The Thesis

See
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/311985882
A Numerical Model of Flow and Settling in

Sedimentation Tanks in Potable Water
Treatment Plants
Thesis · February 2008

DOI: 10.13140/RG.2.2.11654.37441
CITATIONS READS
3 376
1 author:
Ali hadi Ghawi

University of Al-Qadisiyah
43 PUBLICATIONS 90 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Modeling of Particulate Matter Dispersion in Street Canyons in Baghdad City View project
All content following this page was uploaded by Ali hadi Ghawi on 31 December 2016.
The user has requested enhancement of the downloaded file.

SLOVAK UNIVERSITY OF TECHNOLOGY IN BRATISLAVA
FACULTY OF CIVIL ENGINEERING
Department of Sanitary and Environmental Engineering
A Numerical Model of Flow and Settling in Sedimentation

Tanks in Potable Water Treatment Plants
PhD Thesis
Ing. Ali Hadi Ghawi
Supervisor
Prof. Ing. Jozef Kriš, PhD.
Bratislava, February 2008

A Numerical Model of Flow and Settling in Sedimentation Tanks in Potable Water Treatment Plants
SLOVAK UNIVERSITY OF TECHNOLOGY IN BRATISLAVA

FACULTY OF CIVIL ENGINEERING
Department of Sanitary and Environmental Engineering
A Numerical Model of Flow and Settling in Sedimentation

Tanks in Potable Water Treatment Plants
PhD Thesis
Study Field
5.1.6 Sanitary Engineering - Water Construction
Doctoral: Ing. Ali Hadi Ghawi
Supervisor
Prof. Ing. Jozef Kriš, PhD.
Bratislava, February 2008
II
To the illiterate who taught Arab Nation ……………………… The Noble Prophet Mohamed
To the lips was wished plenty of what is good whenever the spoke……………...
To the one who carried me in weakness and stayed awake caring for my comfort
………………………………………………………………………………...My Dear Mother
To the good heart...to my... first teacher and my model in life……………….. My Dear Father
To the aspect of our childhood and vigor of our health
To those for whom the stage of waiting long to the sources of fulfillment…of promise
To those in whom I find my strength in life…………………………………My Dear Brothers
To the companions of the long road …………………………………… ...My Faithful Friends
To you all we present this humble effort
III
Acknowledgement
At first thanks to Allah All-Mighty God for enabling me to complete this work. Cordial thanks to my
supervisor Prof. Ing. Jozef Kriš, PhD. for his valuable guidance and advices through this work.
I would like to record my thanks to the Hrinova and Holic Water Treatment Plant Staffs for providing part
of the technical support.
I would like to express my sincere appreciation to my family for their continuous encouragement; especial
thanks go to my father for his great support and patience.
Finally, my thanks are due to the Staff of Slovak University of Technology in Bratislava Faculty of Civil
Engineering Department of Sanitary and Environmental Engineering and to all My Friends for their interest and
support during this study.
IV
Abstract
Sedimentation tanks are the workhorses of any water purification process. Sedimentation is
the process of separating solids from the liquid stream. The performance of Sedimentation
tanks (ST) depends on several interrelated processes and factors. A Sedimentation tank
Computational fluid dynamics (CFD) model has been developed to include the following
factors: axisymmetric hydrodynamics, discrete settling, and flocculated discrete settling,
flocculation with 13 classes of particles sizes, temperature changes, various external and
internal geometry configurations, scraper movement, hydraulic loading, variation of
coagulant doses and iron removal. The ST model reproduces the major features of the
hydrodynamic processes and velocity and solids distribution on sedimentation tanks designs.
When the model is executed with the field derived settling characteristics, it can accurately
predict the effluent suspended solids concentrations. The model has been developed and
tested using field data from the Hrinova and the Holic Water Treatment Plants located at
Slovakia. These tanks were selected because performance data are available for model
calibration and validation, and because they suffered from the relatively poor hydraulic
performance. The ST CFD model was used for the case study of the effect of adding several
tanks modifications including flocculation baffle, energy dissipation baffles, perforated
baffles and relocated effluent launders, were recommend based on their field investigation on
the efficiency of solids removal. For the simulated trials, the best combination was an inlet
baffles and replace effluent weir by inboard launder with a perforated baffle. The baffles
modifications can considerably reduce the strength of the density flow and increase the solids
detention time in the tank; the effluent quality can be improved by more than 60% for any
cases. Proper launder modifications can be used to improve local flow pattern near the
effluent weir and to re-distribute the effluent flow along the tank longitudinal direction.
The usually unknown and difficult to be measured particle density is found by matching the
theoretical to the easily measured experimental total settling efficiency. The proposed
strategy is computationally much more efficient than the corresponding strategies used for the
simulation of wastewater treatment.This work deals with the development a specialized
strategy for the simulation of the treatment of potable water in sedimentation tanks. The
strategy is based on the CFD and iron removal model. The coagulant dosage in daily
operation is reduced by more than 50%. After calibration, the model was used to evaluate
different internal configurations.
The changes in temperature on STs play an important role on the performance of STs. Scrape
is important in the settling process and play a big role in changing the flow field. All the
guidelines used in the conventional approach are based on ideal transport assumptions. In this
dissertation we improved the STs guidelines design procedure. The fairly good agreement
between model predictions and field data indicates that the present modeling has achieve a
status that it can be used by design engineers to optimize the design of new tanks and to
diagnose the performance of existing tanks
In general this study demonstrated that CFD could be used in reviewing settling tank design
or performance and that the results give valuable insight into how the tanks are working. It
can be inferred that CFD could be use to evaluate settling tank designs where the tanks are
not functioning properly.
V
Abstrakt
Sedimentačné nádrže sú ťažiskom v každom procese čistenia vody. Sedimentácia je proces, v
ktorom sa separujú tuhé látky od kvapalnej zložky. Činnosť sedimentačných nádrží závisí od
niekoľkých vzájomne súvisiacich procesov a faktorov. Výpočtový model pre dynamiku
kvapalín CFD (softvér pre počítačové modelovanie prúdenia kvapalín) v sedimentačnej
nádrži zahŕňa nasledovné faktory: axisymetrická hydrodynamika, oddelené usadzovanie a
flokulačné oddelené usadzovanie, flokuláciu s trinástimi triedami veľkosti častíc, zmeny
teploty, rôzne vonkajšie a vnútorné konfigurácie geometrie, pohyb stieracieho zariadenia,
hydraulické zaťaženie, zmeny v dávkovaní koagulantu a odstraňovanie železa. Model
sedimentačnej nádrže prenáša hlavné charakteristiky hydrodynamického procesu, rýchlosti a
distribúcie tuhých látok do konštrukcie sedimentačných nádrží. Keď je model vytvorený na
základe usadzovacích charakteristík odvodených z prevádzky, môže presne predpovedať
koncentrácie rozptýlených častíc v odtoku. Model bol vytvorený a testovaný s použitím
prevádzkových údajov z úpravní vody v Hriňovej a Holíči. Tieto nádrže boli vybrané na
základe toho, že, je z nich možné získať údaje pre kalibráciu a validáciu a navyše majú
nedostatočný hydraulický výkon. Počítačové modelovanie dynamiky kvapalín (CFD) v
sedimentačnej nádrži bolo použité na prípadovú štúdiu, ktorá sa zaoberala vplyvmi
modifikácií nádrží, vrátane flokulačných upokojovacích zariadení, usmerňovačov rozptýlenej
energie, perforovaných upokojovacích zariadení a vymenených žľabov odtoku, ktoré boli
odporučené na základe prevádzkového výskumu účinnosti odstraňovania tuhých látok.
Najlepšou kombináciou v simulovaných pokusoch sa ukázali prítokové upokojovacie
zariadenia (usmerňovače) a nahradenie prepadového žľabu vnútorným prepieračom
s perforovaným upokojovacími zariadeniami. Modifikácie usmerňovače môžu výrazne znížiť
silu prúdenia a zvýšiť dobu zdržania tuhých látok v nádrži. Kvalita odtoku z úpravne vody sa
tým môže zlepšiť o viac ako 60 %. Vhodné modifikácie prepieračov môžu byť použité na
zlepšenie profilu prúdenia pri prepadovom žľabe a na redistribúciu prúdenia odtoku pozdĺž
nádrže.
Väčšinou neznáma a veľmi ťažko zmerateľná hustota častíc sa zistí spojením teoretickej
a ľahko zmerateľnej experimentálnej celkovej účinnosti usadzovania. Navrhnutá stratégia je
výpočtovo oveľa účinnejšia v porovnaní s príslušnými stratégiami použitými na simuláciu
čistenia odpadovej vody. Táto práca sa zaoberá vývojom špecializovanej stratégie pre
simuláciu úpravy pitnej vody v sedimentačných nádržiach. Táto stratégia vychádza
z výpočtového modelu pre dynamiku kvapalín a modelu odstraňovania železa. Dávka
koagulantu v každodennej prevádzke je nižšia o viac ako 50 %. Po kalibrácii bol model
použitý na vyhodnotenie rôznych interných konfigurácií.
Zmeny teploty v sedimentačných nádržiach majú dôležitú úlohu pri prevádzke
sedimentačných nádrží. Zhrabovacie zariadenie je dôležité pri procese usadzovania a má
význam pri zmene pola prúdenia. Všetky smernice využité v bežnom prístupe sú založené na
predpokladoch ideálneho transportu. V tejto dizertačnej práci sme vylepšili smernice pre
navrhovanie usadzovacích nádrží. Prijateľná zhoda modelových predpovedí s údajmi
z prevádzky naznačuje, že súčasné modelovanie dosahuje stav, ktorý môže byť využitý
konštruktérmi na optimalizáciu návrhov nových nádrží a na diagnostiku konštrukcií už
existujúcich nádrží.
Vo všeobecnosti táto štúdia demonštruje, že CFD môže byť použité pri revízii návrhov alebo
konštrukcii usadzovacích nádrží, a že výsledky poskytujú hodnotný náhľad do toho, ako
nádrže pracujú. Na záver je možné konštatovať, že CFD je vhodné na vyhodnotenie
konštrukčných návrhov usadzovacích nádrží v prípadoch, kde takéto nádrže nepracujú
správne.
VI
Table of Contents
Title Page
Acknowledgment IV
Abstract V
Abstrakt VI
VII
Table of Contents
X
List of Figures XIII
List of Tables XIV
List of Abbreviations XV
List of Symbols and Units
Chapter 1………………………………………………………………………………………… 1
1 Introduction and objectives ……………………………………………………….……….. 1
1.1 Background and Problem Definition …………………………………………………………. 1
1.2 Numerical Investigation……………………………………………………………………….. 2
1.3 Scope and Objectives …………………………………………………………………………. 4
1.4 Dissertation Organization……………………………………………………………………... 5
Chapter 2………………………………………………………………………………………… 6
2 Basic Concepts and Literatures Review………………………………………………….. 6
2.1 1ntroduction…………………………………………………………………………………… 6
2.2 Computational Fluid Dynamics ………………………………………………………………. 6
2.3 Previous Investigations………………………………………………………………………... 7
2.3.1 Clarifiers and Sedimentation Tanks……………………………………………………... 7
2.4 Insight gained from past CFD efforts ………………………………………………………… 10
2.5 Classification of Process Tanks……………………………………………………………….. 11
2.5.1 Phase Separation Tanks…………………………………………………………………. 11
2.5.1.1 Function and purpose …………………………..………………………………. 11
2.5.1.2 Performance and design………………………………………………………… 11
2.6 Types of settling tanks ………………………………………………………………………... 12
2.7 Ideal sedimentation tank with a horizontal flow pattern……………………………………… 14
2.7.1 Classification of settling behavior……………………………………………………… 14
2.7.1.1 Models Based on Discrete Particles Settling…………………………………… 15
2.7.1.2 Ideal Sedimentation Design Assumptions……………………………………… 16
2.8 Factors influencing the capacity and performance of settling tanks 17
2.8.1 Physical and chemical influences ………………………………………………………. 17
2.8.2 Hydraulic influences ……………………………………………………………………. 17
2.8.3 Solids removal mechanisms …………………………………………………………….. 18
2.8.4 Weir loading and sedimentation tank inlet………………………………………………. 19
2.8.5 Particle Size……………………………………………………………………………… 19
2.8.6 Temperature effect ………………………………………………………………………. 19
2.9 Conclusions …………………………………………………………………………………… 20
Chapter 3………………………………………………………………………………………… 21
3 Mathematical Modelling of Settling Tanks…………………………………………………... 21
3.1 1ntroduction…………………………………………………………………………………… 21
3.2 Design of settling tanks ………………………………………………………………………. 22
3.2.1 Overflow rate……………………………………………………………………………. 22
3.2.2 Residence time…………………………………………………………………………... 23
3.2.3 Reynolds number (Re) ………………………………………………………………….. 23
3.2.4 Froude number (Fr) ……………………………………………………………………... 24
3.2.5 Hazen number (Ha) ……………………………………………………………………... 25
3.2.6 The densimetric Froude number (Fd) ………………………………………………….. 25
3.2.7 Balance of forces………………………………………………………………………… 26
VII
3.3 Fluid Dynamics ……………………………………………………………………………….. 26

3.4 Processes in Settling Tanks …………………………………………………………………… 26
3.4.1 Flow in Settling Tanks…………………………………………………………………... 26
3.5 Modeling Equations …………………………………………………………………………... 27
3.5.1 Conservation of mass……………………………………………………………………. 28
3.5.2 Conservation of momentum…………………………………………………………….. 28
3.5.3 Conservation of energy ………………………………………………………………… 29
3.6 Constitutive relations………………………………………………………………………….. 29
3.6.1 Newton’s generalised law of viscosity………………………………………………….. 29
3.7 Source terms…………………………………………………………………………………... 29
3.7.1 Buoyancy………………………………………………………………………………... 29
3.8 Governing equations for Newtonian fluids……………………………………………………. 29
3.8.1 Navier-Stokes equations………………………………………………………………… 29
3.8.2 Scalar transport equation………………………………………………………………… 30
3.8.3 Energy equation…………………………………………………………………………. 30
3.9 Turbulence and turbulence modelling………………………………………………………… 30
3.9.1 Properties of turbulent flows……………………………………………………………. 30
3.9.2 Calculation methods for turbulent flows………………………………………………... 31
3.9.3 High Reynolds number k-ε model………………………………………………………. 31
3.9.4 Wall functions…………………………………………………………………………… 32
3.10 Governing equations for turbulent Newtonian fluids………………………………………... 32
3.11 Representation of the Motion of Solids …………………………………………………….. 33
3.12 Boundary and operational conditions………………………………………………………... 34
3.13 Conclusions ………………………………………………………………………………….. 34
Chapter 4………………………………………………………………………………………. 35
4 Numerical Computation of Flows……………………………………………………………... 35
4.1 Discretization of the domain: grid generation………………………………………………… 35
4.2 Discretization of the equations……………………………………………………………….... 36
4.2.1 Discretization schemes…………...……………………………………………………… 37
4.2.2 Final discretized equation ………………………………………………………………. 38
4.2.3 Alternative numerical techniques….…………………………………………………….. 38
4.3 Solution methods……………………………..………………………………………………. 38
4.3.1 Solution algorithms for Navier-Stokes equations …………………………...………….. 39
4.3.2 Residuals………………………………………………………………………………… 40
4.3.3 Convergence criteria…………………………………………………………………….. 40
4.4 Numerical techniques used in Fluent …………………………………………………………. 40
4.5 Conclusions …………………………………………………………………………………… 41
Chapter 5…………………………………………………………………………………………………………………………………… 42
5 Experimental Techniques for Model Calibration and Validation……………………………………….. 42
5.1 Introduction……………………………………………………………………………………. 42
5.2 Validation of CFD results…………………………………………………………………….. 43
5.3 Experimental Studies…………………………………………………………………………. 43
43
5.3.1 Settling velocity……………………………………………………………………………………………………………… 44
5.3.1.1 Measurement of Discrete Settling Velocities……………………………………………………… 44
5.3.1.2 Calibration of the Settling Sub-Model ………………………………………………
45
5.4 Solids concentration…………………………………………………………………………… 46
5.5 Liquid velocity………………………………………………………………………………… 47
5.6 Particle size distribution……………………………………………………………………….. 47
5.6.1 Particle size distribution-Experimental determination………………………………….. 48
5.7 Full scale measurements ……………………………………………………………………... 48
5.7.1 Plants layout ………………..…………………………………………………………… 49
5.7.2 Settling tanks description of Holic water treatment plant ………………………………. 51
5.7.3 Settling tanks description of Hrinova water treatment plant ……..……………………... 54
VIII
5.8 Temperature …………………………………………………………………………………... 59

5.9 Conclusions …………………………………………………………………………………… 59
Chapter 6…………………………………...…………………………………………………………….. 61
6 Model Development, Applications and Results……………………………………………… 61
6.1 Introduction……………………….…………………………………………………………… 61
6.2 Modelling the settling tank …………………………………………………………………… 61
6.2.1 Mechanistic description of the settling tank: contributing mathematical equations ……. 62
6.2.1.1 Continuity and Momentum Equations ………………………………………… 63
6.2.1.2 k-ε Turbulence Model…………………………………………………………... 63
6.2.1.3 Energy Equation………………………………………………………………… 64
6.2.2 Particle Trajectory Calculation…………………………..……………………………… 64
6.2.3 The influence of particle structure …………………………………………………….. 65
6.2.4 Simulation………………………………………………………………………………. 66
6.3 Simulation of existing sedimentation tanks…………………………………………………… 68
6.3.1. Hrinova treatment plant ……………………………………………………………….. 68
6.3.2 Holic water treatment plant……………………………………………………………… 74
6.4 Simple improvements to the existing sedimentation tank…………………………………….. 78
6.4.1 Hrinova water treatment plant…………………………………………………………… 78
6.4.1.1 Behavior of Tanks with Modifications………………………………………….. 81
6.4.2 Holic water treatment plant …………………………………………………………….. 82
6.4.2.1 Proposed Tank Modifications ………………………………………………… 82
6.4.2.2 Behavior of Tanks with Modifications………………………………………….. 84
6.5 Proposals Improvement in Holic and Hrinova WTPs…………………………………………. 85
6.5.1 Solids distribution……………………………………………………………………….. 85
6.5.2 Comparison of Settling Models…………………………………………………………. 89
6.6 Temperature effect ……………………………………………………………………………. 89
6.6.1 Temperature measurements…………………………………………………………….. 89
6.6.2 Settling velocity correction factor……………………………………………………… 91
6.7 Water Treatment by Enhanced Coagulation Operational Status and Optimization Issues……. 96
6.7.1 Correlations of Single Variable versus Effluent Fe……………………………………... 96
6.7.2 Correlations of Two Variables versus Effluent Fe………………………………………. 97
6.7.3 Correlations of Three Variables versus Effluent Fe ………………………………….. 98
6.7.4 Coagulant Selections…………………………………………………………………….. 98
6.7.5 Irons Removal Model …………………………………………………………………… 98
6.7.6 Applications of Fe removal Model ……………………………………………………. 99
6.8 Modelling the scraper mechanism………………………………………………………………………. 100
6.9 Design procedures and guidelines…………………………………………………………….. 103
6.10 Validation of the Model …………………………………………………………………….. 105
6.10.1 Holic WTP …………………………………………………………………………... 105
6.10.2 Hrinova WTP………………………………………………………………………….. 107
6.10.3 Validation of solids concentration profiles …………………………………………… 109
6.11 Conclusions …………………………………………………………………………………. 109
Chapter 7 ………………………………………………………………………………………………… 111
7 Conclusions and Recommendations …………………………………………………………. 111
7.1 Development of a CFD model ……………………………………………………………….. 111
7.1.1 Modelling of the scraper ……………………………………………………………….. 112
7.1.2 Particle size distributions……………………………………………………………….. 112
7.1.3 Model validation ……………………………………………………………………….. 112
7.1.4 Iron removal ……………………………………………………………………………. 112
7.1.5 Temperature ……………………………………………………………………………. 113
7.2 Proposed modifications……………………………………………………………………….. 113
7.3 Current design guidelines……………………………………………………………………… 114
7.4 Suggestions for Further Research…………………………………………………………….. 114
References……………………………………………………………………………………… 116
IX
List of Figures
Figure Title Page

Figure 1.1 Overview of the Settling Tank Project…………………………………….. 3
Figure 2.1 Rectangular settling tank with different solids removal mechanisms:
travelling-bridge (top) and chain-and-flight (bottom) type collectors …….. 12
Figure 2.2 Lamella settling tank……………………………………………………….. 13
Figure 2.3 Vertical flow settling tank (Dortmund type): side view (left), and plan view
(right)……………………………………………………………………….. 13
Figure 2.4 Circular settling tank with scraper mechanism and flocculator …………… 14
Figure 2.5 Types of Sedimentation…………………………………………………….. 14
Figure 2.6 Definition diagram for particle terminal settling velocity………………….. 15
Figure 2.7 Sedimentation Basin Design (Ideal Basin)…………………………………. 16
Figure 2.8 Example of flow pattern in a settling tank (---: equal solids concentration
in mg/l) …………………………………………………………………….. 18
Figure 3.1 Flow Processes in a Rectangular ST ………………………………………. 27
Figure 4.1 A finite number of small control volumes (cells) by a grid………………… 35
Figure 4.2 Simple 2D domain showing the cell centers and faces ……………………. 36
Figure 4.3 Solution Procedure…………………………………………………………. 41
Figure 5.1 Process of developing CFD code…………………………………………… 42
Figure 5.2 Laboratory settling column………………………………………………… 45
Figure 5.3 Measured velocity profiles in the clarifier ………………………………… 47
Figure 5.4 Particle size distribution in the influent of the Holič and Hrinova ST……... 48
Figure 5.5 Layout of Holic WTP………………………………………………………. 49
Figure 5.6 Layout of Hrinova WTP……………………………………………………. 50
Figure 5.7 Schematic of Hrinova WTP………………………………………………… 50
Figure 5.8 Schematic representation of the Holic sedimentation tank in 2D………….. 52
Figure 5.9 Picture of Holic sedimentation tank ……………………………………….. 52
Figure 5.10 Schematic representation of the Holic ST inlet and hopper………………... 53
Figure 5.11 Picture of The horizontal settling tank of Hrinova WTP…………………... 55
Figure 5.12 Schematic representations of the Hrinova ST and WTP diagram…………. 56
Figure 5.13 Data monitoring in Hrinova WTP................................................................. 57
Figure 5.14 Daily turbidity measures from 7 June 2006 to 14 august 2006…………….. 58
Figure 6.1 CFD Model………………………………………………………………… 61
Figure 6.2 Flow chart of computation sequence………………………………………. 65
Figure 6.3 Effect of particles number on the number of iterations required to achieve a 66
converged solution………………………………………………………….
Figure 6.4 A flocculation channel……………………………………………………… 68
Figure 6.5 The inlet of Hrinova Settling Tank…………………………………………. 69
Figure 6.6 Rectangular settling tanks at the Hrinova Water Treatment Plant. ……….. 69
Figure 6.7 Diagram of The horizontal settling tank of Hrinova WTP…………………. 70
Figure 6.8 The outlet weir of Hrinova Settling Tank………………………………….. 70
Figure 6.9 The Picture of effluent from ST……………………………………………. 70
Figure 6.10 Impact of Diurnal Flow Variations on Settling Tank Effluent……………... 71
X
Figure 6.11 Diurnal Flow Variations in Settling Tank Effluent………………………… 71

