype bell inlet velocity = 0.9 m/s Prototype bell inlet velocity = 1.2 m/s Prototype bell inlet velocity = 1.5 m/s Prototype bell inlet velocity = 1.8 m/s Prototype bell inlet velocity = 2.1 m/s Prototype bell inlet velocity = 2.4 m/s P M P M P M P M P M P M Flow (P = m3/s) (M = ℓ/s) 1.02 3.22 1.36 4.29 1.70 5.36 2.04 6.44 2.37 7.51 2.71 8.58 Bell inlet velocity (m/s) 0.90 0.28 1.20 0.38 1.50 0.47 1.80 0.57 2.10 0.66 2.40 0.76 Froude number 0.262 0.262 0.350 0.350 0.437 0.437 0.524 0.524 0.612 0.612 0.699 0.699 Reynolds number

1.1.E+06 3.4.E+04 1.4.E+06 4.6.E+04 1.8.E+06 5.7.E+04 2.2.E+06 6.8.E+04 2.5.E+06 8.0.E+04 2.9.E+06 9.1.E+04

Weber number

13 318 133 23 665 237 36 966 370 53 221 532 72 429 724 94 661 947

P = prototype M = model

(48)

3.3 Design guidelines to calculated critical submergence

Various design guidelines have been published for the calculation of critical submergence. This paragraph summarises the critical submergence requirements as proposed by different authors. Aspects such as approach flow velocities and suction pipework velocities are also addressed for completeness. The various guidelines discussed in the sub-sections below are compared in

Figure 3.3.

3.3.1 The hydraulic design of pump sumps and intakes (Prosser, 1977)  Approach flow to inlet bell ≤ 0.3 m/s;

 Critical submergence, S ≥ 1.5D (D = inlet bell diameter). The recommended mean velocity at the inlet bell is 1.3 m/s; and

 Velocity in suction pipework ≤ 4 m/s.

3.3.2 Centrifugal pump handbook (Sulzer Brothers Limited, 1987)  Approach flow to inlet bell ≤ 0.3 m/s;

 Critical submergence, S ≥ 1.5D (D = inlet bell diameter). The recommended maximum velocity at the inlet bell is 1.3 m/s; and

3.3.3 Pump intake design (Hydraulic Institute, 1998)  Approach flow to inlet bell ≤ 0.5 m/s;

 Critical submergence, S = D(1 + 2.3 Fr). The acceptable inlet bell velocities for a pump flow exceeding 1 260 ℓ/s ranges from 1.2 m/s to 2.1 m/s, with a recommended inlet bell velocity of 1.7 m/s; and

 Velocity in suction pipework ≤ 2.4 m/s.

3.3.4 Pump station design (Jones et al., 2008)  Approach flow to inlet bell ≤ 0.3 m/s;

 Critical submergence, S = D(1 + 2.3 Fr). The maximum allowable velocity at the inlet bell is 1.5 m/s, with 1.1 to 1.2 m/s recommended as the optimum bell intake velocity; and

(49)

3.3.5 Pump handbook (Karassik et al., 2001)  Approach flow to inlet bell ≤ 0.4 m/s;

 Critical submergence, S = D(1 + 2.3 Fr). It is recommended that inlet bell velocities be related to the pumping head as follow:

o Vbell ≤ 0.76 m/s for pumping heads up to 5 m (i.e. 15 feet); o Vbell ≤ 1.20 m/s for pumping heads up to 15 m (i.e. 50 feet); o Vbell ≤ 1.70 m/s for pumping heads greater than 15 m; and

 Velocity in suction pipework ≤ 2.4 m/s.

The reason for linking the pumping head to the maximum allowable bell inlet velocities is to prevent the velocity head loss at the bell inlet to be a large percentage of the total pumping head.

3.3.6 Swirling flow problems at intakes (Knauss, 1987)  Approach flow to inlet bell ≤ 0.3 m/s;

 Critical submergence:

o S = D(1.5 + 2.5 Fr) was proposed by Paterson and Noble (1982); o S = D(1 + 2.3 Fr) was proposed by Hecker (Knauss, 1987); o S = D(0.5 + 2.0 Fr) was proposed by Knauss (1987); o S ≤ 1.5D was proposed by Prosser (1977); and

No recommendation was made by Knauss on the formula to be used for calculating critical submergence, nor were any guidelines provided for allowable bell inlet velocities. Knauss did, however, refer to the guidelines recommended by Prosser (1977) for approach and suction pipework velocities.

3.3.7 Gorman-Rupp design guidelines (Strydom, 2010b)

It is proposed by Gorman-Rupp that critical submergence be determined as per the data presented in Table 3.2.

Physcial hydraulic model investigation of critical submergence for raised pump intakes

PHYSICAL HYDRAULIC

MODEL INVESTIGATION OF

CRITICAL SUBMERGENCE

FOR RAISED PUMP INTAKES

SH KLEYNHANS

DECLARATION

SINOPSIS

SYNOPSIS

ACKNOWLEDGEMENTS

CONTENTS

LIST OF FIGURES

LIST OF TABLES

LIST OF ABBREVIATIONS

1. INTRODUCTION

1.1

Background to the research project

1.2

Motivation for the study

1.3

Objective of study

1.4

Layout of the thesis

2. METHODOLOGY

2.1

Physical model study

2.2

Modelling scenarios

2.3

Criteria to decide on minimum submergence in proto-type

3. LITERATURE REVIEW

3.1

Fundamentals and theory of vortex formation

(

(

)

3.2

Impacts of scale effects in physical models

3.3

Design guidelines to calculated critical submergence