Figure 6.12 Picture of Hrinova Horizontal Sedimentation Tank shows the flocculant
solids observed on the rest of ST……………………………….................... 72
Figure 6.13 3-D grid generation of a ST………………………………………………… 73
Figure 6.14 Velocity vector on XY area………………………………………………… 73
Figure 6.15 Calculated stream lines for existing sedimentation tank…………………… 74
Figure 6.16 Schematic representation of the Holic sedimentation tank in 3D………….. 74
Figure 6.17 Picture for Holic settling tank……………………………………………… 75
Figure 6.18 Picture of the Hopper………………………………………………………. 75
Figure 6.19 Outlet weir………………………………………………………………….. 75
Figure 6.20 Outlet weir…………………………………………………………………. 76
Figure 6.21 Velocity contours of existing tank (m/s)…………………………………… 76
Figure 6.22 Solids concentration profile for existing tank……………………………… 76
Figure 6.23 Solids concentration profile for existing tank……………………………… 77
Figure 6.24 Flow Pattern and Sludge Blanket (in a vertical section along tank central
axis) in Existing Tank……………………………………………………… 77
Figure 6.25 The model simulated velocities and solids distributions for the original and
modifications tanks (in a vertical section along tank central axis) (a.
original tank, b. Perf. Baffle, c. Inboard Launder and Perf. Baf, and d.
Inboard Launder……………………………………………………………. 78
Figure 6.26 Schematic representation of the Hrinova sedimentation tank with
modification………………………………………………………………… 79
Figure 6.27 A Plot of Flocculent Suspended mg/l vs. Time with The Initial Setup of
The Settling Tank………………………………………………………….. 80
Figure 6.28 A Plot of Turbidity vs. Time Showing the Effects of Decreasing Inlet
Velocity and Adjusting the Tank Configurations…………………………... 80
Figure 6.29 Comparisons of Solids Distributions on Surface Layer between Existing
and Modified Tanks………………………………………………………… 81
Figure 6.30 Tank with baffle and launder Modifications (A and B)……………………. 82
Figure 6.31 Flow Pattern (A) and Sludge Blanket (B) (in a vertical section along tank
central axis) in Tank with Modification 1,2,3, and 6………………………. 83
Figure 6.32 Flow pattern as the change of inlet configuration Streamline contour
(Uin=0.23 cm/sec, Cin=500 ppm)…………………………………………… 83
Figure 6.33 Comparisons of Solids Distributions on Surface Layer between Existing
and Modified Tanks (1, 2, 3, and 6)……………………………………..… 85
Figure 6.34 Predicated percents of solids settled for each particle size class…………… 86
Figure 6.35 Contours velocity (m/s) for the existing and modified settling tank in
Hrinova ST…………………………………………………………………. 87
Figure 6.36 Flocs concentration (kg m-3) along the tank bottom for the standard and the
modified tank for particle class size 2 in Hrinova and Holic STs………….. 87
Figure 6.37 Flocs concentration (kg. m-3) along the tank bottom for the Standard and
the Modified Tank for inboard launder and a perforated baffle modified for
particle class size 3 at Hrinova ST………………………………………… 88
Figure 6.38 Flocs concentration (kgm−3) along the tank bottom for the standard and the
modified tank for particle class size 4 in Hrinova and Holic STs………….. 88
XI
Figure 6.39 Calculation results with Fr =0.003 in Hrinova WTP………………………. 88

Figure 6.40 Calculation results with Fr =0.003 in Holic WTP…………………………. 89
Figure 6.41 Temperatures for Hrinova sedimentation tanks on 13/2/06 near the surface
(0.3 m depth) at distance 1 m, 9 m, and 29 m from the inlet of the tank….... 90
Figure 6.42 Temperatures for Hrinova sedimentation tanks on 13/2/06 (2.5 m below
the surface) at distance 1 m, 9 m, and 29 m from the inlet of the tank…….. 90
Figure 6.43 Ratios of VsT2 / VsT1 and μT2 /μT1 for Different suspended solid (SS)
concentrations……………………………………………………………… 92
Figure 6.44 Effect of Temperature on Settling Velocity………………………………... 92
Figure 6.45 Effect of Influent Temperature Variation on the Internal Temperature
Distribution a) time 120 min, b) 220 min. at ΔT= +1ºC, SS= 75 mg/L……. 93
Distribution a) 120 min, b) 220 min. at ΔT= +1ºC, SS= 30 mg/L………… 94
Figure 6.47 Temperature Stratification under the Effect of a Surface Cooling Process
for an Influent Temperature Equal to 26.5ºC………………………………. 95
for an Influent Temperature Equal to 11.0ºC………………………………. 95
Figure 6.49 Correlation between effluent Fe and ESS concentration…………………… 96
Figure 6.50 Correlation between ESS and Coagulant dosage…………………………... 96
Figure 6.51 Correlation between effluent Fe and flocculation chemical dosage………... 97
Figure 6.52 Correlation between Fe removal and chemical dosage…………………….. 97
Figure 6.53 Correlative relationships between relative dosage and dosage/Fe removal
ratio revealed in this study ………………………………………………… 98
Figure 6.54 Impact of Settling Tank Effluent SS and Flocculation Chemical Dosage on
Irons Removal Efficiency [(Fe – Feo)/Fe = Dosage /(0.4145*SS +
1.0992*Dosage)]……………………………………………………………. 99
Figure 6.55 Settling tank flow capacity in base design and optimized alternative……… 100
Figure 6.56 Picture shows Hrinova ST scraper………………………………………...... 101
Figure 6.57 Picture shows Holic ST scraper…………………………………………….. 101
Figure 6.58 ST scraper for Holic and Hrinova…………………………………………... 101
Figure 6.59 Effect of scraper on solids concentration profiles………………………….. 102
Figure 6.60 Validation of the ESS Simulated by the Model…………………………….. 106
Figure 6.61 Comparison between predicted and measured Velocity Profiles…………... 106
Figure 6.62 Detail of measured and simulated solids concentration profiles mg/l……… 107
Figure 6.63 Input profiles for the 2-day pre-simulation to initialise the settling tank
model……………………………………………………………………….. 108
Figure 6.64 Comparison between predicted and measured solids concentration……….. 108
Figure 6.65 Simulated and experimental particle size distribution in the effluent of the
existing tank………………………………………………………………… 109
XII
List of Tables
Table Title Page

Table 2.1 Current phase separating tank guidelines…………………………………... 12
Table 3.1 Horizontal sedimentation tank guidelines …………………………………. 22
Table 3.2 Horizontal sedimentation tank overflow rates …………………………….. 23
Table 3.3 Recommended typical values of the constants in the k-ε turbulence
model.............................................................................................................. 33
Table 5.1 Discrete Settling Velocities of Large Flocs for Holic ad Hrinova WTPs.… 45
Table 5.2 Discrete Settling Velocities of Medium Flocs……………………………... 46
Table 5.3 Physical and hydraulic data during study periods, and settling tank data….. 53
Table 5.4 Observed data on 16/01/2006 (Holic WTP)…….………………………….. 54
Table 5.5 Observed data 24/07/2006 (Holic WTP)..………………………………….. 54
Table 5.6 Observed data February 13/2006 (Hrinova WTP)………………………… 58
Table 5.7 Observed data June 19/2007 (Hrinova WTP)……………………………… 59
Table 5.8 measured temperature in winter (average) and summer (average) at
Hrinova WTP………………………………………………………………. 59
Table 6.1 Classes of particles used to account for the total suspended solids in the
STs in Holic and Hrinova STs……………………………………………… 67
Table 6.2 Performance data for modeled settling tank………………………………... 79
Table 6.3 Performance data for modelled settling tank……………………………….. 81
Table 6.4 Summary of Loading and Effluent Concentration in Tank………………… 84
Table 6.5 Settling velocity and dynamic viscosities for summer and winter
temperature…………………………………………………………………. 91
Table 6.6 Predicted ESS values for Different Temperature Variations………………. 93
Table 6.7 The conventional design procedure according to van Duuren (1997) with
its corresponding assumptions and deviations from real tanks…………….. 103
Table 6.8 Adjusted design procedure according to kawamaura (2000) with its
corresponding assumptions and implications. ……………………………... 104
Table 6.9 Proposed CFD enhanced design procedure………………………………… 105
Table 6.10 Comparison of model predictions with field data………………………….. 105
Table 6.11 Comparison of model predictions with field data………………………….. 109
XIII
List of Abbreviations
The most commonly used symbols and their general meanings are listed below:
Abbreviations Description
0D, 1D, 2D, 3D Zero-dimensional, One-, Two- and Three-dimensional
ADI Alternating Direction Implicit
ADV Acoustic Doppler Velocimetry
CFD Computational Fluid Dynamics
DAF Dissolved Air Flotation
DPM Discrete Phase Model
EPA Environmental Protection Agency
ESS Effluent Suspended Solids
Fd densimetric Froude number
FDM Finite Difference Method
FEM the Finite Element Method
Fr Froude number
FVM Finite-Volume Method
H Hazen number
HLPA Hybrid Linear/Parabolic Approximation
KE Kinetic energy
MTS Staiger-Mohilo
NTU Nephelometric Turbidity Units
ODEs Ordinary Differential Equations
PDEs Partial Differential Equations
PE Potential energy
PISO Pressure Implicit with Splitting of Operators
PSD Particle Size Distribution
QUICK Quadratic Upstream Interpolation for Convective Kinetics
RANS Reynolds-Averaged Navier-Stokes
Pe Peclet Number
Re Reynolds number
RTD Residence Time Distribution
SIMPLE Semi-Implicit Method for Pressure-Linked Equations
SIMPLEC SIMPLE-Consistent
SIMPLER SIMPLE-Revised
3 -2 -1
SOR Surface Overflow Rate [L L T ]
-3
SS Suspended Solids (concentration) [ML ]
SST Secondary Settling Tank
ST Sedimentation Tank
STESS Settling tank effluent SS concentration
-3
TDS Total Dissolved Solids [ML ]
Type I Discrete particles
Type II Flocculating Free Particles
Type III Hindered particle groups.
Type IV Compression particles
WHO Water Health Organization
WTP Water Treatment Plant
XIV
List of Symbols and Units

English symbols
Abbreviation Description Unit

c Solids concentration…………………………………….. [M/L3]
C1, C2, C3 Modelling constants……………………………………. [-]
CD Drag coefficient………………………………………… [-]
ci Mass fraction of species i……………………………….. [-]
cmin Concentration on non-settling flocs…………………….. [M/L3]
Cμ Turbulence model constant…………………………… .. [-]
D Fractal dimension……………………………………….. [-]
2 -1
Di Laminar diffusivity of species i………………………… [L T ]
DP Particle diameter………………………………………... [L]
-2 -2
E Total energy …………………………………………… [ML T ]
H Tank depth …………………………………………….. [L]
h Depth of the water ……………………………………… [L]
K Constant………………………………………………… [-]
L Tank length ……………………………………………. [L]
-1
n overall settling effectiveness……………………………. [LT ]
ni Percent of solids settled for each particle size class……. [-]
-1 -2
P Static pressure…………………………………………... [ML T ]
-2 -2
Q Heat generated in control volume………………………. [ML T ]
-2 -2
q Heat flux per through control volume face……………... [ML T ]
R The wetted perimeter of the tank ………………………. [L]
T Temperature……………………………………………. Co
t Time ……………………………………………………. [T]
T , T' Mean and fluctuating component of the temperature…. Co
-1
U Average horizontal velocity in the tank ………………. [LT ]
-1
u Velocity in the tank …………………………………… [LT ]
-1
U,V Time-averaged velocities in the X and Y direction…… [LT ]
-1
uj Fluid velocity vector in i-direction……………………... [LT ]
-1
uP Particle velocity………………………………………… [LT ]
3
V Total process tank volume……………………………… [L ]
-1
VsT1 , VsT2 Settling velocities at temperatures T1 and T2 respectively [LT ]
X Axial coordinate
Y Vertical coordinate
-1
Vsp Terminal settling Velocity……………………………… [LT ]
2
A Surface area……………………………………………... [L ]
-1
Q Rate of flow…………………………………………….. [L3 T ]
L, W Length, and Width …………………………………….. [L]
S Particle specific gravity…………………………………
s
dp Diameter of the particle………………………………… [L]
S,W,E,N South, West, East, North
Si Source term
XV
Greek symbols
Abbreviation Description Unit

feff Equivalent solid sphere……………………………………………. [-]
Cμ Turbulence modelling constant……………………………………. [-]
fb Body force per unit mass
feff Equivalent solid sphere. …………………………………………... [-]
-2
g Gravitational acceleration ………………………………………… [LT ]
2 -2
k Turbulent kinetic energy …………………………………………. [L T ]
2 -1
Γ Turbulent diffusivity ……………………………………………... [L T ]
δij Kronecker delta function…………………………………………... [-]
Δx, Δy Mesh size…………………………………………………………...
2 -3
ε Turbulent dissipation rate ………………………………………… [L T ]
-1 -1
μ Laminar viscosity …………………………………………………. [ML T ]
-1 -1
μeff Effective viscosity ………………………………………………… [ML T ]
-1 -1
μ Water dynamic viscosity ………………………………………….. [ML T ]
ref
-1 -1
μ Laminar viscosity …………………………………………………. [ML T ]
-1 -1
μt Turbulent viscosity ……………………………………………….. [ML T ]
-1 -1
μT1 , μT2 Dynamic viscosities of the mixtures………………………………. [ML T ]
2 -1
ν Kinematic viscosity ……………………………………………….. [L T ]
2 -1
νeff Effective viscosity…………………………………………………. [L T ]
2 -1
νt Isotropic eddy viscosity …………………………………………… [L T ]
Πij Stress tensor representing normal and shear stresses
-3
ρ Fluid mixture density……………………………………………… [ML ]
ρP Particle density…………………………………………………….. [L]
-3
ρs Dried solids density……………………………………………….. [ML ]
-3
ρw Density of clear water .…………………………………………… [ML ]
σk Turbulent Schmidt number for k……………………………….….. [-]
ρP Particle density…………………………………………………….. [L]
σε Turbulent Schmidt number for ε…………………………………... [-]
-1 -1
μeff Effective viscosity ………………………………………………… [ML T ]
φ Scalar
XVI
Chapter 1
Introduction and Objectives
1.1 Background and Problem Definition

Sedimentation is perhaps the oldest and most common water treatment process. The principle
of allowing turbid water to settle before it is drunk can be traced back to ancient times. In
modern times a proper understanding of sedimentation tank behavior is essential for proper
tank design and operation. Generally, sedimentation tanks are characterized by interesting
hydrodynamic phenomena, such as density waterfalls, bottom currents and surface return
currents, and are also sensitive to temperature fluctuations and wind effects.
On the surface, a sedimentation tank appears to be a simple phase separating device, but
down under an intricate balance of forces is present. Many factors clearly affect the capacity
and performance of a sedimentation tank: surface and solids loading rates, tank type, solids
removal mechanism, inlet design, weir placement and loading rate etc. To account for them,
present-day designs are typically oversizing the settling tanks. In that way, designers hope to
cope with the poor design that is responsible for undesired and unpredictable system
disturbances, which may be of hydraulic, biological or physic-chemical origin.
To improve the design of process equipment while avoiding tedious and time consuming
experiments Computational Fluid Dynamics (CFD) calculations have been employed during
the last decades. Fluid flow patterns inside process equipment may be predicted by solving
the partial differential equations that describe the conservation of mass and momentum. The
geometry of sedimentation tanks makes analytical solutions of these equations impossible, so
usually numerical solutions are implemented using Computational Fluid Dynamics packages.
The advent of fast computers has improved the accessibility of CFD, which appears as an
effective tool with great potential. Regarding sedimentation tanks, CFD may be used first
for optimizing the design and retrofitting to improve effluent quality and underflow
solids concentration. Second, it may increase the basic understanding of internal
processes and their interactions. This knowledge can again be used for process
optimization. The latter concerns the cost-effectiveness of a validated CFD model where
simulation results can be seen as numerical experiments and partly replace expensive field
experiments (Huggins et al. 2005).
From a hydraulic point of view, a distinction has to be made between primary and secondary
settling tanks in terms of density effects. In secondary clarifiers the increased density (due to
large particle mass fraction) gives rise to a couple of characteristic flow features such as
density waterfall phenomenon near the inlet of the clarifier and solid cascading phenomenon
in clarification of suspended solids (Kim et al. 2005). In the case of potable water treatment
the solid mass fraction is even smaller than that of the primary clarifiers.
Much research has been done on secondary sedimentation tanks for wastewater treatment.
Larsen 1977 was probably the first who applied a CFD model to several secondary clarifiers
and, although his model incorporated several simplifications, he demonstrated the presence of
a “density waterfall”, which is a phenomenon that causes the incoming fluid to sink to the
tank bottom soon after entering. Shamber and Larock 1981 used a finite volume method to
solve the Navier-Stokes equations, the k-ε model and a solids concentration equation with a
settling velocity to model secondary clarifiers. McCorquodale et al. 1991 developed a model
using a combination of finite element methods (for the stream function) and finite difference
methods (for the boundaries). McCorquodale and Zhou 1993 investigated the effect of
various solids and hydraulic loads on circular clarifier performance, whereas Zhou et al.
-1-
1994 linked the energy equation with the Navier-Stokes equations to simulate the effect of
neutral density and warm water into a model clarifier.
Krebs et al. 1995 used the Phoenics code to model different inlet arrangements and
evaluated the effect of inlet baffle position and depth. Deininger et al. 1998 improved the
Champion3D numerical model and predicted the velocity and solids distribution in a circular
secondary clarifier, whereas Kim et al. 2005 have recently performed a numerical simulation
in a 2-D rectangular coordinate system and an experimental study to figure out the flow
characteristics and concentration distribution of a large-scale rectangular final clarifier in
wastewater treatment.
With respect to primary sedimentation tank, where the solids concentration is limited and
discrete settling prevails, Imam et al. 1983 applied a fixed settling velocity and used an
averaged particle velocity. Stamou et al. 1989 simulated the flow in a primary sedimentation
tank using a 2D model in which the momentum and solid concentration equations were
solved but not linked to account for buoyancy. Adams and Rodi 1990 used the same model
as in 1989 and did extensive investigations on the inlet arrangements and the flow through
curves. More advanced is the work of Lyn et al. 1992 that accounts for flocculation where
six different size classes with their respective velocities were considered. Frey 1993 used the
VEST code to determine the flow pattern in a sedimentation tank. The flow profiles were
then used by the TRAPS code to determine particle tracks. Van der Walt 1998 used the 3D
Flo++ code to determine the sensitivity of a primary sedimentation tank behavior on a
number of geometric, fluid and solids transport properties and simulated the existing Vaalkop
sedimentation tanks using a 3D pseudo two-phase model demonstrating how the inlet
geometry was the main cause of the poor desludging capacity.
Generally, many researchers have used CFD simulations to describe water flow and
solids removal in settling tanks for sewage water treatment. However, works in CFD
modelling of sedimentation tanks for potable water treatment, rectangular
sedimentation tanks, and iron removal by sedimentation tank in surface and
groundwater treatment plants have not been found in the literature. Moreover, the
physical characteristics of the flocs may not be such significant parameters in the flow
field of sedimentation tanks for potable water, due to the much lower solids
concentrations and greater particle size distributions than those encountered in
wastewater treatment.
1.2 Numerical Investigation

Settling by gravity is of great importance in water and waste water treatment where settling
tanks can account for 30% of total plant investment. Despite the practical importance of these
tanks, current design practice relies heavily on empirical formulae which do not take full
account of the detailed hydrodynamics of the system.
In recent years efforts have been made to replace empirical design methods by mathematical
models which accurately reproduce the physical processes involved in sedimentation tanks,
Stamou and Rodi (1984). The basic differential equations governing the flow and
concentration field can be assembled and solved by numerical methods on computers. In this
way the effects of geometric changes in tank configuration and variations in other parameters,
such as influent flow rates and the sedimentation characteristics of suspended solids, can be
predicted. This would make a contribution to optimising tank design and operational
efficiency.
The effects of other physic - chemical processes such as flocculation and particle break-up,
can be included in the mathematical model thereby indicating what effects these processes
have on the overall efficiency of the tank
-2-
Computational Fluid Dynamics (CFD) is a well established technique for numerical

simulation by which any flow behavior can be described. The advantage of high-speed and
large-memory computers has enabled CFD to obtain solutions for many flow problems
including compressible or incompressible, laminar or turbulent, reacting or non-reacting,
single phase or multiphase, and steady or transient flows. Multiple combinations of the above
flows can also be solved with the help of CFD.
Design of sedimentation tanks for water and wastewater treatment processes are often
based on the surface overflow rate of the tank. This design variable is predicated on the
assumption of uniform unidirectional flow through the tank. Dick (1982), though,
showed that many full-scale sedimentation tanks do not follow ideal flow behavior because
suspended solids removal in a sedimentation tank was often not a function of the overflow
rate. Because of uncertainties in the hydrodynamics of sedimentation tanks, designers
typically use safety factors to account for this nonideal flow behavior (Abdel-Gawad and
McCorquodale, 1984b).
Non-ideal flow and the performance of Sedimentation Tank (ST) depend on several
interrelated processes; for simplicity, these processes have been divided into six groups: (A)
hydrodynamics, (B) settling, (C) turbulence, (D) sludge, (E) flocculation, and (F) heat
exchange and temperature changes. At the same time, these processes depend on numerous,
also interrelated factors, that include: (1) the geometry of the tank, including inlet and outlet
configurations, sludge withdrawal mechanisms, internal baffles; (2) loading, including solids
and hydraulic loading, and time variations; (3) the nature of the floc, including the settling
properties and the tendency to aggregation and break up; and (4) the atmospheric conditions,
including ambient and water temperature, Naturally, the weight of these processes and factors
is variable, and therefore neglecting assumptions can be made. However, a complete model
for ST must include sub-models for the six aforementioned groups, allowing for the
representation of the interrelated factors. Obviously this is not an easy task, and so far it has
not been completed (to the knowledge of the author).
It can be concluded from the discussion that the current ways in which STs are designed
and modified could and should be improved. Providing a tool that might lead to
sedimentation tank optimization, as well as understanding, quantifying and visualizing
the major processes dominating the tank performance, are the main goals of this
research.
Retrofit and
Improvement of
Settling Tanks
"Stability" Design
Monitoring "Dynamics"
Monitoring
On-Line:
y Cogulation Doses On-Line:
2D-Settling Tank
y pH y Cogulation Doses Modelling
y Temperature y pH CFD
Off-Line: y Particles Size Distribution Prediction of Velocities, Temperature,
Effluent: y Temperature Iron and Solids Concentration Profiles
y Iron Off-Line:
y Manganese y Settling Velocity
y SS y SS
Sludge
y Solids Concentrations Virtual
Optimal
Experimental
Design
Figure 1.1 Overview of the Settling Tank Project
-3-
1.3 Scope and Objectives

This research focuses on the development of a CFD Model that can be used as an aid in
the design, operation and modification of sedimentation tanks. This model represents in a
2D scheme the major physical processes occurring in STs. However, effect of scrapers and
inlet are also included, hence the CFD Model definition. Obviously, such a model can be a
powerful tool; it might lead to rectangular sedimentation tanks optimization, developing
cost-effective solutions for new sedimentation projects and helping existent
sedimentation tanks to reach new-more demanding standards with less expensive
modifications. An important benefit is that the model may increase the understanding of
the internal processes in sedimentation tanks and their interactions. A major goal is to
present a model that can be available to the professionals involved in operation,
modification and design of sedimentation tanks; in this respect, the model was developed
following two premises: first; a non-commercial code for the solver was developed; second,
the recalibration of the model for the application to specific cases was designed to be as
straightforward as possible, i.e., whenever possible the theory with the simplest calibration
parameters was used.
In this respect, the STs Project was started in 2005. It is a joint project of VEGA and KEGA
(Department of Sanitary and Environmental Engineering) at The Faculty of Civil Engineering
(The Slovak University of Technology in Bratislava). The ultimate goal of the project is to
develop a new CFD methodology for the analysis of the sediment transport for multiple
particle sizes in full-scale sedimentation tanks of surface and groundwater potable
water treatment plants with high iron concentration. The CFD package FLUENT 6.3.26
was used for the case study of the effect of adding several tank modifications including
flocculation baffle, energy dissipation baffles, perforated baffles and relocated effluent
launders, were recommend based on their field investigation on the efficiency of solids
removal. An overview of the outline of the project is given in Figure 1.1.
The specific objectives of this research include:
Improve the operation and performance of horizontal sedimentation tank in Slovakia
which have been identified as operating poorly, by predicting the existing flow,
coagulant dose to remove iron and flocculent concentration distribution of the
sedimentation tank by means of CFD techniques.
Develop a mathematical model for sedimentation tanks in 2D;
Introduce a flocculation submodel in the general ST model,
Introduce a temperature submodel in the general ST model.
Design CFD model for simulation of sedimentation tanks, i.e. grids and numerical
descriptions.
Develop a model calibration procedure, including the calibration of the settling
properties, and validate the models with experimental data.
Evaluate the suitability of CFD as a technique for design and research of rectangular
sedimentation tanks for drinking water treatment plants and iron removal.
Use CFD to investigate the effects of design parameters and operational parameters.
Finally, a CFD model was developed to simulate the full scale rectangular sedimentation
tanks at the Holic and Hrinova purification works in Slovakia. The CFD simulations of
the Holic and Hrinova tanks were done by setting up standard cases for each, i.e. a
configuration and operating conditions that represented the physical tanks as they were built,
and then varying different aspects of the configuration or operating conditions one or two at a
time to determine the effect. Discrete particles in dilute suspension were simulated, as it is
the applicable type for the operating conditions in rectangular sedimentation tanks for potable
water treatment.
-4-
1.4 Dissertation Organization

The structure of the present work is the following: Chapter 1 introduces the topic, presents a
short description about ST modeling, and discusses the problem, the dissertation scope and
objectives, and the organization of the document.
Chapter 2 presents a basic concepts and literature review on the topics related to the
dissertation. Hydrodynamics of settling tank, settling properties of the sludge, turbulence
modeling, flocculation, and temperature effects are the major topics discussed in this chapter.
Chapter 3 discusses the modeling principles of hydraulic and solids transport in settling
tanks. Although it principally describes the tank in two and/or three dimensions, simple
models will be briefly mentioned as well. The complexity of the former models increases
further by including phenomena such as turbulence. Because turbulence has a significant
influence on flow behaviour, it cannot be neglected in the model. Of course, the complete
model can only be solved when the initial hydraulic and solids fields are known; also the
conditions at the boundaries of the tank should be defined.
Chapter 4 will therefore deal with some general computational issues. To calibrate and
validate a model, experiments are indispensable. Data about solids settling velocities are
typically determined off-line. To confront the model with reality, however, it is preferable to
collect data in situ. Depending on the model, information such as liquid and/or particle
velocities, particle size distributions and solids concentrations are needed. This is not always
self-evident as instruments mostly are invasive and may alter the measurement conditions.
Hence, care should be taken with the interpretation of data. In this respect, Chapter 5 gives
an extensive overview of measurement techniques used for this purpose with their advantages
and disadvantages. Where possible, data from this research will be used for illustration.
Chapter 5 presents the CFD-based simulation strategy developed with respect to the specific
features and conditions of a potable water sedimentation tank. It describes the ways of
particle trajectories calculation, based on their small mass loading, and of the handling of the
different particles size classes. In addition, the influence of particle structure is discussed and
a method for particle density calculation is developed. Also chapter 6 presents the outcomes
of the standard and the modified tank simulations concerning the flow pattern and the solids
distribution and discusses the CFD model validity and the influence of the improvement on
the solids settling. Chapter 6 deals with the simulation results of a case study, in which a
settling tank at Holic and Hrinova WTP (Slovakia) is considered. The modelling procedure
followed is discussed step by step. Data about sedimentation has been collected. To validate
the model, calculated solids concentration profiles are compared with measured profiles. The
predicted flow field, on the other hand, is studied by means of a simplified flow-pattern test.
Note that all validations have been conducted under non-steady conditions due to the
diurnally changing influent conditions of the treatment plant.
Finally, Chapter 7 states the general and specific conclusions of the research.
Recommendations for improving the model and future research are also presented in this
chapter. Appendices provide background, general and detail information, and most of the
data collected during the development of this research.
-5-
Chapter 2
Basic Concepts and Literatures Review
2.1 Introduction
Successful design and reliable operation of new water treatment facilities and the constant
upgrading of existing plants has always been, and continues to be a major concern for all
world water professionals. The main objective is to ensure the adequate treatment process and
to guarantee its maximum efficiency.
Sedimentation is perhaps the oldest and most common water treatment process. The principle
of allowing turbid water to settle before it is drunk can be traced back to ancient times. The
Assyrians, Egyptians, Incas and Romans made extensive use of this technique (Forbes and
Dijksterhuis, 1963).
Sedimentation tanks are the workhorses of any water purification process. It is thus crucial
for the sedimentation tank to be operated to its full potential. It is not only the chemical
aspects of flocculation that cause problems, however. Hydraulics also play a prominent part.
Over-design of plant is common, leading not only to unnecessary capital expenditure, but
also to water wastage in the form of excessive sludge. Inadequate design causes overloading
of filters, leading to frequent backwashing, which also wastes a significant percentage of
treated water. Many plants are already a few decades old and do not incorporate the latest
developments in technique, e.g. inlet design. Sedimentation tank performance is strongly
influenced by hydrodynamic and physical effects such as density driven flow, gravity
sedimentation, flocculation and thickening. In turn the velocity and density patterns in tanks
influence these processes and are therefore of great interest to design engineers.
The primary performance indicator for sedimentation tanks is the fraction of the solids
present in the raw water removed by the sedimentation step. If the intricate balance of forces
(hydrodynamic and physical effects) is not taken into account it may lead to poorly
performing tanks:
1. The tank may not remove enough suspended solids. This results in high outlet
turbidity levels that affect downstream processes. High outlet turbidity can be
overcome by higher coagulant dosages or increased backwash frequencies, but at
an increased operating cost.
2. The tank may also not remove enough settled solids. These results in suspended
solids that end up in areas where it cannot be removed with the integral sludge
removal devices. Accumulation of sludge can eventually lead to tank short-
circuiting and requires tanks to be taken out of service for cleaning.
2.2 Computational Fluid Dynamics
Computational fluid dynamics (CFD) is the numerical solution of the mathematical equations
describing the behaviour of fluids (air, water, and mixtures), to enable investigation of details
of the flow regime (velocities, pressures, and fluxes) to be analyzed at a spatial and temporal
resolution that is difficult and/or expensive to achieve by other means (observation, direct
measurement, and inference).
Computational Fluid Dynamics (CFD) is the analysis of systems involving fluid flow by
means of computer-based simulation. It is a research tool and a design tool and it is
complementary to theory and experiments. CFD can also be described as a method to
investigate and simulate fluid flow by means of iterative calculations on computers. It was
developed originally to study aerodynamics, but has since been applied to many different
types of flow under a great variety of conditions. A recent application (since ca 1990) is to
-6-
use it for the simulation of unit processes in water treatment, e.g. chlorine and ozone
contactors, sedimentation tanks and sludge thickeners.
The basic technique utilized in CFD is to divide the region or component in or through which
the flow is to be investigated (the flow domain) into a grid of thousands of small blocks.
Boundary types and conditions for the domain (inflow, outflow, pressure, and symmetry),
values for inflow (flow rate, concentrations, temperature) and physical values for the fluid
(density, viscosity) are specified. The computer then calculates values for velocity
components, pressure, momentum, energy and concentrations for each cell iteratively until
the values converge for a steady state solution, or for a required number of iterations that
yield sufficient accuracy for nonsteady flow cases. The results can then be visualized by
displaying it graphically by looking at views (slices) through the flow domain, for example as
velocity vectors, pressure contours or particle tracks. It can also be imported into applications
like spreadsheets for further mathematical analysis.
The application of CFD to the simulation of sedimentation tanks is fairly recent because it
involves modelling of the movement of solid particles in water, i.e. two-phase flow. This
increases computer memory and processor speed needs. With the advent of PC's with
processor speeds in the Giga-Herz range and Gigabytes of RAM it has become feasible and
has put CFD as a tool within the scope of standard consulting practice in the design and
operation of water purification and treatment works.
Modelling and simulations mean nothing, of course, independent of the reality they are
supposed to represent. The accuracy of the simulation must be checked, or validated, against
data obtained from real operating systems.
2.3 Previous Investigations
2.3.1 Clarifiers and Sedimentation Tanks
Although CFD is an advanced modelling technique it is not a novel technique. CFD
techniques were applied to secondary clarifiers in the late 1970s. Larsen (1977) applied a
very basic CFD model to several secondary clarifiers in Sweden. Although his model
incorporated several simplifications, he was the first to show the presence of a ‘density
waterfall’. A density waterfall is a phenomenon that causes the incoming fluid to sink to the
tank bottom soon after entering. This phenomenon is caused by the density differences of the
incoming fluid and the water already in the tank. Density differences can be caused by the
presence of suspended and dissolved solids as well as temperature differences. Several
investigators have used CFD techniques since Larsen.
The first 2D clarifier model was presented by Larsen (1977). His model, developed for
rectangular clarifiers, was based on the equations of motion, continuity and an exponential
equation relating settling velocity to concentration. He introduced the concept of stream
function and vorticity, and the generation of vorticity by internal density gradients and shear
along solid boundaries. Diffusivity was assumed equal to eddy viscosity, which was
computed on the basis of the Prandtl mixing length theory.
Schamber and Larock (1981) introduced the k-ε turbulence into a finite element model to
simulate neutral density flow in the settling zone of a rectangular tank. Imam et al. (1983)
developed and tested a numerical model to simulate the settling of discrete particles in
rectangular clarifiers operating under neutral density conditions. A two-step Alternating
Direction Implicit (ADI), weighted upwind-centered finite difference scheme was used to
solve the 2D sediment transport and vorticity-transport stream function equations. Abdel-
Gawad and McCorquodale (1984a) applied a strip integral method (SIM) to a primary
circular settling tank in order to simulate the flow pattern and dispersion characteristics of the
flow under steady conditions. The authors expanded their work (Abdel-Gawad and
McCorquodale, 1985) coupling the hydrodynamics with a transport model to simulate the
transport and settling of primary particles in circular settling tanks; the model was restricted
-7-
to the neutral density case. Celik et al. (1986) presented a numerical Finite-Volume Method
(FVM) using the k-ε turbulence model for predicting the hydrodynamics and mixing
characteristics of rectangular settling tanks.
DeVantier and Larock (1986) used a finite element approach with a k-ε turbulence model
and solids concentration to model clarifier behaviour. They linked the solids concentration
with a body force term in the y-momentum equation and adjusted the turbulence model for
buoyancy. DeVantier and Larock (1987) solved the two-dimensional x- and r-momentum
equations, the k-ε turbulence model and the solids transport model. They solved the equations
by using the Galerkin finite element method and the Newton-Raphson solver. Solutions were
obtained for the eddy viscosity, velocity vectors and solids concentration contours for various
inlet concentrations. A sensitivity analysis of the constants of the k-ε model was done at the
inlet and was found not to be critical. They also found that the settling velocity, which was
assumed to be constant, affected the removal efficiency considerably. Inlet concentrations of
less than 1400 mg/l were used. Numerical instability was reported at higher inlet
concentrations.
Stamou, et al., (1989) used a 2D model to simulate the flow in a primary sedimentation tank.
The momentum and solid concentration equations were solved, but not linked to account for
buoyancy. They did, however, note that the variation of the Schmidt number for turbulent
diffusion of the solids did not have a marked influence (a 100% increase in Schmidt number
resulted in only a 1.2% change in predicted removal efficiency). Numerically generated RTD
curves showed a definite lag compared to experimental data. Eddy viscosity was shown to
vary by a factor of 100 throughout the tank. A constant eddy viscosity can therefore not be
used.
Casonato and Gallerano (1990) developed a self-adaptive finite difference model. A novel
bottom boundary condition was used, where the concentration at the bottom wall was
determined as a function of the concentration, particle size and shear velocities.
Zhou and McCorquodale (1992) developed a mathematical model to predict the velocity
and concentration distribution of a non-uniform flocculated suspension for turbulent density
stratified flow in secondary clarifiers. The k-ε turbulence model was used (Rodi, 1980),
together with conservation equations for mass, momentum and sediment concentration. The
cylindrical polar co-ordinate system was used, and a 2 dimensional model of a centrally fed
clarifier with a circumferentially placed withdrawal weir was constructed and solved. The
SIMPLE solution algorithm of Patankar and Spalding (1972) was employed.
McCorquodale et al. (1991) developed a model using a combination of finite element
methods (for the stream function) and finite difference methods (for the boundaries). The
vorticity was transformed to an ordinary differential equation. They studied the effect of
unsteady inflow condition on the efficiency of clarifiers. They concluded that a strong density
induced bottom current increased the recirculation in the tank when high inlet concentrations
were experienced. Taken over a period of a day, the unsteadiness increased the effluent
concentrations if compared to the steady state flow.
Krebs (1991) modelled a final rectangular settling tank by approximating turbulent flow with
a constant turbulent viscosity. He used PHOENICS to set up a model in which the velocity
distribution and sludge concentration in the tank were at steady state conditions. He used inlet
baffles to reduce the intensity of density currents. They showed that an optimal slot width
exists at a densimetric Froude number of 1. It was also demonstrated that intermediate
transverse baffles could reduce the bottom current effect. It is, however, important to ensure
that the section between the baffle and the inlet be cleaned to prevent another density current
forming as a result of sludge accumulation.
Stamou (1991) applied a curvature-modified k-ε model to simulate the neutral density flow
in a sedimentation tank. The curvature modified turbulence model showed significant
-8-
improvement over the standard k-ε model based on the separation length. It was also found
that the RTD curve shape was independent of the turbulent Schmidt number. A phenomenon
previously confirmed by Stamou, et al., (1989).
Zhou and McCorquodale (1992a) showed that the effluent quality of full-size plants could
be accurately predicted. They also demonstrated the effectiveness of inlet baffles. In reaction
to their paper Hager and Ueberl (1993) recommended the inclusion of 3D effects to account
for horizontal swirl at inlets. They also suggested that the inlet baffle should be sized as close
as possible to a densimetric Froude number of one. This subsequently reduced the effluent of
the nearest overflow by 50%. Samstag, et al., (1992) reviewed some of the previous clarifier
modelling efforts and showed the ability of CFD to predict velocity and solid concentration
distributions.
Continuing with the same line of investigation, Bretsher et al. (1992) considered his use of
porous walls normal to the flow in rectangular sedimentation tanks. They assumed steady
state conditions, constant turbulent viscosity and the settling velocity as a function of the
sludge volume fraction. Pure water experiments as well as two-phase buoyancy conditions
were investigated. The walls were modelled with different porosity fractions, representing
equally spaced holes. This resulted in a more even velocity distribution, with the velocity at
the bottom of the tank reduced. They further partitioned the walls into several chambers and
they found that the sludge level difference was reduced in steps from one wall to the other.
Krebs et al., (1992) investigated the use of perforated baffles to improve tank performance.
McCorquodale and Zhou (1993) investigated the effect of various solids and hydraulic
loads on circular clarifier performance. It was found that clarifier performance is strongly
related to the densimetric Froude number and not so much the Reynolds number. It was also
found that for a given solids loading an optimum Fd exists that results in a minimum effluent
concentration. In this case it was found to be between 0.5 and 1.
Frey et al. (1993) used the VEST code to determine the flow pattern in a sedimentation tank.
The flow profiles were then used by the TRAPS code to determine particle tracks. The model
was used to determine tank efficiency and optimise the tank geometry. Good agreement
between the model and experimental results was reported. Unfortunately it was only used for
very low solid concentrations. Density effects were therefore excluded.
Dahl et al., (1994) compared model results with measurements on a lab-scale secondary
settling tank. They included an advanced formulation, describing the rheology of the settled
sludge near the tank bottom.
A Bingham plastic formulation related the shear stress, the yield strength, the plastic viscosity
and the velocity gradients near the tank bottom. Good agreement was obtained between
model and measured velocity profiles and suspended solids concentration (R2 ranged between
0.84 and 0.99).
Zhou et al., (1994) linked the energy equation with the Navier-Stokes equations to simulate
the effect of neutral density and warm water into a model clarifier. An algebraic version of
the k-ε model was used to accurately model the mixing of the warm and cold water.
Comparisons were made with the standard k-ε model and the k-ε model adapted for
buoyancy. The algebraic k-ε version showed improved prediction of velocity profiles
compared to the standard k-ε turbulence model, but only a marginal improvement compared
to the k-ε turbulence model adapted for buoyancy. Comparing the temperature profiles, the
algebraic k-ε model performs much better than the other two k-ε models.
Krebs, et al., (1995) used the PHOENICS code to model different inlet arrangements. They
evaluated the effect of inlet baffle position and depth, inlet dissipating devices such as angled
bars and deflection of inlet jets. This reiterated a previous recommendation that the inlet slot
width should be sized for a densimetric Froude number of unity. The presence of angled bars
showed insensitivity to the densimetric Froude number.
-9-
Matko et al. (1996) modelled circular sedimentation tanks using another computational fluid
dynamics program, CFDS-FLOW3D. They used tracer tests to compare their numerical
model with both full-scale and pilot-scale tanks with different geometries. They also
performed simulations on the effect of a vertical and horizontal inlet arrangements. Results
from their pilot-scale model gave good agreement with their experimental data of residence
times in terms of shape and tail of the curve. CFD simulations of the full-scale tank using a
steady value of the inflow did not agree with the experimental data, but agreed favourably
when using a variable inflow rate.
Ekama et al. (1997) compiled a manual that summarised much of the previous modelling
efforts by the past and present research groups. Empirical, theoretical, modelling and
practical aspects were discussed.
Marais, et al., (2000) compared a 1D model based on the idealized flux theory and a 2D
hydrodynamic model.
Lakehal et al. (1999) and Armbruster et al. (2001) presented a model for unsteady
simulation of circular clarifiers that included the sludge blanket in the computation domain. A
rheology function was included that accounted for the increased viscosity of highly
concentrated sludge mixtures. Stamou et al. (2000) applied a 2D mathematical model to the
design of double-deck secondary clarifier. They modelled each tank independently adjusting
the boundary conditions for the independent cases.
De Clercq (2003) presented an extensive study in SSTs that included calibration and
validation with both lab-scale and full-scale investigations. He implemented sub-models that
account for the rheology of the sludge, the Takacs solids settling velocity and the scraper
mechanism.
Kim et al. 2005 have recently performed a numerical simulation in a 2-D rectangular
coordinate system and an experimental study to figure out the flow characteristics and
concentration distribution of a large-scale rectangular final clarifier in wastewater treatment.
In recent years, CFD commercial programs have become fast and user-friendly and have been
widely used by engineers in many fields. Two of the most common CFD packages are
PHOENICS and FLUENT. Examples of PHOENICS applications can be found in Krebs
(1991), Dahl et al. (1994), Krebs et al. (1995), De Cock et al. (1999) and Brouckaert and
Buckley (1999). Laine et al. (1999), Jayanti and Narayanan (2004), Kris and Ghawi
(2006, 2007a, 2007b, 2007c, 2007d, 2007e and 2007f) used FLUENT for their simulation of
the 2D hydrodynamics of settling tanks.
2.4 Insight gained from past CFD efforts
Many lessons can be learned from past CFD efforts. The insights gained in terms of the
transport mechanisms include:
• The behaviour of a clarifier/sedimentation tank is influenced by the presence of
suspended solids. The modelling of the fluid transport can therefore not be separated
from the transport of the suspended solids. The fluid transports the suspended solids
while the fluid is also dragged along by the suspended solids.
• Clarifiers also experience turbulent flow. Accurate clarifier models therefore need to
incorporate turbulence models that can account for the turbulent fluid and solids
transport. The k-ε model has been the most popular approach despite its known
shortcomings.
The tank geometry has a profound influence on tank behaviour, such as:
• The inlet conditions and shape have been identified as the most important
geometrical aspect of a sedimentation tank. Inlets should be designed to introduce
as little energy as possible into the tank. Inlets those are much higher than the
floor of the sedimentation tank should be avoided, due to large potential energy
- 10 -
differences. Velocities should be damped at the inlet to avoid jets issuing into the
tank. This can be achieved by the introduction of inlet slots and baffles.
• The outlet of a tank should be positioned in the area of lowest solids
concentration. This is generally difficult to establish prior to modelling of the
tank.
• Internal baffles that damp the density currents can improve tank performance
(Krebs et al, (1992)). Zhou and McCorquodale (1993) warned that if the slot it
too small it can also lead to undesirable resuspension of solids.
• External effects such as the effect of wind and temperature can cause severe
short-circuiting.
Several discretisation approaches have been used. The finite volume method appears to be the
most popular approach for several reasons. It is much easier to discretize and it divides a flow
domain into a finite number of volumes (a concept that is fundamentally linked to a
continuum).
2.5 Classification of Process Tanks
One of the prime objectives of water treatment plants is to remove suspended solids from
natural occurring surface and ground water. At the heart of water treatment plant lays the
tanks that affect suspended solids removal. However, potable water cannot be produced
without processes that precede and follow suspended solids removal. The continued
deterioration of water sources forced the water industry to develop new treatment processes
and include a number of processes that go beyond conventional suspended solids removal. It
is for this reason that the meaning of the term process tank is extended to include all the
processes that take place in a defined space from the source to the tap. This includes tanks of
all shapes and sizes; deep tanks, small tanks, large tanks, long tanks, narrow tanks and many
more.
2.5.1 Phase Separation Tanks
2.5.1.1 Function and purpose
The purpose of phase separation tanks is to reduce the concentration of suspended solids. At
the core of most water treatment plants is a phase separating tank. The suspended solids can
be removed by the mechanisms of:
1. Sedimentation/Thickening/Clarification (solids and liquid)
2. Flotation (DAF) (solids, liquid and gas)
3. Filtration (solids, liquid and medium)
The force that drives sedimentation and flotation is the density differences between the water
and the dispersed second or third phases. Filtration is dominated by shear and attraction
forces. The movement of the liquid phase (hydrodynamics) plays a pivotal role in the
behaviour of a phase separating device. Characteristics of these tanks are complex
geometries, complex flow patterns, the presence of two or more phases and strong buoyant
and shear forces. Tanks vary from medium to long retention times (minutes to hours). Phase
separation tanks rely on convection, diffusion, sedimentation, flotation, buoyancy and
attrition mechanism. Examples of phase separating tanks include:
1 Settling tanks (horizontal settling tanks, vertical upflow tanks, clarifiers and
thickening tanks)
2 Dissolved Air Flotation (DAF)
3 Filtration (rapid sand filters, slow sand filters and membrane filters)
2.5.1.2 Performance and design
The performance indicators and design guidelines for phase separation tanks are more
complex as the mechanisms used in each tank type are different and the geometrical
- 11 -
configurations are diverse. Only a few of the performance indicators that are commonly used
are listed in Table 2.1.
Table 2.1 – Current phase separating tank guidelines.
Process Tank Performance indicator
Overflow rate
Sedimentation tank Froude number
Reynolds number
Overflow rate
Froude number
Thickness
Reynolds number
Phase separation tank
Underflow rate
Filtration rate
Sand filter
Backwash rate
Overflow rate
DAF Air concentration
Recycle rate
2.6 Types of settling tanks
Four types of settling tanks can be found in practice, i.e. circular, rectangular, lamella and
vertical flow configurations. Although square tanks are used on occasion, they are not as
efficient in solids removal as circular and rectangular tanks. Solids accumulate in the corners
of the square tanks and are swept over the weirs by the movement of the solids collector
mechanism (Tchobanoglous & Burton, 1991).
Figure 2.1: Rectangular settling tank with different solids removal mechanisms:
travelling-bridge (top) and chain-and-flight (bottom) type collectors
When rectangular (Figure 2.1), lamella (circular or rectangular) (Figure 2.2), the vertical flow
settling tank (Figure 2.3) and circular (Figure 2.4) settling tanks are properly designed, they
show the same removal efficiency and capacity (Parker et al., 2001). The design of circular
tanks is however preferable for a number of constructional and operational costs. Firstly, the
investment costs for large-volume circular tanks are lower than for rectangular tanks. It is
only when large combined units of rectangular settling tanks are being built that construction
costs are more or less the same (Parker et al., 2001). Circular tanks also offer the advantage
- 12 -
of allowing solids removal at the bottom by scraper blades, which is cheaper and safer to
operate and with a longer life than chain conveyers (Kalbskopf, 1970).
Figure 2.2: lamella settling tank

If the horizontal dimension of the settling tank is relatively small compared to the water depth
(ratio of horizontal to vertical distance less than 2), the flow can be assumed to be
predominantly vertical (Ekama et al., 1997). Therefore, this type of tank is called the vertical
flow settling tank. With such tank geometry and positioning the inlet structure in the solids
blanket results in a fluidisation of the sludge. The solids blanket then acts like a floc filter that
even removes very fine particles (Fuchs & Staudinger, 1999).
The performance of conventional settling tanks may be improved by the installation of tubes
or parallel plates to establish laminar flow. As a result, these tanks have a very small settling
distance; the solids settled in the tubes or on the plates slide out due to gravitation. The major
drawback of these systems is the tendency to clog because of biofouling.
If limited land area is available, tray or multiple-storey settling tanks may be an option
(Stamou et al., 2000). With this type of settling tanks, two tanks are built on top of each
other to save space. Although they can be operated in parallel.
In this dissertation rectangular sedimentation tanks have been used in Hrinova and
Holic water treatment plant. Inlet velocities to settling tanks are in the region of 0.1 m/s to
0.6 m/s. Slower influent velocities may cause sedimentation in the inlet feed system and
higher velocities may case the flocs to be broken up by high shear, this is known as floc
disruption. Settleable solids concentration in the influent is typically in the range 500 to 5000
mg/l. Solids densities are typically 1066 kg/m3 to 1300 kg/m3, although in this study a solids
density of 1066 kg/m3 was used in the simulation of the settling tank in Chapter 6.
Figure 2.3: Vertical flow settling tank (Dortmund type): side view (left), and plan view
(right)
- 13 -
Figure 2.4: Circular settling tank with scraper mechanism and flocculator (Ekama et
al., 1997)
2.7 Ideal sedimentation tank with a horizontal flow pattern

Sedimentation is a solid-liquid separation that utilizes gravity to remove suspended solids
used in water and wastewater treatment.
An understanding of the principles governing the various forms of sedimentation behavior is
essential to the effective design and operation of sedimentation tanks.
Type I
Type II
Type III
Type IV
Figure 2.5 Types of Sedimentation
2.7.1 Classification of settling behavior

Settling particles can settle according to four different regimes (Figure 2.5), basically
depending on the concentration and relative tendency of the particle to interact: 1) Type I
discrete particle, 2) Type II flocculent particles, 3) Type III hindered or zone, and 4) Type
IV compression. In Holic and Hrinova settling process is dominated by regimes 1 and 2. A
description of the four classes is presented elsewhere (e.g. Takacs et al, 1991; Ekama et al.,
1997) and won’t be repeated here. This review focuses on the equations that have been
previously presented to model one of the four regimes.
Type I and Type II (discrete particles in dilute suspension) was simulated, as it is the
applicable type for the operating conditions in rectangular sedimentation tanks for potable
water treatment.
- 14 -
Figure 2.6 Definition diagram for particle terminal settling velocity
2.7.1.1 Models Based on Discrete Particles Settling

A diagram for settling of an idealized spherical particle is shown below (Figure 2.6). The
settling of discrete particles, assuming no interaction with the neighing particles, can be
found by means of the classic laws of sedimentation of Newton and Stokes. Equating
Newton’s law for drag force to the gravitational force moving the particle, we get Equation
2.1.
4 ( ρ P − ρ l ) gd P
VsP =
3 ρl CD ………………………………………………………………….………..2.1
where Vsp is the terminal settling velocity of the primary particle; CD is the drag coefficient; g
is the acceleration due to gravity; ρp and ρl are the particle and liquid density respectively;
and dp is the diameter of the particle. The drag coefficient is a function of the Reynolds
number (Re) and the particle shape. For settling particles Re is defined as:
VsP d P
Re = ……………………………..……………………………………………………….…....….2.2
v
In practice, it is found that CD is a function of the Reynolds Number, Re, and, for spherical
particles, it can be represented by the following expressions
24
Re < 1, CD =
Re
24 3
1 < Re < 104, CD = + 1 + 0.34
Re (Re ) 2
103 < Re < 105, CD ≈ 0.4

Substituting the above expression for Re < 1 (laminar flow) in Equation 2.1, results in the
following equation, known as Stoke’s Law:
g (S s − 1) 2
Vsp = d P …………………………………………………………….…....…. 2.3
18 v
Where S is the particle specific gravity.
s
The general conclusion, that Vsp depends on a particular diameter, particle density and, under
some conditions, also on fluid viscosity and hence on temperature, is important in
- 15 -
understanding sedimentation behavior. Furthermore, in practical sedimentation tanks, the

terminal settling velocity is quickly reached, so, for non-flocculent particles and uniform fluid
flow the settling velocity is constant throughout the settling time. This fact can be usefully
applied to a study of settling in an ideal sedimentation tank to provide an important design
principle for sedimentation processes. Idealized representations of rectangular horizontal flow
are shown in Figure 2.7.
Figure 2.7 Sedimentation Basin Design (Ideal Basin)
The ideal rectangular horizontal flow sedimentation tank is considered divided into four
zones (Figure 2.7)
1. Inlet zone - in which momentum is dissipated and flow is established in a uniform
forward direction
2. Settling zone - where quiescent settling is assumed to occur as the water flows
towards the outlet
3. Outlet zone - in which the flow converges upwards to the decanting weirs or launders
4. Sludge zone - where settled material collects and is moved towards sludge hoppers for
withdrawal. It is assumed that once a particle reaches the sludge zone it is effectively
removed from the flow.
2.7.1.2 Ideal Sedimentation Design Assumptions
The critical particle in the settling zone of an ideal rectangular sedimentation tank, for design
purposes, will be one that enters at the top of the settling zone, at point H, and settles with a
velocity just sufficient to reach the sludge zone at the outlet end of the tank. The velocity
components of such a particle are V in the horizontal direction and Vsp, the terminal settling
velocity, in the vertical direction.
From the geometry of the tank it is apparent that the time required for the particle to settle, to,
is given by
H
to = = L/vs …………………………………………………………….…....…………..…2.4
Vsp
but, since Vs = Q/W.H, then Vs = Q/W.L, where Q is the rate of flow, and L, W and H are
the length, width and depth of the tank, respectively. Since the surface area of the tank, A, is
W.L, then
Vs = Q/A …………………………………………………………….…....…………………………2.5
According to this relationship, the slowest-settling particles, which could be expected to be

completely removed in an ideal sedimentation tank would have a settling velocity of Q/A.
- 16 -
Hence this parameter, which is called the surface loading rate or overflow rate, is a
fundamental parameter governing sedimentation tank performance.
This relationship also implies that sedimentation efficiency is independent of tank depth - a
condition that holds true only if the forward velocity is low enough to ensure that the settled
material is not scoured and re-suspended from the tank floor.
Design of sedimentation tanks for water treatment processes are often based on the surface
overflow rate of the tank. This design variable is predicated on the assumption of uniform
unidirectional flow through the tank. Dick (1982), though, showed that many full-scale
sedimentation tanks do not follow ideal flow behavior because suspended solids removal in a
ST was often not a function of the overflow rate. Because of uncertainties in the
hydrodynamics of STs, designers typically use safety factors to account for this nonideal flow
behavior (Abdel-Gawad and McCorquodale, 1984a).
The ideal sedimentation design assumptions

1.Homogeneous inlet zone (same particle size distribution at all depths)
2.Discrete particle (Type I) settling
3.Uniform horizontal flow in sedimentation zone
4.Outlet zone to transfer uniform flow to discharge flow
5.Particles are not re-suspended from sludge zone
6.All particles with settling velocity greater than the critical velocity settle to the sludge zone
Non-ideal flow behavior can be the result of the following (Wells, 1990):
1. Inlet and outlet geometry
2. Inflow jet turbulence
3. Dead zones in the tank
4. Resuspension of settled solids
5. Density currents caused by suspended solids and temperature differentials
2.8 Factors influencing the capacity and performance of settling tanks

The previous sections described the settling tank complexities. Not only is its capacity and
removal efficiency, also many internal processes play a role that is physical and chemical
origin, and hydraulics influences. However, a clear distinction is difficult to make.
2.8.1 Physical and chemical influences
These factors are all related to floc aggregation and floc breakup, i.e. flocculation and
deflocculation.
Flocculation consists of two discrete steps, i.e. transport and attachment of particles. Before
particles can collide they first have to be transported in order to approach each other; this is
achieved by local variations in fluid/particle velocities. Depending on the transport
mechanism three types of flocculation exist:
1. Perikinetic flocculation, i.e. aggregation by the random thermal Brownian motion of
particles
2. Orthokinetic flocculation, i.e. aggregation by imposed velocity gradients due to mixing
3. Differential sedimentation, i.e. aggregation by differences in settling velocities
2.8.2 Hydraulic influences
In this respect, Baud & Hager (2000) observed tornado vortices in the corners of rectangular
settling tanks. They were capable of scouring the top of the solids blanket and significantly
reduce the solids removal efficiency. The hydraulics of settling tanks therefore has a large
influence on the efficiency of the Water Treatment Plant (WTP).
- 17 -
Figure 2.8: Example of flow pattern in a settling tank (---: equal solids concentration in
mg/l) (Bretscher et al., 1992).
The first theory about the efficiency of settling tanks was developed by Hazen (1904) for
individual particle settling in a uniform flow. Anderson (1945) discovered that the flow is far
from uniform because of density stratification. The solids-loaded influent has a higher density
than the ambient water and, hence, plunges as a density jet to the bottom of the tank; this is
the so-called density current (Figure 2.8). As a result, a secondary counter-current is induced
at the surface; even a three- or four-layered structure in the flow field can be experimentally
observed (Larsen, 1977; Van Marle & Kranenburg, 1994). The density current is
characterised by high velocities and appears in the vicinity of the solids blanket (Kinnear &
Deines, 2001). Therefore, settled solids may resuspend with increasing flow rates and can be
transported to the effluent weirs; consequently, the effluent quality deteriorates. However,
van Marle & Kranenburg (1994) indicated that density currents also might be favourable
for clarification. The layered flow pattern in the settling tank may result in high solids
removal efficiencies, since it tends to reduce short circuiting. If short circuiting is prevented,
the hydraulic detention time in the layered flow (with alternating flow directions) will equal
that of uniform flow. The latter is beneficial for the settling tank performance (Hazen, 1904).
Krebs et al. (1999) came to similar conclusions; however, when strong density effects
prevail, as is usually the case in settling tanks, increased tank length are recommended. In
addition, experiments of Konicek & Burdych (1988) show flocculation in the density
current, being beneficial for the solids removal efficiency.
2.8.3 Solids removal mechanisms
The solids removal mechanism is directly related to the type of settling tank. According to
Günthert (1984), this device has two functions: (i) collection of settled and thickened sludge,
and (ii) solids transportation from the point of settling to the hopper. Additionally, short
solids retention times, small disturbance in solids settling, and a quick conveyance at a
maximum solids concentration are desired. The type of removal mechanism depends more or
less on the type of settling tank.
Two different sludge collector systems are commonly used in rectangular settling tanks: (i)
traveling-bridge and (ii) chain-and-flight type collectors (Figure 2.1). For the travelling
bridge type the mechanism with mounted scrapers travels up and down the tank on wheels or
on rails supported on the sidewalls. The chain-and-flight consists of a pair of looped
conveyor chains, at which solids removal blades from wood or fibreglass are mounted.
Hence, the solids are scraped from the bottom and transported to the hopper.
How the scraper works, is still under debate. Two theories are commonly accepted. Firstly,
the scraper is considered from a mechanical perspective, i.e. it pushes the solids to the hopper
(Narayanan et al., 2000). Secondly, the scrapers are not truly conveying the solids, but are
merely resuspending it (Kinnear & Deines, 2001). During a specialised European COST
- 18 -
meeting on settling tanks (COST Action 624, 14-15 November 2002, Prague), the mechanism
of the scraper action was recognised as one of the issues still to be solved.
2.8.4 Weir loading and sedimentation tank inlet
Due to the different nature of the fluid introduced into a tank and the fluid withdrawn from a
tank, the inlet and outlet should be treated differently. The hole size in a weir has only a local
effect and will not affect the flow patterns in a sedimentation tank significantly. (Weir detail
may be important for practical reasons such as scum removal.) Weir positioning is more
important than the velocities through the weir holes. Low weir velocities will not prevent
short-circuiting, but the correct positioning of the weir can. The correct position depends on a
number of factors and cannot be isolated from the hydrodynamic behaviour of the rest of the
tank.
The overemphasis of detail relating to weir loading should not lead to a neglect of detail
relating to the tank inlet. In some cases the advice given will not have the desired effect.
This statement is criticized for the following reasons:
¾ A spreading flow will certainly improve the distribution of flocculated water at the inlet,
but this will not aid sedimentation as such. It is also unlikely that a spreading inlet will
promote plug flow. It depends on how the flocculated water is introduced. If it is not
introduced close to the tank bottom it can result in a strong density current, causing a
short-circuit, not plug flow.
¾ Reduced outlet velocities will not limit short-circuiting as the outlet velocities are not the
main cause of short-circuiting. The inlet conditions, resulting density currents and poorly
positioned overflows are often the cause for sedimentation tank short-circuiting.
2.8.5 Particle Size
The size and type of particles to be removed have a significant effect on the operation of the
sedimentation tank. Because of their density, sand or silt can be removed very easily. The
velocity of the water-flow channel can be slowed to less than one foot per second, and most
of the gravel and grit will be removed by simple gravitational forces. In contrast, colloidal
material, small particles that stay in suspension and make the water seem cloudy, will not
settle until the material is coagulated and flocculated by the addition of a chemical, such as an
iron salt or aluminum sulphate.
The shape of the particle also affects its settling characteristics. A round particle, for
example, will settle much more readily than a particle that has ragged or irregular edges.
All particles tend to have a slight electrical charge. Particles with the same charge tend to
repel each other. This repelling action keeps the particles from congregating into flocs and
settling.
2.8.6 Temperature effect
Recently, Computational fluid dynamics (CFD) software has become easy to use, fast and
user-friendly. This new generation software offers an inexpensive means of testing and
optimizing hydraulic operation of both existing constructions and those under design.
The effects of temperature on settling velocities and sedimentation in general have been
largely recognized and debated. Hazen (1904) suggested that particles settle faster as the
water becomes warmer. He stated that “a given sedimentation basin will do twice as much
work in summer as in winter.” This is maybe a bold statement, but the influence of
temperature differentials in the settling tank performance have been demonstrated by several
researches. In this respect, Wells and LaLiberte (1998) suggested that in the presence of
temperature gradients in the settling tanks, such as during periods of winter cooling, the
temperature effects are important and should be included in the modelling of the settling tank.
The atmospheric cooling process was earlier studied by Larsen (1977); he suggested that the
cooled-denser water sinks and is replaced by rising warmer water. He also suggested the
removal efficiency of a tank may vary over the year with a minimum during the winter
- 19 -
season when cooling rates are at a maximum. Similar effects were observed by Kinnear
(2004). Kinnear (2004) found that the excess effluent suspended solids increases as the air
temperature decreases.
McCorquodale (1977) showed that a diurnal variation in the influent of the order of ± 0.2°C
may produce short circuiting in sedimentation tanks. Larsen (1977) and McCorquodale
(1987) showed that the direction of the density current in settling tank may be defined by the
difference between the inflow and ambient fluid temperature. A cooler influent produces a
bottom density current, while in the cases of a warmer influent the density current is along
the surface. Studies done by Zhou et al. (1994), and Wells and LaLiberte (1998) support
these findings. Wells and LaLiberte (1998) found that temperature differences affect the
hydrodynamic of settling tank.
Another temperature-effect to take into consideration is probably the direct effect on the
settling properties of the sludge. Surucu and Cetin (1990) suggested that the zone settling
velocity decreases as the temperature of the tank increases.
2.9 Conclusions
This chapter focused on a general introduction to water treatment and the settling tank as
process unit. As demonstrated in this chapter, the settling tank clearly plays an important role
in the treatment of water. Physical and chemical influences may also alter the settling tank
operation. These factors are linked to the flocculation of sludge and the hydraulic pattern in
the settling tank. Large and dense flocs obviously settle faster and increase the removal
efficiency of the tank. Further, solids sedimentation can interact with internal flows; too high
liquid velocities may scour solids from the solids blanket interface.
To minimise the adverse effects on the removal performance, many systems have been
proposed in literature. Retrofitting and new settling tank designs try to cope with the static
and dynamic system disturbances. In this respect we applied Computational Fluid Dynamics
(CFD) modelling to account for hydraulic influences on the solids removal. This approach
incorporates the two-dimensional and/or three-dimensional geometry of the settling tank. By
modelling the internal physical processes it increases our system understanding. If it is well
validated, the CFD model calculations may be even regarded as numerical experiments that
can replace the costly field experiments. However, for now this is only possible for very
simple flows; settling tanks still demand a lot of research.
This dissertation has to be seen as an effort to optimise CFD models for settling tanks by
developing submodels for, e.g., iron concentration, temperature effect and scraper
mechanism. In this respect, the next chapter will introduce modelling techniques for settling
tanks, with special emphasis on CFD.
- 20 -
Chapter 3
Mathematical Modelling of Settling Tanks
3.1 Introduction
This chapter reviews the basics of settling tank modelling. Settling tank models are tools used
to represent the physical and chemical processes in the real system. Engineers mostly
consider mathematical and physical models. The latter models are scale models where the
real process is mimicked on a small scale; however, true similarity cannot be achieved
because some scale-effects occur in settling tanks (Ekama et al., 1997).
The simplest mathematical models consist of mathematical relations between known inputs
and outputs. These statistical and empirical models are sometimes called black box because
they provide little insight into the process physics. This type of models is therefore restricted
in use within the boundaries of calibration. In their most complex form, however,
mathematical models can describe the important processes occurring in the system by solving
the partial-differential equations of continuity, momentum, energy, and transport of dispersed
solids; in addition, realistic boundary conditions must be provided. For that reason, these
Computational Fluid Dynamics (CFD) models can be called deterministic or glass box
models. They reveal the role of fundamental laws on the process performance. Because of
their origin deterministic models can be applied beyond their range of validity, albeit with
caution.
Mathematical models also can be classified by their spatial resolution. There are simple
models that do not consider any spatial variability of certain state variables; i.e. an identical
state is assumed for every point in space. Besides these simple zero-dimensional (0D)
models, there also exist very complex three-dimensional (3D) models that account for the
state variability in space. In addition, models can simulate steady-state or unsteady conditions
in the system.
In a real plant there are many factors influencing the performance and the capacity of the
settling tank; many boundary and flow conditions cannot be reflected in the 1D model. There
are four categories of unconsidered influences,
1. Geometry, e.g. shape of the basin, inlet and outlet arrangements, and baffles
2. Flow, e.g. density effects causing non-uniform velocity profiles. This may result in
short-circuits from the inlet to the outlet, resuspension of settled flocs and turbulence
3. Solids removal mechanism, which results in many unsteady effects
4. Environmental, e.g. wind shear, air and inlet water temperature
The prediction of the settling tank performance is therefore a matter of calibration. Due to the
above-mentioned influences it is highly questionable whether 1D models will ever be able to
predict the dynamics of effluent quality. Also, different internal structures, e.g. baffles,
cannot be investigated by means of these models. Hence, more advanced models are
needed.
2D and 3D models have the potential to describe the internal flow pattern and the appropriate
solids and solutes transport phenomena. Their application is mainly related to the evaluation
of internal structural changes, e.g. baffles, on the settling tank efficiency. This fundamental
modelling also gives insight in the process physics. Since they are based on fundamental
conservation laws, 2D and 3D models can in principle be employed outside their range of
calibration. However, care has to be taken. A disadvantage of the models is that their use is
- 21 -
computationally very demanding and, therefore, they still cannot be used for control
purposes. Instead, they reveal the internal flow pattern of the system; this information can be
used to generate the proper model structure that requires less computation time.
3.2 Design of settling tanks
The insight gained from past CFD simulations raises a number of questions regarding the
conventional design process and guidelines. The basis for the conventional approach was
already laid at the turn of the century. Since then many researchers have discovered ‘holes’ in
this idealised and over-simplified approach. With the aid of CFD techniques, some of the
shortcomings inherent to the idealised approach are brought to the surface. Typical design
guidelines are listed in Table 3.1.
Table 3.1: Horizontal sedimentation tank guidelines ([1] Van Duuren, 1997, [2]
Kawamura, 2000)
Typical design guidelines for horizontal sedimentation tanks
Aspect Guideline Source
Inlet velocity <1m/s [1]
0.5-1.5 m/h [1]
Overflow rate
0.83-2.5 m/h [2]
<72 m/h [1]
Longitudinal velocity
18-66 m/h (mean flow) [2]
<10 m3/h.m [1]
Weir overflow rate
9-13 m3/h.m [2]
Solids loading 1-1.3 kg/m2.d [1]
>8 h [1]
Retention time
1.5-4 h [2]
Number of tanks 2 [1], [2]
Length <100 m [1]
>3:1 [1]
Length/width ratio
>4:1(6:1) [2]
<5m [1]
Water depth
3-5 m [2]
Width 3 m (in case of longitudinal baffles) [2]
Width/water depth ratio 3:1-6:1 [2]
Water depth/length ratio >1:15 [2]
Floor slope 1:60 [1]
Sludge drain > 1.4 m/s [1]
Desludging water loss 3% [1]
Re <20 000 [2]
Fr >10-5 [2]
Effluent turbidity 5-7 NTU [2]
It is evident from Table 3.1 that:

• All guidelines are based on average values across the tank.
• A horizontal sedimentation tank is designed primarily by considering the average
vertical fluid velocity.
• No reference is made to the density affected hydraulic nature of the tank.
• No guidance is given on the inlet of a sedimentation tank. On the other hand, the
weir loading (of lesser importance) is specified in detail.
A few design guidelines that are often used are discussed below.
3.2.1 Overflow rate
This common design parameter is often used to size settling tanks and gives the impression
that water flows vertically upwards and solids vertically downwards. This sentiment is
reflected by a recent publication on design guidelines for water treatment plants.
- 22 -
“It should again be noted that while the particles settle downwards, the clarified water flows
upwards” (van Duuren, 1997)
Water flows primarily in a horizontal direction. The overflow rate can at the most indicate
the tank surface area in relation to the flow-through rate. In addition:
¾ The water is only withdrawn from the weirs. This is only from a small portion of the
tank and not from the entire surface area. Using the entire sedimentation tank area in
this context is also misleading.
¾ The water does not rise uniformly from the tank bottom to the surface. At the same
time the solids do not sink to the bottom uniformly. In fact the water movement is
influenced by the solids and moves predominantly in a horizontal direction.
¾ The effect of stagnant zones, short circuiting and density currents are disregarded by
the overflow rate concept.
¾ The leading mechanism driving sedimentation is not the flow of the water through the
tank, but the density differences between the water and the solids. Two tanks with
the same overflow rate, but with different inlet suspended solids concentration will
behave entirely differently.
¾ The wide range of tank overflow rates used in the industry (Table 3.2) is not
surprising as the overflow rate is an unrealistic design parameter that disregards
many of the aspects at the core of sedimentation tank performance.
The overflow rate is therefore an unrealistic design parameter, and will reveal by the
present CFD analysis.
Table 3.2: Horizontal sedimentation tank overflow rates (Van Duuren, 1997).
Horizontal sedimentation tank overflow rates (m/h)
Germany USA Belgium South Africa Slovakia
0.5-4 0.8-12 1.8-5.4 1-2.8 0.5-4
3.2.2 Residence time
The residence time is a design guideline related to the overflow rate, that is based on ideal
flow patterns and does not relate to the actual residence time. Actual residence times in
sedimentation tanks are always different from idealised residence times. The presence of
density currents will result in deviations from idealised plug flow. The residence time is
entirely dependent on the hydrodynamic behaviour of the tank and cannot be reduced to a
single number based on flow rate and tank volume. A high residence time T will not
necessarily result in a good removal efficiency. If the inlet conditions are not favourable the
removal efficiency will be poor despite the long residence time. The insight of the present
CFD analysis will suggests differences between the idealized residence time and the actual
tank removal efficiency.
3.2.3 Reynolds number (Re)
The use of non-dimensional parameters has the advantage that it can be used irrespective of
scale. At least this is the underlying assumption. Unfortunately this is not always the case.
The main reason is that the hydrodynamic behaviour of any tank is inseparably linked to its
geometry. Non-dimensional parameters therefore aim at generalising fluid behaviour, but in
the process, loose sight of the importance of tank geometry. Three such parameters are often
inappropriately used for this purpose: the Reynolds number, the Froude number and the
Hazen number. In addition, as sedimentation tanks are affected by shear, turbulence and
gravity forces, it can be expected that these aspects were included in the design guidelines.
However, this is not currently the case.
Consider for example the Reynolds number. The Reynolds number is given by:
ρ .U .H
R e
=
μ
…………..……………………………………………………………………………………3.1
- 23 -
Where U – The average horizontal velocity in the tank (L/t)

μ – The molecular (laminar) viscousity (F.t/L2)
H – A length dimension (tank depth or length) (L)
ρ - Fluid mixture density (M/L3)
The Re number relates the momentum forces to viscous forces. The inclusion of a length
parameter indicates the expected length scale of turbulence. This means that the flow regime
can change from laminar to turbulent by simply changing the length scale from a few
millimetres to tens of meters. Most water treatment process tanks experience turbulent flow
for velocities larger than 1 mm/s and dimensions larger than 10 m.
The use of the Reynolds number is therefore limited in the following way:
¾ The use of Re is limited to establish if turbulent flow can be expected or not.
¾ Re is based on the molecular viscosity. The area of the sedimentation tank that
experiences laminar flow and viscous dissipation is limited.
¾ The Reynolds number only indicates the range of eddy size that can be expected.
It does not give an indication of the hydrodynamic behaviour that is conducive to
settling.
¾ Using the Reynolds number based on average horizontal velocities can lead to
erroneous results. If the bottom current is used as a guide the Re number can
increase by a factor of 10.
¾ It has also been shown by McCorquodale and Zhou (1993) that the bottom
current is insensitive to Re.
¾ Stamou, Adams and Rodi (1989) used the turbulent viscosity μt (as calculated by
the k-ε model) to calculate the Re number for turbulent flow. It was found that
flow fields are independent of turbulent Ret.
A Re number based on laminar viscosity and average velocities (Eq.3.1) does not relate to a
real sedimentation tank.
3.2.4 Froude number (Fr)
The Froude number is given by:
U2
F r gR …………..………………………………………………………………………………………3.2
=
Where R – The wetted perimeter of the tank (L)

g – The gravitational acceleration (L/t2)
The Froude number relates the average tank velocity to the gravitational acceleration. The use
of Fr is limited for the following reasons:
¾ Due to the fact that the water surface in a sedimentation tank is essentially
horizontal, the Froude number is of lesser importance (Van Marle and
Kranenburg (1994)).
¾ Fr is calculated as the average horizontal velocity of the tank. This is a crude
approximation as the local velocities, especially in the bottom current of the tank,
can be orders of magnitude larger than the average velocity.
¾ Fr is only applicable to the ideal plug flow case and does not account for flow
with density effects.
The Froude number relates average values of parameters unrelated to the mechanisms driving
sedimentation tank behaviour. Although gravity plays a fundamental role in sedimentation,
the formulation used in (3.2) does not distinguish between the effect of gravity on the fluid
and the suspended solids. The Froude number for a neutral density tank and a tank with an
inlet concentration of 15 000 mg/l will be identical. Its use as a design guideline is therefore
limited.
- 24 -
3.2.5 Hazen number (Ha)

The Hazen number relates the ideal settling time to the ideal residence time. This formulation
dates back to Hazen in the early 1900s and is given by.
v .L
H a = Uset.H …………..……………………………………………………………………………………3.3
Where L – The tank length (L)
H – The tank depth (L)
vset Settling velocity of suspension (L/t)
The Hazen number is limited for the following reasons:

¾ It is based on average values across the entire surface area, although the water is
only withdrawn from the tank from a limited area.
¾ It assumes that particles settle uniformly along the length of the tank, with a
uniform vertical settling velocity.
¾ It is only applicable to plug flow. In reality the flow at the tank bottom will be
orders of magnitude higher than the average flow. This will result in very low
Hazen number at the tank bottom. The Hazen number for the surface return
current will be negative!
3.2.6 The densimetric Froude number (Fd)
The densimetric Froude number (Fd) needs special mention. The densimetric Froude number
relates the local convection to the local density differences and does not assume tank
averaged values. It is a more appropriate non-dimensional parameter, that relates the largest
forces present in a sedimentation tank.
u
F d = (Δρ / ρ ) .g.h …………..………………………………………………………………….…………3.4
Where u – The velocity in the tank (L/t)
h – The depth of the water (L)
p - the density of the mixture (M/L3)
The density difference is given by equation 3.5:
Δρ = c[( ρ p − ρ ) / ρ p ] …………..…………………………………………………………………………3.5
Where pp -The density of the dried solids (M/L3)
c – Solids concentration (M/L3)
The use of Fd has the following advantages:
¾ Relates the inlet velocity (kinetic energy) to the density differences (potential
energy). This gives an indication of the strength of the density ‘waterfall’ that can
be expected as well as the resulting bottom density current.
¾ It gives an indication of tank stability. If the ratio is less than unity the gravity
effects will dominate the momentum forces. Conversely the momentum forces
will dominate gravity forces.
¾ It is position specific and does not assume tank averaged values.
The square of the densimetric Froude number indicates the ratio of the local kinetic and
potential energy (Krebs, 1991). If this ratio given by equation (3.6) is larger than unity, it
means that more kinetic energy is present than potential energy. In the case where the ratio is
less than unity, it means that more energy is available as potential energy. In other words,
large density differences are present at a high level in the tank that can be converted to kinetic
energy. Such a condition can lead to strong density currents and needs to be carefully
dissipated to avoid short-circuiting.
- 25 -
2 ρ .u 2
F d
=
Δρ .g .h
…………..…………………………………………………………………………………3.6
3.2.7 Balance of forces

If one considers the magnitude of the forces present in a sedimentation tank, it is evident that
buoyancy forces dominate convection forces, which in turn dominate viscous forces. The use
of non-dimensional parameters should therefore also be considered in order of significance.
Firstly, the densimetric Froude number, then the Reynolds number and then the Froude
number.
3.3 Fluid Dynamics
Enquiry into the transport of water started many centuries ago. The Babylonians, Egyptians
and Romans constructed canals, aqueducts, cisterns, screw-pumps and water mills. In the
Middle Ages water and wind driven mills were constructed. The scientific enquiry into the
behaviour and characteristics of water movement brought about the science that is known
today as fluid mechanics, fluid dynamics, hydrodynamics and hydraulics. The current
mathematical formulation of the observed lawfulness for fluids has only been developed in
the last century. This was as a result of contributions on many diverse fields’ viz.
mathematics, physical sciences, and applied mathematics. The Navier-Stokes equations that
form the basis of fluid mechanics, were developed by Navier and Poisson (circa 1830s) and
later refined by de Saint Venant and Stokes (circa 1840s). The reason why it took more than a
century to apply these equations to real life problems, is probably due to the fact that there is
no exact solution of the Navier-Stokes equations. This led to the development of two
branches of fluid mechanics.
The hydraulics branch focused on the practical application of fluid science and developed
many empirical formulations of specific fluid transporting devices. The other branch focused
on the theoretical hydrodynamics of non-viscous flow. Prandtl showed empirically and
theoretically the importance of viscous forces that dominate the boundary layer.
In this dissertation CFD is applied to a settling tank. More background information about the
modelling of flows is given in the next sections. Due to its significant impact on the flow
field, turbulence will be considered as well. Because the settling tank aims at a gravitational
separation of solids from the liquid, transport of solids deserves more attention. To conclude
this chapter, an overview of different boundary conditions as applied in literature will be
given.
3.4 Processes in Settling Tanks
3.4.1 Flow in Settling Tanks
Since the initial theory of settling in an ideal basin presented by Hazen (1904), many
researchers have made contributions to a better understanding of the flow processes in a
settling tank. Camp (1945) identified that the hydrodynamic presented in a real tank deviates
from the ideal presented by Hazen due to four major reasons: (1) flocculation process in the
ST, (2) retarding in settling due to turbulence, (3) the fact that some of the fluid passes
through the tank in less time than the residence time (short-circuiting), and (4) the existence
of density currents in the ST. Camp (1945) stated that “short-circuiting” is exhibited by all
tanks and is due to differences in the velocities and lengths of stream paths and it is
accentuated by density currents. Camp defined density currents as a flow of fluid into a
relatively quiet fluid having a different density, and identified that the differences in density
may be caused by differences in temperature, salt content, or suspended matter content.
Larsen (1977) divided the settling tank into four zones and identified some of the processes
occurring in each one of these: (1) the inlet zone, a part of the tank in which the flow pattern
and solids distribution is directly influenced by the energy of the influent. Mixing and
- 26 -
entrainment are important features in this zone. (2) The settling zone, in which Larsen
described two currents, a bottom current and a return current separated by a nearly horizontal
interface. (3) The sludge zone, located at the bottom of the tank containing settled material
which moves horizontally. (4) The effluent zone, which is the part of the settling tank in
which the flow is governed by the effluent weirs. Figure 3.1 shows the zones and the flow
pattern suggested by Larsen (1977) for rectangular settling tanks.
Figure 3.1 Flow Processes in a Rectangular ST (Larsen, 1997)
Larsen also identified that the flow in settling tanks is maintained and affected by major
energy fluxes:
1) Kinetic energy (KE) associated with the inlet flow.
2) Potential energy (PE) associated with influent suspension having a higher density than
the ambient suspension.
3) Wind shear at the free surface transferring energy to the basin.
4) Surface heat exchange that in the case of atmospheric cooling may produce water
with higher density and therefore supply a source of potential energy.
5) Energy flux associated with water surface slope.
6) Energy losses due to internal friction and settling.
In the matter of energy fluxes affecting the flow in settling tanks, Larsen presented the
following conclusions: (1) The KE is mainly dissipated in the inlet zone, and in addition to
defining the flow pattern in this zone, the influent is diluted by entrainment. (2) The PE of
Suspended Solids (SS) is partly dissipated at the inlet and partly converted to KE through the
density current which forms a flow along the bottom of the basin. The flow rate of the bottom
currents is higher than the inlet flow rate due to the additional flow supply by a counter-flow
in the upper layer (caused by the density current). (3) Gravity adds a small amount of energy
to this flow. (4) The energy leaving the system, kinetic energy of the outflow and potential
energy of the SS leaving the tank, is negligible as a component of the settling tank. (5) Wind
shear and heat exchange may be of significance. These energy contributions affect mainly the
upper layers where turbulence mixing may be enhanced. (6) All the energy inputs cause
turbulence, which greatly affect the flow field and concentration distributions in the settling
tank. Thus, the amount of SS in the effluent may depend on these energy inputs.
3.5. Modeling Equations
The hydrodynamic and solids stratification of settling tanks have been successfully described
by application of the following governing equations and conservation laws:
a) Continuity equation (conservation of fluid mass).
b) Fluid momentum equations (conservation of momentum).
c) Mass transport equation, including the modeling of the settling behavior of the
particles (conservation of particulate mass).
- 27 -
d) Energy equations (conservation of energy).

e) Turbulence modeling equations.
Continuity, momentum and mass transport equations have been used in all 2D and 3D models
to describe the flow pattern in STs. Few modifications have been introduced in these
equations since the original work of Larsen (1977), except in the treatment of the settling
velocities [major modifications in the differential equations are presented in the work of
Wells and LaLiberte (1998). On the other hand, different turbulence models have been
proposed and used with different levels of success, and few models have included energy
considerations.
The following conservation equations can be used to describe two-dimensional, unsteady,
turbulent, and density stratified flow in a settling tank using either rectangular or cylindrical
co-ordinates:
The following sections will discuss the governing equations of continuum mechanics (fluid
mechanics), turbulence modelling and the numerical techniques applied to solve these
equations.
3.5.1 Conservation of mass
The first term of equation 3.7 represents the rate of change of the density in the control
volume and the second term represents the rate of mass flux passing through the control
volume surfaces per unit volume
∂ρ ∂
+ ( ρu j ) = 0 .
∂t ∂x j …………..…………………………………………………………………………3.7
where ρ - fluid density

uj - fluid velocity vector in the j direction
In cases where there is more than one species, it is possible to specify the mass fraction of the
different species to make up the total mass of a control volume as described by equation 3.8.
∂( ρci ) ∂
+ ( ρu j ci ) = 0 …………..…………………………………………………………………3.8
∂t ∂x j
where ci - mass fraction of species i
3.5.2 Conservation of momentum
The conservation of momentum, generally referred to as Newton’s second law, was first
proposed by Descartes. In its control volume form it can be presented by equation 3.9. The
conservation of momentum equation balances the forces acting on the control volume with
the momentum change of the control volume.
∂ ∂
( ρu i ) + ( ρui u j ) = ρf + ∇. ∏ ij …………..………………………………………………3.9
∂t ∂x j b
Where Π ij - stress tensor representing normal and shear stresses
fb - body force per unit mass
The first term represents the rate of change of momentum per unit volume in the control
volume, the second term represents the rate of change of momentum lost by convection
through the control volume surfaces per unit volume. The first term on the right-hand side
represents the body forces acting on the control volume per unit volume. The second term on
the right-hand represents the surface forces per unit volume. The surface forces are applied
by external stresses and consist of normal and shear stresses.
3.5.3 Conservation of energy
Based on the First Law of Thermodynamics the energy equation is given below (3.10). The
first term on the left represents the rate of increase of total energy per unit volume. The
- 28 -
second term on the left represents the total energy lost per unit volume through the control
volume surface. The first term on the right represents the rate of heat produced per unit
volume. The second term on the right is the heat lost by conduction per unit volume. The
third term represents the work done by body forces on the control volume. The last term
represents the work done on the control volume by surface forces (Anderson et al., 1984).
∂E ∂ ∂Q
+ ∇.( Eui ) = − ui .q + ρf .ui + ∇.(∏ ij .ui ) …………..…………………………3.10
∂t ∂xi ∂t
where E - total energy = ρ (internal energy + kinetic energy + potential energy)
q - heat flux through control volume surfaces
Q - heat generated in a control volume
3.6 Constitutive relations
The constitutive relationships apply the governing equations to specific types of fluids. Only
a selection of these constitutive relations is discussed below.
3.6.1 Newton’s generalised law of viscosity
For fluids exhibiting a Newtonian character, a linear relationship exists between the fluid
stress and fluid strain, expressed as follows (Anderson et al., 1984).
⎡⎛ ∂u ∂u ⎞⎤
⎢⎜ ∂u j 2 ⎟⎥
∏ ij = − ρ δ ij + μ ⎢⎜ i + − δ ij k ⎟⎥
⎢⎜ ∂x j ∂x
i
3 ∂x
k ⎟⎥
⎣⎝ ⎠⎦
where δij - Kronecker delta function (δij = 1 if i=j , δij = 0 if i≠j)
The first term on the right-hand side represents the stresses due to pressure exerted on the
control volume. The second term (inside the square brackets) represents the viscous shear
and normal stresses on the control volume.
3.7 Source terms
A number of special source terms can be linked to the conservation laws to account for
buoyancy and chemical reactions.
3.7.1 Buoyancy
Buoyancy is added as a body force source term in the momentum equation (see equation 3.9)
to account for the gravitational force that acts on suspended matter in water. One way to
account for this force is to use the difference between the clear liquid density and the density
of the dried solids. This body force term ( fb ) is only applicable in the vertical direction:
f b = ( ρ p − ρ f ).g
where g - gravitational acceleration
3.8 Governing equations for Newtonian fluids
Substitution of Newton’s law of viscosity in the conservation of momentum equation, results
in a set of equations describing the behaviour of laminar fluid flow. These equations are
commonly known as the Navier-Stokes equations and form the basis of CFD simulations.
3.8.1 Navier-Stokes equations
⎡ ⎛ ∂ u ⎞⎟ ∂ u ⎤⎥
∂ ∂ ∂P ∂ ⎢ ⎜ ∂ui j 2 k + ρf
(ρui ) + (ρuiu j ) = − + μ⎜ + ⎟ − δ ij
∂t ∂x j ∂x ∂ x j ⎢⎢ ⎜ ∂ x ∂x ⎟ 3 ∂x ⎥
⎥ b ........……3.11
i ⎣ ⎝ j i ⎠ k ⎦
The first term represents the rate of change of momentum, the second term represents the
rate of momentum lost by convection through control volume surfaces. The first term on the
right represents the pressure gradient forces acting on the control volume. The second term
on the right represents the normal and shear stresses acting on the control volume surfaces.
- 29 -
The last term represents the body forces acting on the control volume. All terms are
expressed per unit volume.
3.8.2 Scalar transport equation
The convection and diffusion of dissolved and suspended substances can be modelled by a
convection-diffusion equation:
∂( ρci ) ∂ ∂ ∂c
+ ( ρui ci ) = ( Di i ) + Si………..………………………………………………3.12
∂t ∂xi ∂xi ∂xi
where Di - laminar diffusivity of species i.
3.8.3 Energy equation
Using Fourier’s law of diffusion for incompressible fluids the energy equation (3.10) can be
expressed as follows:
∂E ∂ ∂Q
+ ( Eui ) = + k∇ 2 .q + ρf .ui + ∇.(∏ ij .ui )…………..……………………………3.13
∂t ∂xi ∂t
3.9 Turbulence and turbulence modelling
The equations described above represent laminar fluid flow. In most water treatment
processes the fluid flow is, however, turbulent. The additional effort required to model
turbulent flow is significant. In fact to date it is not possible to describe the phenomenon of
turbulence in its full complexity and detail.
Before the turbulence modelling choices are described a few qualitative differences between
laminar and turbulent flow are highlighted.
3.9.1 Properties of turbulent flows
Some of the differences between laminar and turbulent flows according to Rodi (1993) and
various others include:
• Turbulent flows are inherently unsteady and three-dimensional, i.e. velocities,
pressure, temperature, etc. vary continuously in time and space.
• Turbulent flows are diffusive. The fluctuations of convection fluxes lead to strong
momentum, heat and mass transfer across the flow direction. Turbulent stresses are
much higher than viscous stresses causing higher diffusion rates and increased
“mixing”.
• Turbulent flows are dissipative. The kinetic energy of fluid is by virtue of turbulence
transformed into heat at a much faster rate than for laminar flows. Higher friction drag
and pressure drops occur.
• Turbulent fluctuations decay in a non-linear and anisotropic fashion close to walls.
• Turbulence is damped by free surfaces and stable stratification.
• Turbulent disturbances grow, become non-linear and interact with neighboring
disturbances. The mutual interactions lead to tangling and attenuation of eddies.
• Turbulence is affected by geometry and buoyancy.
Rodi (1993) also states that the large eddies are influenced by the boundary conditions of the
flow. The eddy size is of similar magnitude as the flow domain. The small eddies are
determined by viscous forces. The spectrum of eddy sizes increases with increasing Reynolds
number. Momentum is primarily transferred by the large-scale turbulent motion. It is
therefore important to model the large-scale motion with a turbulence model.
The large scale eddies interact with the mean flow in a significant way. Energy is abstracted
from the mean flow and fed into the large scale eddies. These eddies are stretched as vortex
elements and passed onto smaller and smaller eddies. The small scale eddies dissipate the
energy through viscous forces as heat. Rodi (1993) refers to this as ‘energy cascading’ as the
energy is cascading from the main flow to the large eddies and in turn to decreasing smaller
- 30 -
scale eddies until it is eventually dissipated as heat. The rate at which energy can be
dissipated is therefore dependent on the amount of energy that can initially be transferred to
the large scale eddies. In the case of buoyant forces (such as in a sedimentation tank or in a
dissolved air flotation tank) there is also an interaction between the potential energy of the
particle/bubble, the mean flow and the turbulent motion. Buoyant forces can enhance or
dampen turbulent motion.
3.9.2 Calculation methods for turbulent flows
It is generally not practical to solve turbulent flows numerically, in all its detail, as a huge
number of grid points and small time steps are required to resolve all the small space and
time scales of turbulent motion. In order to resolve the small-scale turbulent motion a mesh
with spacing smaller than the length scale of the smallest turbulent eddies, and a time step
smaller than the time scale of turbulent fluctuations have to be implemented.
Rodi (1993) estimated that a scale of the order 10-3 of the flow domain is required to model
the small-scale turbulence effectively. To resolve the turbulent flow of small-scale eddies,
effectively the scale of the numerical grid needs to be even smaller.
In the mean time other measures have been implemented to approximate turbulence in the
form of turbulence models. According to Anderson et al. (1984) a number of turbulence
model categories can be distinguished.
• Basic turbulence models use correlations such as the friction factor as a function of
Reynolds number or one or more ordinary differential equation. This method is
simple, but has a limited application.
• Turbulent or eddy viscosity models are based on equations obtained by averaging
procedures. The k-ε model is an example of such a turbulence model. This is probably
the most commonly used turbulence model as it balances economy of calculation,
accuracy and generality.
• Reynolds stress models or stress equation models also use averaging procedures, but
do not make use of the Boussinesq approximation to relate Reynolds stresses. The
Reynolds stresses are calculated directly.
• Direct numerical simulation (DNS) is the most exact way to model turbulence. The
unsteady three dimensional Navier-Stokes equations are solved without averaging or
approximation. However, the domain for which the computation is performed must be
at least the size of the largest turbulent eddy.
For engineering applications a balance is often struck between accuracy, economy and speed.
For this reason the k-ε model has found widespread application. The sections below discuss
the eddy viscosity turbulence model approach.
3.9.3 High Reynolds number k-ε model
The k-ε model calculates the turbulent viscosity in terms of two additional equations,
turbulent kinetic energy ( k ) and dissipation rate of turbulent kinetic energy ( ε ).
k 2 …........................………..…………………………………………………………………3.14
μt = ρC μ
ε
where Cμ - modelling constant The eddy viscosity model assumes that the effect of
turbulence can be represented as a form of increased viscosity. The effective viscosity is
represented by the sum of the laminar and the turbulent viscosity.
μ
eff
= μ + μt
The turbulent kinetic energy (k) is calculated as a partial differential equation:
- 31 -
⎛μ ⎛ ⎞⎞
∂ ∂ ∂ ⎜ eff ⎜ ∂ k ⎟ ⎟ 2 ∂ ui ⎡ ∂ ui ⎤
( ρk ) + ( ρu j k ) = ⎜ ⎜ ⎟⎟ − ⎢μ t ( ) + ρ k ⎥ + μ t ( Pk + Pb ) − ρε
∂t ∂x j ∂ x j ⎜⎜ σ ⎜ ∂ x ⎟ ⎟⎟ 3 ∂ x i ⎣ ∂ xi ⎦
⎝ k ⎝ j ⎠⎠
……………..………………………………….......................…............……………………………3.15
Where
∂u ⎛ ∂u i ∂u j
⎜
⎞
⎟
P = i ⎜ ∂x
+
⎟
k ∂x ⎝ ∂xi ⎠
j j
g ∂ρ
P =− i
b ρσ t ∂xi
The first term of equation 3.15 describes the rate of change of k, the second term the
convective transport. The first and second terms on the right describe the diffusive transport,
the third term production of k due to fluid shear and buoyancy and the fourth term
represents the viscous dissipation (Rodi, 1993).
The dissipation rate of turbulent kinetic energy (ε) is given by flowing equation
⎛ ⎛ ⎞⎞
∂ ∂ ∂ ⎜ μt ⎜ ∂ε ⎟⎟
( ρε ) + ( ρu j ε ) = ⎜ ⎜ ⎟⎟ +
∂t ∂x j ∂ x j ⎜⎜ σ ⎜∂ x j ⎟ ⎟⎟
⎝ ε ⎝ ⎠⎠
⎡ ⎞⎤
ε⎢ 2 ∂ u i ⎛⎜ ∂ u 2
i + ρ k ⎟ ⎥ − C ρ ε …...............……………………3.16
C μ
1k⎢ t k
( P + C P ) − μ
3 b 3 ∂ x ⎜ t∂ x ⎟⎥ 2
k
⎣⎢ i⎝ i ⎠ ⎦⎥
where σε - turbulent Schmidt number for ε
σk - turbulent Schmidt number for k
C1, C2, C3, Cμ - modelling constants listed in Table 3.3.
Table 3.3: Recommended typical values of the constants in the k-ε turbulence model.
Cμ [-] C1[-] C2 [-] C3 -] σε [-] σk [-]
0.09 1.44 1.93 0.8 1.3 1.0
The first term of equation 3.16 describes the rate of change of ε, the second term the
convection. The first term on the right describes the diffusion, the second the generation of
vorticity due to vortex stretching and the third term the viscous destruction of vorticity
(Rodi, 1993).
3.9.4 Wall functions
The flow close to walls is retarded by viscous forces. In order to accurately model the
boundary layer, significant grid refinements are required. An attempt to overcome this is to
assume a profile that approximates the boundary layer near the wall. This is done by
introducing a wall function. For more detail regarding wall functions refer to Rodi (1993).
For low Reynolds number turbulent flow, special adjustments are made to the wall functions.
3.10 Governing equations for turbulent Newtonian fluids
The combination of the Reynolds averaging procedure results in the following governing
equations for turbulent flow. The Navier-Stokes equations are given by (Rodi, 1993):
⎡ ⎛ ∂u ⎞⎟ ∂u ⎤⎥ ∂
∂ ∂ ∂P ∂ ⎢ ⎜ ∂ui j 2 k −
(ρui ) + (ρuiu j ) = − + ⎢μ ⎜ + ⎟ − δ ij μ ⎥ ∂x ((ρui u j ) + ρf b
∂t ∂x j ∂x ∂x ⎜
j ⎢⎣ ⎝ j∂x ∂x ⎟ 3 ∂x
i i⎠ k ⎥⎦ j
…………..……………………………………….....................................................…………………………3.17
- 32 -
Written in its divergence conservative form, for two-dimensional incompressible flow, the
Navier-Stokes equations are given as ( Zhou et al., 1992):
∂( ρu ) ∂( ρuu ) ∂( ρuv) ∂P ⎛⎜ ∂ ∂u ∂ ∂u ⎞⎟
+ + =− + (μ ) + (μ ) +
∂t ∂x ∂y ∂x ⎜⎝ ∂x eff ∂x ∂y eff ∂y ⎟⎠
⎛ ∂ ∂u ∂ ∂v ⎞⎟
+ ⎜⎜ (μ ) + (μ )
⎝ ∂x
eff ∂x ∂y eff ∂x ⎟⎠ …………..…………………………………………………3.18
∂( ρv) ∂( ρuv) ∂( ρvv) ∂P ⎛⎜ ∂ ∂v ∂ ∂v ⎞⎟

+ + =− + (μ ) + (μ ) +
∂t ∂x ∂y ∂y ⎜⎝ ∂x eff ∂x ∂y eff ∂y ⎟⎠
⎛ ∂ ∂u ∂ ∂v ⎞⎟
+ ⎜⎜ (μ ) + (μ ) ⎟ + f …………..………………………………………………3.19
∂x eff ∂ y ∂y eff ∂x b
⎝ ⎠
3.11 Representation of the Motion of Solids
The two-phase flow of solids and water in sedimentation can be treated as dispersed flow.
There are essentially two ways of calculating the flow field in dispersed flow. These are (1)
the Eulerian-Eulerian approach and (2) the Eulerian-Lagrangian approach.In the first
approach, both the phases are assumed to be so thoroughly mixed that they can be regarded
as interpenetrating continua. Thus, in any small volume within the flow domain (‘‘cell’’), a
fraction of the volume is occupied by one phase, the rest being occupied by the other phase (a
generalization to arbitrary number of phases is also possible). Conservation equations for
each phase are then written (Drew and Lahey 1979) introducing a new variable, phase
fraction, to represent the fact that only part of the volume is being occupied by that phase.
The equations for the two phases are coupled by the phase fraction and the interaction terms
representing the exchange of mass, momentum, and heat between the two phases. In the
Eulerian-Lagrangian approach, the motion of the two phases is calculated separately without
explicitly accounting for the volumetric fraction occupied by the other phase. The continuous
phase flow field is calculated first using the traditional Eulerian approach. The motion of the
particulate phase is then calculated using a Lagrangian approach in which the trajectories of
individual particles are calculated through the flow domain. Several such particle histories are
required to obtain a representative particle track. The continuous phase affects the particle
motion through drag, buoyancy, and other forces. The effect of the particulate phase on the
continuous phase is represented through source/sink terms appearing in the conservation
equations for the continuous phase.
Of the two approaches, the Eulerian-Eulerian approach is less time-consuming than the
Eulerian-Lagrangian approach which requires the generation of many particle trajectories to
correctly account for effects such as turbulent dispersion. However, the behavior of
individual particles or particles with specific characteristics can be evaluated more easily in
this approach and a better physical picture can be obtained. The principal disadvantage, apart
from the need for much larger computational resources, is that interparticle interaction cannot
be taken into account. Both effects lead to significant errors if the dispersed phase
(volumetric) fraction is higher than, say, 10–12%. Also since each trajectory calculation is
independent of others, the information of volume fraction at a given location is not readily
available. Hence it cannot be used to take account of the physical presence of other
phases/particles for phenomena such as agglomeration. Thus, the model cannot be used for
Type III settling conditions. The Eulerian-Eulerian approach should be used in such cases.
For the cases of sedimentation tank considered in the present study, the Eulerian-Lagrangian
approach is the natural choice to examine in detail the settling process and the influence of
- 33 -
the field on it. In the present study, the effect of the particulate phase on the fluid motion is
also neglected. Some justification for this will be provided later.
3.12 Boundary conditions
The Navier-Stokes equations are a system of elliptic partial differential equations. The
computational grid to which these equations are applied needs to include boundary conditions
in order to guarantee a unique solution.
These boundary values are specified on the external and internal surfaces of a computational
grid. Typical boundary conditions include:
Inlet boundaries specify the mass, momentum, turbulence levels, temperature and
density that enters a flow domain. The type of inlet conditions depends on the type of
flow that is considered.
Outlet boundaries are often specified as zero gradient boundaries. In special cases the
mass flow rate can also be specified at the outlet boundary.
A symmetry plane is specified where only a section of a domain is considered. This
type of boundary condition is only valid when symmetric conditions exist in the tank.
Wall surfaces are specified on all internal and external walls where the flow is
retarded by viscous forces. Special wall functions are used to approximate the flow
depending on the turbulent nature of the flow.
In water treatment process tanks free surface boundaries are very common. Free surfaces are
treated slightly different from symmetry plane boundaries (Gibson and Rodi (1989). Free
surface boundaries can also include the effect of wind on the free surface.
3.13 Conclusions
As presented in this chapter settling tanks can be modelled in different ways; each is
characterised by its own application possibilities and restrictions. Here, the technique of
computational fluid dynamics was highlighted because it is the topic of this dissertation. In
CFD studies, the settling tank is modelled in 2D and/or 3D. The method is able to illuminate
the internal hydraulics of the system. To this end, not only the Navier-Stokes equations are
needed, but also turbulence and solids transport have to be modelled. It was also recognised
that knowledge about the behaviour of the suspension is crucial. Besides these operational
parameters, no model can be solved without knowledge of the boundary conditions of the
system. A literature review on this topic was also covered in the last section of the chapter.
Finally, to solve the large set of partial differential equations, together with the constitutive
relations for sedimentation, a spatial discretisation of the settling tank is necessary. This will
be briefly discussed in the next chapter. Also the numerical algorithms to solve the equations
will be dealt with.
- 34 -
Chapter 4
Numerical Computations of Flows
The differential equations presented in chapter 3 describe the continuous movement of a fluid
in space and time. To be able to solve those equations numerically, all aspects of the process
need to be discretized, or changed from a continuous to a discontinuous formulation. For
example, the region where the fluid flows needs to be described by a series of connected
control volumes, or computational cells. The equations themselves need to be written in an
algebraic form. Advancement in time and space needs to be described by small, finite steps
rather than the infinitesimal steps. All of these processes are collectively referred to as
discretization. In this chapter, discretization of the domain, or grid generation, and
discretization of the equations are described. A section on solution methods is also included.
Figure 4.1 A finite number of small control volumes (cells) by a grid
4.1 Discretization of the domain: grid generation

To break the domain into a set of discrete sub-domains, or computational cells, or control
volumes, a grid is used. Also called a mesh, the grid can contain elements of many shapes
and sizes. In 2D domains, for example, the elements are usually either quadrilaterals or
triangles. A series of line segments (2D) connecting the boundaries of the domain are used to
generate the elements. Structured grids are always quadrilateral (2D) or hexahedral (3D), and
are such that every element has a unique address in I and J space, where I and J are indices
used to number the elements in each of the two computational directions (Figure 4.2). The I
and J directions can, but need not be aligned with the coordinate directions x and y.
In general, the density of cells in a computational grid needs to be fine enough to capture the
flow details, but not so fine that the overall number of cells in the domain is excessively
large, since problems described by large numbers of cells require more time to solve.
Nonuniform grids of any topology can be used to focus the grid density in regions where it is
needed and allow for expansion in other regions.
- 35 -
Figure 4.2 Simple 2D domain showing the cell centers and faces
4.2 Discretization of the equations

Several methods have been employed over the years to solve the Navier-Stokes equations
numerically, including the Finite Difference Method (FDM), Finite Element Method (FEM),
spectral element, and Finite Volume Method (FVM). The focus of this study is on the finite
volume method, which is described in detail below. Once the method and terminology have
been presented, the other methods will be briefly discussed (Section 4.1.3).
To illustrate the discretization of a typical transport equation using the finite volume
formulation (Patankar, 1980; Versteeg, 1995), a generalized scalar equation can be used
with the rectangular control volume shown in Figure 4.2. The scalar equation has the form:
∂( ρφ ) ∂ ∂ ∂φ
+ ( ρ u iφ ) = ( Di ) + Si ...……………………………………….(4.1)
∂t ∂xi ∂xi ∂xi
The parameter Di is used to represent the diffusion coefficient for the scalarφ. If is one of
the components of velocity, for example, Diwould represent the viscosity. All sources are
collected in the term Si. Again, if is one of the components of velocity, Si would be the sum
of the pressure gradient, the gravitational force, and any other additional forces that are
present. The control volume has a node, P, at its center where all problem variables are
stored. The transport equation describes the flow of the scalar into and out of the cell
through the cell faces. To keep track of the inflow and outflow, the four faces are labeled
with lower case letters representing the east, west, north, and south borders. The neighboring
cells also have nodes at their centers, and these are labeled with the capital letters E, W, N,
and S.
The first step in the discretization of the transport equation is integration over the control
volume. The volume integral can be converted to a surface integral by applying the
divergence theorem. Using a velocity in the positive x-direction, neglecting time-dependence,
and assuming that the faces e and w have area A, the integrated transport equation takes the
following form:
⎛ ⎡ dφ ⎤ ⎡ dφ ⎤ ⎞
( ρ eueφe − ρ wu wφ w ) A = ⎜ D ⎢ ⎥ − D ⎢ ⎥ ⎟ A + S ...……………………….(4.2)
⎜ e ⎣ dx ⎦ w ⎣ dx ⎦ ⎟
⎝ e w⎠
- 36 -
where S is the volume integral of the source terms contained in Si. This expression contains
four terms that are evaluated at the cell faces. To obtain the face values of these terms as a
function of values that are stored at the cell centers, a discretization scheme is required.
4.2.1 Discretization schemes

Since all of the problem variables are stored at the cell center, the face values (the
derivatives) need to be expressed in terms of cell center values. To do this, consider a steady-
state conservation equation in one dimension without any source terms:
d d ⎛ dφ ⎞
( ρuφ ) = ⎜⎜ D ⎟ ...…...……………………………………………………….(4.3)
dx dx ⎝ i dx ⎟⎠
This equation can be solved exactly. On a linear domain that extends from x = 0 to x = L,
corresponding to the locations of two adjacent cell nodes, with = 0 at x = 0 and = L
at x = L, the solution for at any intermediate location (such as the face) has the form:
⎧ x ⎫
exp⎨ Pe − 1⎬
φ = φ0 + (φ L − φ0 ) ⎩ L ⎭ .....................……………..……………………….(4.4)
exp( Pe − 1)
The Peclet number, Pe, appearing in this equation is the ratio of the influence of convection
to that of diffusion on the flow field.
ρuL .............................................................…………………………………….(4.5)
Pe =
Di
Depending on the value of the Peclet number, different limiting behavior exists for the
variation of between x = 0 and x = L. These limiting cases are discussed below, along with
some more rigorous discretization, or differencing schemes that are in popular use today.
Central differencing scheme
For Pe = 0, there is no convection, and the solution is purely diffusive. This would
correspond to heat transfer due to pure conduction, for example. In this case, the variable
varies linearly from cell center to cell center, so the value at the cell face can be found from
linear interpolation.
When linear interpolation is used in general, i.e when both convection and diffusion are
present, the discretization scheme is called central differencing. When used in this manner, as
a general purpose discretization scheme, it can lead to errors and loss of accuracy in the
solution. One way to reduce these errors is to use a refined grid, but the best way is to use
another differencing scheme. There is one exception to this rule.
Upwind differencing schemes
For Pe >>1, convection dominates, and the value at the cell face can be assumed to be
identical to the upstream, or upwind value, i.e. w = W. When the value at the upwind
node is used at the face, independent of the flow conditions, the process is called first order
upwind differencing. A modified version of first order upwind differencing makes use of
multi-dimensional gradients in the upstream variable, based on the upwind neighbor and its
neighbors. This scheme, which makes use of a Taylor series expansion to describe the
upwind gradients, is called second order upwind differencing. It offers greater accuracy
than the first order upwind method. It uses larger ‘stencil’ for 2nd order accuracy,
essential with tri/tet mesh or when flow is not aligned with grid; slower convergence
Power law differencing scheme
For intermediate values of the Peclet number, 0 < Pe < 10, the face value can be computed as
a function of the local Peclet number, as shown in Eq. (4.4). This expression can be
approximated by one that does not use exponentials, involving the Peclet number raised to an
- 37 -
integral power. It is from this approximate form that the power law differencing scheme
draws its name. This first order scheme is identical to the first order upwind differencing
scheme in the limit of strong convection, but offers slightly improved accuracy for the range
of Peclet numbers mentioned above.
QUICK differencing scheme
The QUICK differencing scheme (Leonard and Mokhtari, 1979) is similar to the second
order upwind differencing scheme, with modifications that restrict its use to quadrilateral or
hexahedral meshes. In addition to the value of the variable at the upwind cell center, the value
from the next upwind neighbor is also used. Along with the value at the node P, a quadratic
function is fitted to the variable at these three points and used to compute the face value. This
scheme can offer improvements over the second order upwind differencing scheme for some
flows with high swirl.
4.2.2 Final discretized equation
Once the face values have been computed using one of the above differencing schemes, terms
multiplying the unknown variable at each of the cell centers can be collected. Large
coefficients multiply each of these terms. These coefficients contain information that includes
the properties, local flow conditions, and results from previous iterations at each node. In
terms of these coefficients, Ai, the discretized equation has the following form for the simple
2D grid shown in Figure 4.2:
A pφ p = A φ + A φ + A φ + A φ = ∑ A φ .......………………………….(4.6)
N N S S E E W W i i
For a complex, or even a simple flow simulation, there will be one equation of this form for
each variable solved, in each cell in the domain. Furthermore, the equations are coupled,
since, for example, the solution of the momentum equations will impact the transport of every
other scalar quantity. It is the job of the solver to collectively solve all of these equations with
the most accuracy in the least amount of time.
4.2.3 Alternative numerical techniques
As mentioned earlier, other methods for solving the Navier-Stokes equations exist. The finite
difference, or Taylor series formulation replaces the derivatives in Eq. (4.1) with finite
differences between the variable storage sites (cell centers). The variation of the variable
between storage sites is ignored during the solution process. While this is an acceptable
method for some classes of flows, it is not the best choice for general purpose CFD analysis.
The finite element method uses piecewise linear or quadratic functions to describe the
variation of the variable f within a cell. This formulation is based on the method of weighted
residuals.
Weighting functions are chosen so that the coefficients of the linear or quadratic functions
result in the smallest residual (error) when f is substituted into the conservation equation. The
method is popular for use with structural analysis codes and some CFD codes. The spectral
element method is similar to the finite element method, only polynomials of higher order are
used to describe the variableφ. This method has been used in some specialty CFD codes.
4.3 Solution methods
The result of the discretization process is a finite set of coupled algebraic equations that need
to be solved simultaneously in every cell in the solution domain. For small problems, i.e.
those with fewer than 1000 elements, a matrix inversion can be done. Few problems can be
solved with adequate solution accuracy using such a small cell count, however, so alternative
methods are usually employed. Two iterative methods exist for this purpose. A segregated
solution approach is one where one variable at a time is solved throughout the entire
domain. Thus the x component of the velocity is solved on the entire domain, then the y-
component is solved, and so on. One iteration of the solution is complete only after each
variable has been solved in this manner. A coupled solution approach, on the other hand, is
- 38 -
one where all variables, or at a minimum, momentum and continuity, are solved
simultaneously in a single cell before the solver moves to the next cell, where the process is
repeated. The segregated solution approach is popular for incompressible flows with
complex physics, typical of those found in sedimentation applications.
Typically, the solution of a single equation in the segregated solver is carried out on a subset
of cells, using a Gauss-Seidel linear equation solver. In some cases, the solution time can be
improved (i.e. reduced) through the use of an algebraic multigrid correction scheme.
Independent of the method used, however, the equations must be solved over and over again
until the collective error reduces to a value that is below a pre-set minimum value. At this
point, the solution is considered converged, and the results are most meaningful. Converged
solutions should demonstrate overall balances in all computed variables, including mass,
momentum, heat, and species for example. Some of the terminology used to describe the
important aspects of the solution process is defined below.
4.3.1 Solution algorithms for Navier-Stokes equations
The solving of the Navier-Stokes equation is not a trivial exercise. It is not just a matter of
inserting the equations in a matrix solver in order to get an answer. The main problem arises
from its non-linearity and the fact that there is no direct relation between the velocity and the
pressure. Numerous methods have been devised to overcome this problem. Only the two
most common methods are discussed.
• The SIMPLE algorithm. Semi Implicit Pressure Linked Equations (SIMPLE)
algorithm decouples the pressure from the momentum equations during the solution
process. The pressure is then estimated based on the velocity profile and re-coupled
with the flow field. The iterative process subsequently refines the pressure correction
until convergence is satisfied for the complete set of equations. The algorithm was
first proposed by Caretto et al. (1972) and popularised by Patankar (1980). Various
refinements of the SIMPLE algorithm have been proposed such as the SIMPLER,
SIMPLEC which basically improves the method of pressure correction. In order to
avoid the solution from diverging the solution is underrelaxed. The SIMPLE
algorithm is mainly used for steady state flow.
• The PISO algorithm (Issa, 1986) also decouples the pressure and velocity fields, but
incorporates several intermediate steps to calculate the pressure. The predictor step
predicts what the velocities should be by solving the momentum equation. The
pressure field is then calculated based on the predicted velocities. The new pressure
field is used in the corrector step to recalculate the velocities. The PISO algorithm is
mainly used for unsteady flows. Solving of the equations that arise from CFD
computation grids require iterative methods due to the enormous size. This requires
that the convergence of the solution be checked on a regular basis and at some time a
decision needs to be made to stop the iterative process based on convergence criteria.
Numerous efforts have been made to improve the convergence speed of solvers.
Another criteria for obtaining a correct solution, is the stability of the convergence
process. Instability can be controlled by the solution algorithm, boundary values,
source terms and the particular solver used. Once a solution is reached it is important
to know the level of accuracy of the solution. Techniques exist to determine the errors
that arise during the discretisation and solution process, Jasak (1996).
With respect to settling tank modelling, the SIMPLE algorithm has been frequently used to
solve the unsteady and incompressible RANS equations (Zhou & McCorquodale, 1992).
Instead, Adams & Rodi (1990) and Szalai et al. (1994) applied the SIMPLEC algorithm. In
this dissertation, however, the PISO algorithm was found to be the most robust technique
avoiding possible occurring numerical divergences. Following the solution of the remaining
problem variables, the iteration is complete and the entire process repeated.
- 39 -
4.3.2 Residuals
If the algebraic form of a conservation equation in any control volume (Eq. (4.6)) could be
solved exactly, it would be written as:
A pφ p − ∑ Aiφi = 0 .............................................................……………………………(4.7)
Since the solution of each equation at any step in an iterative calculation is based on inexact
information, originating from initial guessed values and refined through repeated iterations,
the right hand side of the above equation is always non-zero. This non-zero value represents
the error, or residual in the solution of the equation in the control volume.
A φ − ∑ A φ = R p .............................................................…………………………….(4.8)
p p i i
The total residual is the sum over all cells in the computational domain of the residuals in
each cell.
∑ R p = R…….............................................................…………………………………….(4.9)
Since the total residual, R, defined in this manner, is dependent upon the magnitude of the
variable being solved, it is customary to either normalize or scale the total residual to gauge
its changing value during the solution process. While normalization and scaling can be done
in a number of ways, it is the change in the normalized or scaled residuals that is important in
evaluating the rate and level of convergence of the solution.
4.3.3 Convergence criteria
The convergence criteria are preset conditions for the (usually normalized or scaled) residuals
that determine when an iterative solution is converged. One convergence criterion might be
that the total normalized residual for the pressure equation drop below 1 x 10-3. Another
might be that the total scaled residual for a species equation drop below 1 x 10-6.
Alternatively, it could be that the sum of all normalized residuals drop below 1 x 10-4. For
any set of convergence criteria, the assumption is that the solution is no longer changing
when the condition is reached, and that there is an overall mass balance throughout the
domain. When additional scalars are being solved (heat and species, for example), there
should be overall balances in these scalars as well.
Whereas the convergence criteria indicate that overall balances probably exist, it is the wise
engineer who will examine reports to verify that indeed they do.
4.4 Numerical techniques used in Fluent
Whereas Sections 4.1 and 4.2 already mentioned different numerical techniques available in
Fluent, this section will shortly deal with the methods applied in this dissertation.
The Fluent software utilises the finite volume method to solve the governing integral
equations for the conservation of mass and momentum, and (when appropriate) for scalars
such as turbulence and solids concentration. In this dissertation, the so-called segregated
solver was applied; its solution procedure is schematically given in Figure 4.3. Using this
approach, the governing equations are solved sequentially, i.e. segregated from one another.
Because the governing equations are non-linear (and coupled), several iterations of the
solution loop must be performed before a converged solution is obtained.
Concerning the spatial discretisation, the segregated solution algorithm was selected. The
k-ε turbulence model was used to account for turbulence, since this model is meant to
describe better low Reynolds numbers flows such as the one inside our sedimentation tank.
The used discretisation schemes were the simple for the pressure, the PISO for the pressure-
velocity coupling and the second order upwind for the momentum, the turbulence energy
and the specific dissipation. Adams and Rodi 1990 pointed out that for real settling tanks the
walls can be considered as being smooth due the prevailing low velocities and the
correspondingly large viscous layer. Consequently, the standard wall functions as proposed
- 40 -
by Launder and Spalding 1974 were used. The water free surface was modeled as a fixed
surface; this plane of symmetry was characterized by zero normal gradients for all variables
Update properties, e.g. p
Solve numerical equations (u. v velocity)
Solve pressure-correction (continuity) equation

Update pressure, face mass flow rate
Solve turbulence and scalar equations
Converged?
No Yes
Update solution values with converged values at current time
Requested time steps completed?
Take a time step No Yes STOP
Figure 4.3 Solution Procedure

4.5 Conclusions
Many techniques exist to discretise the flow equations and subsequently solve the set of
algebraic or ordinary differential equations. It is emphasised that this dissertation does not
aim for the development of new numerical schemes to solve the flow equations. However, the
software used, Fluent, gives the user the opportunity to choose from a number of numerical
schemes. It was therefore opted to include a chapter providing a short overview of numerical
techniques introducing the reader into this large and complicated field of computational fluid
dynamics.
To solve time-dependent and time-independent flow problems, both the spatial domain and
the governing equations need to be discretised. Space is discretised by setting up a mesh or
grid, whereas the discretisation of the Partial Differential Equations (PDEs) results in a
system of algebraic equations. To perform the latter, three options are available to the
modeller, i.e. finite difference, finite element and finite volume methods. In this respect,
Fluent utilises the finite volume method, which is very effective on irregular meshes.
Although the approaches of the three techniques are different, the discretisation schemes used
are very similar.
The used discretisation schemes were the simple for the pressure, the PISO for the pressure-
velocity coupling and the second order upwind for the momentum, the turbulence energy and
the specific dissipation. Adams and Rodi 1990 pointed out that for real settling tanks the
walls can be considered as being smooth due the prevailing low velocities and the
correspondingly large viscous layer.
- 41 -
Chapter 5
Experimental Techniques for Model Calibration and
Validation
5.1 Introduction
Computational fluid dynamics belong to the category of new tools being used by the engineer
for analysis and design of systems. CFD simulations are widely used by academics, industry
and government to help study and understand basic flow physics. An underlying concern of
all CFD users is the issue of credibility of the results. These concerns arise because the
accuracy of the numerical solution remains uncertain despite the large knowledge base. This
results from inadequate calibration of the models and numerical techniques. Therefore, in
order to establish credibility for the CFD code verification and validation must be performed.
In performing verification and validation both the numerical accuracy and the physical
modelling capabilities are scrutinised.
The numerical accuracy is addressed in the verification process. During verification, errors in
the computational solutions are identified and quantified. They arise from the approximation
of the solution of the governing phenomenological equations.
The validation process identifies and assesses error and uncertainty in the conceptual and
computational models. This task involves the comparison of simulation results with actual
measurements. Experimental fluid dynamics play a dominant role in this portion of code
development and assessment. However, there is a major problem with the use of most
experimental data for validation purposes; it is usually incomplete. Indeed, CFD use of data
puts very stringent requirements on the experiments. These requirements include the need for
a complete set of physical modelling data and a qualification/quantification of the uncertainty
involved with the experimental measurements.
The process of developing (incl. calibration), verifying, and validating a CFD code requires
the use of experimental, theoretical and computational sciences. This process is a closed loop
as presented in Figure 5.1.
PHYSICAL
REALITY
CONCEPTUAL MODEL MODEL

MODEL CALIBRATION VALIDATION
CODE SIMULATED
VERIFICATION REALITY
Figure 5.1 Process of developing CFD code.

The above clearly indicates that good experimental data are indispensable for settling tank
model validation; their quality largely depends on the applied experimental technique.
5.2 Validation of CFD results
The verification of CFD software involves the checking of discretisation algorithms and
solvers to ensure that aspects such as discretisation and solving of equations are performed
- 42 -
correctly. The verification is in most cases performed by the developer of the CFD software.
The validation of the CFD results is left to the user of the CFD software. This involves
comparing the CFD model results with experimental measurements. This is, however, easier
said than done and can in some cases be as challenging (if not more challenging) than the
CFD simulations itself.
This in no way suggests that experimental evidence is not necessary, but it emphasises the
challenges that are involved in collecting experimental measurements to validate information
rich models.
Technological advances have enabled modellers to overcome some of the above-mentioned
challenges. But, the improved accuracy and detail of the flow field come at a healthy price
tag and additional compromises. Kuipers and Van Swaaij (1998) list only a few
compromises:
• A choice needs to be made between local and whole field measurements. Some local
field instruments can be used in full-scale tanks, but most whole field measurements
are limited to scale models.
• Intrusive methods penetrate deeper into the full-scale tank, but can influence the flow
patterns. Non intrusive methods do not influence the flow patterns, but requires a
transparent fluid to assess the flow field.
• By adding the additional complexity of considering the turbulent nature of the flow
field instantaneous or time-averaged measurements can also be considered.
• The decision of using full-scale or pilot scale measurements is often determined by
budget constraints.
5.3 Experimental Studies
In this section we look at experimental studies from the perspective of their suitability for
model testing. One of the key requirements is that field data for the primitive variables used
in the model, principally velocity concentration, temperature and particle size distribution, are
gathered in the experiment. In addition, other properties of the flow field and the physical
properties of the dispersed phase must be measured in order to carry out an accurate
numerical simulation. The density, inlet concentration and settling velocity of the sludge
should ideally be measured in order to fully characterize the sludge. The geometry of the tank
should be well defined and the inlet and outlet conditions well proscribed. Very rare
experimental studies fulfill all of these requirements in drinking water treatment plants.
Stamou and Rodi (1984) carried out a review of sedimentation tank experiments reported in
the literature with aim of identifying those suitable for testing mathematical settling tank
models. They concluded that the study by Imam (1981) was the only one suitable for this
purpose. In this study velocity was measured on a rectangular basin of relatively complicate
inlet and outlet geometry. Lack of satisfactory scope and quality of measurements such as
incomplete velocity profiles due to equipment shortcomings was also a factor. Insufficient
information regarding influent concentration with settling velocities that needed to be
estimated from empirical formulae, also contributed to the unsuitability of these results. It
was noted, however, that some of laboratories (2000-2008) results on settling tanks could be
used for validation studies provided that suitable approximations were made in 2D
simulations of the 3D geometry. Numerical comparisons have been made on some of these
test cases. Laboratories work provides the data for a lock exchange experiment, used in this
study to test the predicted velocity of a dense incursion for a given excess density.
Ghawi (2007) carried out experiments and numerical simulations on a large model scale
rectangular settling tank (Hrinova and Holic WTPs). He measured concentration and velocity
profiles at 7 stations along the tank centre line over a period of time long enough to observe
bed form development. In addition, he measured the dispersed phase density, inlet
concentration and settling velocity, making this the most complete data set for model scale
- 43 -
tanks. This experiment was chosen for numerical comparisons using the Discrete Phase
Model (DPM). Details of the experiment together with the results of the simulation are
described in detail next sections. Settling velocity and dispersed phase density were measured
for each of the cases, however, measurements along the centre line and overall geometry of
the tank were made.
For the purposes of testing the numerical model presented in this thesis on a full scale tank,
the data set gathered by Ghawi and Kris (2007a, 2007b) and laboratories, was selected.
Here, a comprehensive experimental study of a working settling tank at Holic and Hrinova in
Slovak republic was carried out. Velocity and concentration profiles were gathered at 7
stations along the length and 3 stations across the width of the tank for a variety of inlet
conditions and inlet and outlet geometries. Volumetric flow rates through the inlets and
outlets were measured for each test condition studied. Details of the tank geometry and the
experimental conditions for which 3D numerical simulations have been made are given in
next sections.
The remaining of this chapter will therefore discuss different experimental techniques. The
following topics are dealt with,
¾ Settling velocity
¾ Solids concentration
¾ Particle size distribution
¾ Velocity
¾ Temperature
Residence time distribution measurements are not discussed either. The reasons for omitting
these are the relative simplicity of the measurement technique and data handling, and the fact
that they have not been used in the CFD work in this dissertation.
5.3.1 Settling velocity
Before the settling models of Chapter 3 can be applied in a settling tank CFD model, they
need to be calibrated. Experiments are therefore needed to determine the model parameters,
and these should be consistent with the case under consideration. Indeed, the experimental
conditions should be comparable to real-life. As a result, it is important to define the
prevailing settling regime, i.e. discrete settling. The latter is crucial in order to determine the
experimental technique largely depending on the settling regime itself.
Sedimentation has many applications in preparation of potable water as it can remove
suspended solids and dissolved solids that are precipitated. There is some example below:
• Plain settling of surface water prior to treatment by rapid sand filtration (Type I)
• Settling of coagulated and flocculated waters in limesoda softening (Type II)
• Settling of waters treated for iron and manganese content (Type I)
The measurement of the discrete settling velocities are explained in the next sections
5.3.1.1 Measurement of Discrete Settling Velocities
The measurement of the discrete settling velocity of water sample starts with the
identification of the “threshold for discrete particle settling”, and the “lag time”. The test is
done in a transparent pipe, 2 m high with a diameter of 19 cm (Figure 5.2).
The “threshold for discrete particle settling” refers to the total suspended solids concentration
below which the particles settle in a complete discrete settling regime; particles settle as
individual units with no significant interaction with neighboring particles. Experimental
results obtained at the Holic WTP and Hrinova WTP indicates that this threshold is in the
range of 80 to 180 mg/L.
The “lag time” is the time at the beginning of the column batch test during which a
predominant vertical movement of the particles is not observed. Prior to the test, the sample
is agitated in order to produce a homogenous distribution of solids in the column. After that
the column is collocated in the upright position, an initial energy and momentum dissipation
- 44 -
produce the lag time. The lag time lasts for about 1 to 1.5 minutes before the discrete settling
starts.
Even though the settling velocity of individual flocs can be measured using a digital video
technique or a more sophisticated photographic technique, the settling velocity of “large” and
“medium” flocs can also be obtained by visual inspection and direct measurement: using a
halogen light to backlight the settling column, the individual flocs can be identified and
followed, and the settling velocity can be measured using a scale and stopwatch. The
procedure has to be repeated several times in order to get an appropriated number of
individual floc measurements. Kinnear (2002) used 50 individual measurements for
obtaining an average floc settling velocity. In this research average “large flocs” and
“medium flocs” settling velocities were obtained with at least 15 individual measurements
Figure 5.2 Laboratory settling column in Hrinova WTP
Table 5.1 Discrete Settling Velocities of Large Flocs for Holic ad Hrinova WTPs
Large Flocs
Flocs No. Time (s) V(m/h) Flocs No. Time (s) V(m/h)
1 10.5 10.3 11 14.0 7.7
2 14.2 7.6 12 12.3 8.8
3 9.3 11.6 13 14.3 7.6
4 14.2 7.6 14 9.5 11.4
5 11.2 9.6 15 9.9 10.9
6 15.4 7.0 16 10.4 10.3
7 15.0 7.2 17 6.5 16.6
8 6.9 15.9 18 8.1 13.4
9 7.7 14.0 19 6.5 16.6
10 11.6 9.3 20 9.1 11.9
Average Large Flocs Settling Velocity: 10.8 m/h
Standard Deviation: 3.16
Maximum: 16.6 m/h
Minimum: 7.0 m/h
5.3.1.2 Calibration of the Settling Sub-Model
As discussed in section above the discrete settling characterizes of the sludge are determined
using a sludge sample with a concentration equal (or lesser than) to the discrete threshold
concentration. The settling velocities of large and medium flocs are found by direct
- 45 -
measurement in a column batch test using a light source a scale and a stopwatch. The results
of the individual measurements of these types of flocs are reported in Table 5.2 and 5.3 in
Holic and Hrinova WTP.
Table 5.2 Discrete Settling Velocities of Medium Flocs
Medium Flocs
Flocs No. Time (s) V(m/h) Flocs No. Time (s) V(m/h)
1 26.5 4.1 11 25 4.3
2 47.2 2.3 12 30.2 3.6
3 46 2.3 13 29 3.7
4 45 2.4 14 26 4.2
5 30 3.6 15 30 3.6
6 47.9 2.3 16 29 3.7
7 46.5 2.3 17 45 2.4
8 49 2.2 18 38 2.8
9 28 3.9 19 58 1.9
10 40.7 2.7 20 40 2.7
Average Large Flocs Settling Velocity: 3.0 m/h
Standard Deviation: 0.79
Maximum: 4.3 m/h
Minimum: 1.9 m/h
5.4 Solids concentration

Since settling tanks have the task to gravitationally separate solids from water, it is obvious
that solids concentration profiles inside the basin are very relevant for model validation.
Solids concentrations are measured in view of other aspects of settling tank dynamics; e.g.
calibrating models for settling velocity. In view of its importance, a review of different
measurement techniques with a special emphasis on field applications is given below. In the
following discussion, two groups of techniques can be considered, i.e. On-line measurements
and in Off-line measurements (Figure 1.1).
On-line measurements.
Several techniques available on the market demand for in on-line measurements. Sampling
of the solids-liquid suspension, prior to determining the solids concentration in the lab. It is
obvious that any measurement bias from invasive sampling may determine the final accuracy
of the measurement. Because of its importance, van Rijn & Schaafsma (1986) reviewed
different sampling techniques such as mechanical traps, bottles and pumps.
After sampling, the solids concentration has to be determined. Solid concentration profile
(every hour) and sludge bed height (every 10 minutes) were measured on-line with a sensor.
The height of the sludge blanket was defined to be the height where the sludge concentration
reaches 0.8 g/l. The concentration sensors consist a Turbidity sensor (submersion type).
Off-line measurements.
Several techniques are available, i.e. gravimetric methods (APHA, 1992), nuclear methods
(Wren et al., 2000) and image analysis. The most important technique is the gravimetric
determination of solids concentration; APHA (1992) is the internationally standardised
method to determine solids concentration in water treatment. The technique consists of
evaporating the water from the sample and a subsequent determination of the solids weight.
The temperature of drying depends on what fraction of solids to be determined, i.e. total
suspended solids (drying of filter residue at 103-105 °C), total dissolved solids (drying of
filtrate at 180 °C) or fixed and volatile solids (igniting the residue of former two cases at 550
°C). In practice this method is used to calibrate solids concentration sensors; Total suspended
solids concentration of inlet, effluent, and non-settleable suspended solids concentration were
measured according to Standard Methods (APHA et al., 1992).
- 46 -
5.5 Liquid velocity

To validate a CFD model velocity measurements are very important. Although the residence
time distribution describes the hydraulic behaviour of the reactor, it only allows indirect
knowledge about the flow pattern. Hence, to validate the flow field velocity data are
essential. In literature, however, not many references to lab-scale or full-scale velocity
measurements can be found; most CFD modelers refer to the work of e.g. Larsen (1977.
Many possibilities exist to measure liquid velocities which vary between zero and 6 cm/s in
settling tanks (Anderson, 1945; Bretscher et al., 1992; Kinnear, 2000). Kinnear & Deines
(2001) measured complete velocity profiles with a single measurement in the clarifier (Figure
5.3). The application of each technique in practice depends on e.g. lab- or full-scale, solids
concentration and cost. Liquid velocity profiles (every hour) were measured on-line with a
sensor in both water treatment plants.
Figure 5.3 Measured velocity profiles in the clarifier (Kinnear & Deines, 2001)
5.6 Particle size distribution

The behaviour of particles is of major interest when studying settling tanks. In the hindered
settling regime, particle settling is primarily dictated by the prevailing solids concentration.
At low concentrations, however, discrete settling occurs. Here, the particle’s behaviour is
influenced by many factors; amongst them, particle size is the most commonly determined
parameter. Clearly, particle characterisation is crucial for process understanding and
development.
In practice, many measuring techniques of Particle Size Distributions (PSD) exist. A major
characteristic of particle sizing technology is that every technique focuses on different
particle characteristics and, therefore, results of the different methods will be different as
well. Although some researchers combine techniques to cover a larger range of particle sizes
(Li & Ganczarczyk, 1991), such practices should be avoided due to this incompatibility.
Due to different measurement principles particle sizes may be based on particle lengths,
surface areas or volumes. Also the frequency of occurrence may be expressed differently; for
instance, the particle numbers or volume fractions in specific size classes can be utilised.
Knowledge on the particle shape is clearly necessary to enable any transformation between
different size distributions; e.g. transforming a number-frequency to a volume-frequency size
distribution. However, this particle shape information is generally absent and spherical
shapes are adopted in the calculations. More detail about PSD is discussed in the next
chapter.
- 47 -
5.6.2 Particle size distribution-Experimental determination

Samples of incoming and effluent suspensions were taken and analyzed for particle size
distribution using the laser diffraction technique. The location of sampling is very important
and depends on the goal of the study.
When samples are taken at the tank overflow exit, care was exercised in order to sample from
a well-mixed and representative location. Sampling took place during at least twice the
theoretical residence time. If the system is unclear or there are known dead spaces, it might
be considered to prolong the measurement campaign to capture the complete hydraulic
behaviour.
Figure 5.4 Particle size distributions in the influent of the Holič and Hrinova sedimentation tank
Drawbacks of the laser diffraction technique are the assumption of sphericity of particles in
the optical model and the required dilution step to avoid multiple scattering, because this is
not taken into account by the optical model. The latter is checked by means of the
obscuration level, which should be inside a certain level. The possibility of misinterpreting
the size distribution of open porous floccules by assuming them as compact spheres is well
known in the literature but at present laser diffraction techniques are acceptable since there
are no better alternatives.
Figure 5.4 presents the measured particle size distribution in the influent of the Hrinova and
Holič sedimentation tanks. The Figure represents average values of three measurements
conducted for the three sedimentation tanks of the plant. The repeatability expressed as the
average standard deviation of the three measurements was 1.3%.
5.7 Full scale measurements
Over the last two decades, there has been a large growth in the application of computational
fluid dynamics (CFD) to design of facilities. CFD has many advantages over traditional
modelling approaches as it is a low-cost, high speed technique for evaluating engineering
systems that are difficult to simulate in a laboratory or under field conditions. CFD is able to
yield a “virtual prototype” and a good example of this is in its application to the design of
compact and more efficient sedimentation tanks for traditional water treatment plants. CFD
can capture the three-dimensional fluid flow inside a tank and thus help to minimize
turbulence and optimize solids separation. The purpose of the sedimentation tank is to allow
the particles sufficient time to settle out of the water. Turbulence should be minimized so that
the water may move as quiescently as possible. The horizontal velocity through the tank
should be small, and the residence time should be large.
- 48 -
The section below describes the Plants layout. At the same time, the settling tank physical
characteristics will be highlighted. It is further attempted to validate the CFD model both in
terms of hydrodynamics and solids transport.
5.7.1 Plants layout
Groundwater is an important source of municipal drinking water for many small and
medium-sized communities in Slovakia. Many favour groundwater over surface water
because of its excellent and consistent quality, and because, generally, it requires little or no
treatment before consumption. Unfortunately, many groundwater supplies are contaminated
by varying levels of iron and manganese in concentrations that exceed the Slovakian
Drinking Water Guidelines.
When present at very high concentrations, or when existing as either the insoluble oxide or as
an organic complex as may be found in surface water, the removal of iron, and to a lesser
degree manganese, may be accomplished by gravity process using coagulation, flocculation
and sedimentation processes. Also, when accompanied by either high turbidity or colour a
sedimentation step will be required. This may then be followed by high-rate sand or dual
media pressure or gravity filtration to insure the complete removal of any fine precipitates of
iron and manganese.
When dealing with a groundwater supply there is no hard or fast rule to determine under
exactly what circumstances settling would be recommended over either ion exchange or
oxidation followed by either standard filtration or filtration using manganese greensand. The
method chosen would depend on a variety of conditions including raw water iron and
manganese concentration, the chemical form in which they exist, the plant flow rate, and
additional treatment other than iron and manganese removal which may be required (e.g.,
softening, and turbidity or colour removal). Other factors to be considered would be initial
capital cost, operating and chemical costs, ease of operation, requirements for disposal of
waste products, and the effluent quality required.
Figure 5.5 Layout of Holic WTP
Figure 5.5 represents treatment of water obtained from a deep well (groundwater) in Holic
WTP and Figure 5.6 depicts the same treatment of a surface water supply in Hrinova WTP.
The Hrinova and Holic WTP were built to remove iron Fe+3, manganese, turbidity and
organic material. The layout of the Holic can be summirised in Figure 5.5 the layout of the
Hrinova plant can be summarised by the following stages (Figure 5.7),
• Reservoir
- 49 -
• Coagulation tank
• Flocculation tank
• Horizontal settling tank
• Filtration tank
• Disinfection tanks
• Storage tank
.
Figure 5.6 Schematic of Hrinova WTP.
Figure 5.7 Layout of Hrinova WTP.
The majority of iron and manganese treatment systems in Slovakia employ the processes of
oxidation/ sedimentation. The oxidant chemically oxidizes the iron or manganese (forming
a particle), and kills iron bacteria and any other disease-causing bacteria that may be present.
The sedimentation tank then removes the iron or manganese particles.
Oxidation followed by sedimentation is a relatively simple process. The source water must be
monitored to determine proper oxidant dosage, and the treated water should be monitored to
determine if the oxidation process was successful
Before iron and manganese can be settled in Hrinova and Holic WTPs, they need to be
oxidized to a state in which they can form insoluble complexes. Holic WTP use aeration
- 50 -
tank to oxidize iron and manganese. Oxidation involves the transfer of electrons from the
iron, manganese, or other chemicals being treated to the oxidizing agent. Ferrous iron (Fe2+)
is oxidized to ferric iron (Fe3+), which readily forms the insoluble iron hydroxide complex
Fe(OH)3. Reduced manganese (Mn2+) is oxidized to (Mn4+), which forms insoluble (MnO2).
Hrinova and Holic water treatment plant have high iron concentration.
In a surface water supply iron and manganese may be present due to their dissolution from
the associated geologic formations and/or from the decomposition of organic materials. For
example, anaerobic conditions on a reservoir bottom may cause the dissolution of iron and
manganese from the bottom sediments. When seasonal overturns occur due to temperature
gradients, this dissolved iron and manganese will be distributed throughout the water supply.
Although increased levels of iron and manganese may occur during periods associated with
overturns, it is also very likely that these elements will be found at objectionable
concentrations throughout the year.
In the removal of iron and manganese by sedimentation followed by filtration, the mechanism
is the complete oxidation of the elements to the insoluble forms followed by coagulation and
sedimentation to remove the bulk of the metal oxides and final filtration to remove the
remaining precipitated products of oxidation. In a surface water (Hrinova WTP) in which the
iron and perhaps the manganese are already oxidized, coagulation is required to
agglomerate the precipitated oxides and to aid settling. In addition, complementary oxidation
using an oxidizing agent may be required if manganese is present in its reduced form. In
Hrinova WTP has low manganese concentration. Sedimentation will then provide for the
removal of the bulk of the metal oxides and allow a sufficient detention time for the complete
oxidation of iron and manganese.
The most common chemical oxidant in groundwater water treatment in Slovakia is
potassium permanganate. Oxidation using potassium permanganate is frequently applied in
small groundwater systems (Holic WTP). The dosing is relatively easy, requires simple
equipment, and is fairly inexpensive.
Potassium permanganate (KMnO4) is normally more expensive than chlorine and ozone, but
for iron and manganese removal, it has been reported to be as efficient and it requires
considerably less equipment and capital investment.
The dose of potassium permanganate, however, must be carefully controlled. Too little
permanganate will not oxidize all the iron and manganese, and too much will allow
permanganate to enter the distribution system and cause a pink color. Permanganate can also
form precipitates that cause mudball formations on filters. These are difficult to remove and
compromise filter performance.
In general, manganese oxidation is more difficult than iron oxidation because the reaction
rate is slower. A longer detention time (10 to 30 minutes) following chemical addition is
needed prior to filtration to allow the reaction to take place.
There are different filtration media for the removal of iron and manganese, including
manganese greensand, anthra/sand or ironmansand, electromedia, and ceramic.
Manganese greensand is by far the most common medium in use for removal of iron and
manganese through pressure filtration in Holic WTP. The material is coated with manganese
oxide. This treatment gives the media a catalytic effect in the chemical oxidation reduction
reactions necessary for iron and manganese removal. This coating is maintained through
either continuous or intermittent feed of potassium permanganate.
5.7.2 Settling tanks description of Holic water treatment plant
The numerical model is compared with data obtained in a settling tank. In this study, mean
velocity and solids fraction concentration were measured at a number of stations along the
length and across the width of the Holíc settling tanks.
- 51 -
The time taken to gather all the data for one inlet condition was around 3 hours, so no distinct
time was given for the measurements at each station. Because of this, the numerical
simulation was compared with the experimental data at a time when the settled sludge layer
in the simulation was at approximately the same height as that found in the experiment. At
this time, the flow field above the settled sludge layer should be similar in both the
experiment and the simulation.
Figure 5.8 Schematic representation of the Holic sedimentation tank in 2D
Figure 5.9 Picture of Holic sedimentation tank

The horizontal settling tank is shown in Figures 5.8 and 5.9, it is 30 m long, 4.5 m wide and
has a maximum depth at the hopper of 2.5 m. Longitudinal velocity was determined by
means of sensors, a method that involves measuring the time taken for a drifter at the
measurement depth to traverse a given distance from which the velocity could be calculated.
Accuracy is given to +/- 2mm/s.
Solids concentration was measured by gravimetric determination method and sensor.
Interference with concentration measurements occurred when the probe was placed within
0.2 m of the walls or sludge scraper, hence no readings were taken closer than this distance.
Also these measurements were limited to an upper solids fraction of about 0.07, around 1.3
times the inlet solids fraction.
Tank Inlet. Each tank is fed by pipe from flocculation tank. As the water enters the
sedimentation tank. This arrangement is shown in Figure 5.10. One pipe 0.8 m diameter
allows the flocs to leave the flocculation tank and enter into the settling tank (see Figure
5.10).
Tank Outlets. Sludge is withdrawn from the base of the sludge hoppers by a pump. The clear
water is withdrawn from the tank surface by effluent outlet weir located at 29m, along the
width of the tank (Table 5.3).
- 52 -
Figure 5.10 Schematic representations of the Holic sedimentation tank inlet and hopper.
Table 5.3 Physical and hydraulic data during study periods, and settling tank data.
Geometry Value
Tank length 30.0 m
Tank width 4.50 m
Water depth 3.80 m
Hopper depth 2.50 m
Bottom slop 0.0038
Weir length 4.50 m
Weir width 0.70 m
Weir depth 0.50 m
loading Value
SOR 2.7 m/h
Density of water 1000 kg/m3
Particle Density 1066 kg/m3
Tank parameter Value
Average flow rate 80 l/s
Sludge pumping rate 5 l/s
Inflow temperature, 5oC -11oC , and 25oC -27oC
Inflow suspended solids ~ >50 mg/l
Detention time 3.6 hr
Cmin 0.17 mg/l
μ 0.001 N.s/m2
No. of Tank 2
Sludge Scrapers. Settled sludge is moved along the floor of the tank towards the sludge
hopper by a continuous chain scraper. This scraper moves baffles 0.125m deep at a rate of
0.008 m/s towards the inlet.
Inlet Conditions. The total inlet volumetric flow rate was 80 l/s.
The density difference between inlet and effluent is between 1.05-2 kg/m3 for all the
experiments. The sludge density is 1066 kg/m3, which is the density of the hydrated sludge
flocs in suspension. This is low compared to dry sludge densities previously reported between
1066 and 2000 kg/m3, Larsen (1977) and Dahl (1993).
This is rather high in comparison to the inlet solids fraction reported in the previous
simulations.
A rectangular sedimentation tank in Slovakia water treatment plants was selected to
demonstrate the response of rectangular tanks to different internal geometries. This case is
- 53 -
based on the Holic settling tanks describe in this chapter. This tank was selected because
performance data are available for model calibration, and because it represents a marginal
performance case. Table 5.3 shows the main tank dimensions and loadings.
The model hydrodynamic parameters were calibrated before the model was applied. The data
on velocities and concentrations were used to adjust the model bed roughness and vertical
Prandtl-Schmidt number. The data in Table 5.4 and 5.5 shows the water quality of Holic
WTP (Turbidity, pH, Manganese and Iron) before and after treatment in summer and winter.
As stated on all analyses received since treatment started “This water complies with the
recommended guidelines [WHO] for drinking water quality.”
Table 5.4: Observed data on 16/01/2006 (Holic WTP)
Before Treatment After Treatment
Date pH Manganese Iron Turbidity Manganese Iron Turbidity
January (mg/L) (mg/L) (NTU) (mg/L) (mg/L) (NTU)
24/2006
08:00 8.06 2.5 11.2 30 0.6 3.0 10
09:00 8.03 2.2 10.5 25 0.7 3.6 10
10:00 7.76 2.2 10.5 50 0.6 2.8 9
11:00 7.76 2.4 9.3 45 0.6 2.4 11
12:00 8.07 2.4 9.3 30 0.8 2.4 11
13:00 8.10 2.4 8.6 32 0.6 2.4 12
14:00 7.75 2.2 9.5 32 0.3 2.6 11
15:00 7.75 2.0 7.7 28 0.3 2.0 14
16:00 7.75 2.0 10.6 52 0.3 2.2 12
17:00 7.79 2.4 10.4 52 0.2 2.2 10
18:00 7.78 2.3 9.8 45 0.2 2.0 11
19:00 8.10 2.2 8.6 45 0.09 1.9 12
20:00 8.08 2.2 8.7 30 0.08 2.0 14
21:00 8.46 2.2 8.7 30 0.3 2.0 11
Limited 8.5 0.05 0.2 5
Table 5.5: Observed data 24/07/2006 (Holic WTP)

July (mg/L) (mg/L) (NTU) (mg/L) (mg/L) (NTU)
5/2006
08:00 7.06 1.5 11.2 30 0.06 3.2 10
09:00 7.89 2.2 12.5 25 0.17 3.16 10
10:00 7.17 2.0 12.5 50 0.99 2.18 9
11:00 7.17 1.8 10.3 45 1.0 2.24 11
12:00 7.89 1.7 9.3 30 0.8 2.4 11
13:00 8.26 2.0 8.6 32 0.6 2.9 12
14:00 8.24 2.2 9.5 32 0.03 2.2 11
15:00 8.09 2.0 6.7 28 0.3 2.1 11
16:00 8.12 2.0 10.6 52 0.3 2.8 12
17:00 7.37 2.4 10.4 52 0.62 2.2 10
18:00 8.03 2.1 8.8 45 0.2 2.0 8
19:00 8.40 2.2 8.6 45 0.09 1.9 12
20:00 8.30 2.0 9.7 30 0.08 2.0 7
21:00 8.06 2.2 8.7 30 0.63 2.7 11
Limited 8.5 0.05 0.2 5
5.7.3 Settling tank description of Hrinova water treatment plant
- 54 -
The removal of iron and manganese has been the subject of numerous papers over the last 50
years or more. In this study CFD model has used to study iron Fe3+, manganese and turbidity
removals by sedimentation tank in Hrinova water treatment plant
Hrinova WTP was built in 1963. The untreated water is pumped from the reservoir to the
Hrinova WTP by five large submersible pumps and is then dispatched into two parallel
process lines. The water then flows in each line through a clarifier (coagulation, flocculation,
and horizontal sedimentation tank) and filtered through a sand bed. After chlorination, the
water is stored in three underground reservoirs. Five booster pumps, connected to the last
reservoir, ensure the water supply in the Hrinova network.
Figure 5.11: Picture of the horizontal settling tank of Hrinova WTP.
A full-scale rectangular sedimentation tank was investigated, similar to those used in the
potable WTP of the city of Hrinova (Figure 5.11). There are 4 rectangular tanks at the
Hrinova Water Treatment Plant. The plant receives raw water from reservoir and its capacity
is around 200 l/s. The employed processes include coagulation-flocculation, sedimentation,
sludge thickening, filtration through sand and chlorination. The volume of sedimentation tank
is 500 m3 (Figure 5.12). Each tank’s influent flow is introduced through a weir and the water
flow through a perforated inlet baffle. The purpose of this baffle is to distribute the flow
uniformly over the entire cross-sectional area of the tank. Each of the openings has the same
size, and they distributed uniformly across the baffle. There should not be any openings
towards the bottom of the baffle, however, because this could increase scour of the deposited
sludge. The goal when designing this baffle is to achieve some head loss while keeping the
velocity gradients through the ports equal to (or lower than) the velocity gradient in the end
of the flocculator, so as to not break up the flocs. In each tank there is 1 effluent weir, which
goes along the longitude direction and is distributed in a range of 4 meters from the side wall.
The multiple scrapers driven by chain move toward the tank upstream hopper to collect the
settled sludge into the sludge hopper located at the beginning of tank.
The Hrinova WTP uses lime, (NH4)2SO4 (Ammonium Sulphate) and Fe2(SO4)3 (Iron (III)
Sulfate) to flocculate the Fe+3 before entering the sedimentation tanks. The tanks have wedge
shaped sludge hoppers below the inlet. The sludge is withdrawn hydraulically from the
narrowest part of the hopper. A flocculation channel is positioned on top of the sedimentation
tank at the inlet.
The clean water is withdrawn from the end of the tank through overflow weir. It was recently
observed that significant amounts of sludge are not removed from the sedimentation tank
during desludging as it is washed out of the hopper into the horizontal section of the tank.
This requires frequent decommissioning of the tank to remove the sludge buildup.
- 55 -
Figure 5.12 Schematic representations of the Hrinova sedimentation tank and WTP
diagram.
The current study not only focuses on the hydrodynamics of the flow using a relatively
efficient computer simulation model, but also improves the understanding of both flow and
sediment characteristics.
- 56 -
In order to assess the performance of the Water Treatment Plant (WTP) in terms of water
quality, turbidity measurements were taken at the inlet and the outlet of the settling tanks.
The outlet measurements are to be compared with water quality standard values. Also in
Hrinova WTP there is monitoring system to measure deferent chemical parameters (Figure
5.13).
Figure 5.13 Data monitoring in Hrinova WTP.

Water turbidity was measured with the portable turbidimeter HACH Model 2100P from June
th th
7 , 2007 to August 14 , 2007.
This turbidimeter operates on the nephelometric principle of turbidity measurement (scatter
light ratio to transmitted light). The measuring range is 0–1000 NTU with an accuracy of +/-
2% of readings (HACH 2004). The calibration of the turbidimeter is based on three samples
of standard turbidity (20, 100, 800 NTU). The calibration was performed once at the
beginning of June and regularly checked during the duration of the field research.
Daily water samples were collected at the inlet of the settling tanks. For three high-turbidity
events, measurements of the inlet turbidity (and outlet turbidity for two of these events) were
taken at least every fifteen minutes. Water samples were taken with a 1.5 L plastic bottle
filled in an average of 15 s. Afterwards, the collected water was stirred to ensure the
homogeneity of the solution and a sample of 15 ml was collected and its turbidity measured
with the turbidimeter. For the sunny days, the sampling was done at a random time. For rainy
days, several samples were collected and the maximum turbidity recorded. As the samples
during the sunny days always showed turbidity (< 25 NTU), the turbidity on some sunny
days towards the end of the study was not measured.
- 57 -
Daily turbidity measurements are shown in Figure 5.14. The maximum turbidity recorded is
about 190 NTU but a lot of rain events were measured. On sunny days, the turbidity was
about 20 NTU on average and never exceeded 30 NTU.
Figure 5.14 Daily turbidity measures from 7 June 2007 to 14 august 2007
The data in Table 5.6 and 5.7 show the water quality of Hrinova WTP (Turbidity, pH,
Manganese and Iron) before and after treatment in summer and winter. As stated on all
analyses received since treatment started “This water complies with the recommended
guidelines [WHO] for drinking water quality.”
Table 5.6: Observed data February 13/2006 (Hrinova WTP)

February (mg/L) (mg/L) (NTU) (mg/L) (mg/L) (NTU)
27/2006
08:00 7.06 1.5 9.2 40 0.16 2.0 10
09:00 7.19 1.2 8.5 55 0.17 1.6 10
10:00 7.17 2.0 7.5 50 0.19 1.8 9
11:00 7.17 2.1 7.3 45 0.19 2.4 11
12:00 7.15 1.4 8.3 43 0.18 3.4 11
13:00 7.26 2.0 8.6 32 0.06 1.4 12
14:00 7.24 2.0 7.5 32 0.03 2.6 11
15:00 8.09 1.4 7.7 67 0.13 3.0 14
16:00 8.12 1.1 6.6 52 0.03 3.2 12
17:00 7.59 1.4 6.4 52 0.02 2.2 10
18:00 7.30 2.0 7.8 60 0.12 2.1 11
19:00 7.40 2.0 7.6 73 0.09 2.5 12
20:00 7.30 2.3 7.7 45 0.08 1.0 14
21:00 7.06 2.1 7.7 45 0.13 1.0 11
limited 8.5 0.05 0.2 5
- 58 -
Table 5.7: Observed data June 19/2007 (Hrinova WTP)

June (mg/L) (mg/L) (NTU) (mg/L) (mg/L) (NTU)
19/2006
08:00 7.06 1.05 8.2 80 0.16 2.1 20
09:00 7.19 1.02 6.5 66 0.17 1.6 18
10:00 7.17 2.10 8.5 55 0.19 3.8 11
11:00 7.77 1.10 147 55 0.19 2.4 14
12:00 7.89 1.40 7.3 54 0.18 2.5 13
13:00 7.26 2.00 6.6 56 0.06 3.0 51
14:00 7.24 2.20 6.5 77 0.03 2.6 13
15:00 8.09 1.04 7.7 67 0.13 3.0 14
16:00 8.12 1.01 9.6 77 0.03 3.2 12
17:00 8.40 1.70 7.4 77 0.02 2.2 14
18:00 7.30 2.10 7.8 60 0.12 2.0 14
19:00 7.54 2.00 7.6 73 0.09 1.9 16
20:00 7.36 2.03 6.2 65 0.08 1.9 16
21:00 7.76 1.61 6.1 90 0.13 0.6 22
limited 8.5 0.05 0.2 5
5.8 Temperature
Experiments were conducted in Slovakia during the summer and winter at water treatment
plant in Hrinova in Slovakia. In order to examine the effects of surface heating, study
periods were chosen coinciding with summer weather conditions when surface heat losses
would be significant. These conditions were most significant during the day (afternoon) hours
when plant flow rates were high and summer heating was greatest.
In order to measure the temperature distributions in these tanks, thermistors at varying depths
were attached to each tank. Each thermistor was attached to a computer data logger that
recorded temperature continuously during the study periods. The study period for the
sedimentation tanks was between 3 and 4 days in winter and summer in 2006 and 2007.
The data presented in this section was collected at the Hrinova WTP in winter and summer.
This data shows the results of the procedure presented in chapter 6 aimed to determining a
correction factor for the zone settling velocities based on temperature difference.
Table 5.8 summarizes the different temperatures (Ts (summer temperature) and Tw (winter
temperature) measured at Hrinova WTP.
Table 5.8 measured temperature in winter (average) and summer (average) at Hrinova WTP
Sample at Summer Temperature (Ts) Sample at Winter Temperature (Tw)
Inlet temperature 0C Outlet temperature 0C Inlet temperature0C Outlet temperature 0C
27.5 27.6 6.0 6.0
27.5 27.8 4.6 4.8
26.0 26.2 5.8 6.8
25.3 25.3 4.0 5.9
23.7 23.9 3.8 5.0
23.7 24.0 6.2 6.4
21.4 22.0 6.0 6.8
21.4 21.9 5.4 6.6
5.9 Conclusions
In this chapter an extensive review of experimental techniques to calibrate and validate CFD
models was given. Since settling tanks aim to gravitationally separate flocs from the liquid, it
- 59 -
is clear that data about solids settling is crucial. Next to this support for calibration, model
validation has to be performed too in order to check the model’s ability to predict the
system’s response to external disturbances. To this end, simulation results have to be
confronted with local measurements of velocity and solids concentrations. Many methods are
available and the pros and cons of each should be weighted in view of the applicability to the
investigated system. In this respect, the reader is referred to the respective sections of this
chapter that discuss the different measurement techniques. For the purposes of testing the
numerical model presented in this thesis on a full scale tank, the data set gathered by
measurement and laboratories, was selected. Here, a comprehensive experimental study of a
working settling tank at Holic and Hrinova in Slovak republic was carried out. Velocity and
concentration profiles were gathered at 7 stations along the length and 3 stations across the
width of the tank for a variety of inlet conditions and inlet and outlet geometries. Volumetric
flow rates through the inlets and outlets were measured for each test condition studied.
Details of the tank geometry and the experimental conditions for which 3D numerical
simulations have been done in this and next chapter.
- 60 -
Chapter 6
Model Development, Applications
and Results
6.1 Introduction
This chapter presents a case study demonstrating the development, simulation and prediction
power of CFD. The full-scale horizontal settling tanks at the drinking treatment plant of
Hrinova and Holic (Slovakia) were opted for. The WTP is operated by Slovak Water; this
organisation is part of the Bratislava City Council.
Most settling tanks of Slovak Water exhibit a horizontal settling tank. Generally, many
researchers have used CFD simulations to describe water flow and solids removal in settling
tanks for sewage water treatment. However, works in CFD modelling of sedimentation tanks
for potable water treatment have not been found in the literature. Moreover, the physical
characteristics of the flocs may not be such significant parameters in the flow field of settling
tank for potable water, due to the much lower solids concentrations and greater particle size
distributions than those encountered in wastewater treatment. Hence, research was focused on
this type of settling tanks.
The sections below describe the modelling approach. At the same time, the settling tank
physical characteristics will be highlighted. It is further attempted to validate the CFD model
both in terms of hydrodynamics and solids transport.
6.2 Modelling the settling tank
This section deals with the set-up of the settling tank model (Figure 6.1). The fundamental
equations are presented together with their boundary conditions. A real-life settling tank is a
3D system, hence an appropriate spatial simplification has to be realised in order to perform
simulations (with 2500 MHz Pentium 4 and 2 GHz Pentium IV processors, and 512 Mb
RAM) within an acceptable time frame. The lowering of the model dimension to 2D has
important ramifications for the modelling of the inlet structure and the scraper operation.
Different approaches to model them in 2D are dealt with in this chapter.
Pre-processing Equation solved on mesh
. Mesh . Solid y Transport equation

Generator Modeller . Mass y Physical model
. Species mass fraction . Turbulence (k-e Model)
. Phases volume fraction . DPM (Lagrangian Model)
.Solver . Momentum . Phases change
Settings . Moving mesh
. Energy
y Equation of State
. Post-Processing y Supporting physical Models
. Velocity
. Concentration
. Temperature
. Dimensions
. Efficiency
. Material properties
. Boundary conditions
. Initial conditions
Figure 6.1 STs Development CFD Model
- 61 -
6.2.1 Mechanistic description of the settling tank: contributing

mathematical equations
The computational fluid dynamics code FLUENT 6.3.26 has been developed to carry out the
simulations. The code predicts fluid flow by numerically solving the partial differential
equations, which describe the conservation of mass and momentum. A grid is placed over the
flow region of interest and by applying the conservation of mass and momentum over each
cell of the grid sequentially discrete equations are derived. In the case of turbulent flows, the
conservation equations are solved to obtain time-averaged information. Since the time-
averaged equations contain additional terms, which represent the transport of mass and
momentum by turbulence, turbulence models that are based on a combination of empiricism
and theoretical considerations are introduced to calculate these quantities from details of the
mean flow. Either an Eulerian or a Lagrangian approach can be adapted to model particulate
phase. In the literature, Eulerian applications are used for almost all diffusion dominated
problems, so strictly speaking they are only suitable for gas or ultrafine particle study (De
Clercq, and. Vanrolleghem 2002). Due to their versatile capabilities, approaches based on
the Lagrangian method have been applied extensively for many two-phase flow problems. In
these approaches, the fluid is treated as a continuum and the discrete (particle) phase is
treated in a natural Lagrangian manner, which may or may not have any coupling effect with
fluid momentum. De Clercq and Vanrolleghem 2002 mentioned that the Lagrangian model
should not be applied whenever the particle volume fraction exceeds 10–12%.
The trajectories of individual particles through the continuum fluid using the Lagrangian
approach are calculated in FLUENT by the discrete phase model (DPM). The particle mass
loading in a sedimentation tank for potable water treatment is typically small, and therefore, it
can be safely assumed that the presence of particles does not affect the flow field (one-way
coupling). This means that the fluid mechanics problem can be solved in the absence of
particles to find the steady state flow field. Then the particles, whose density and size could
be assigned at will, are released from the inlet and are tracked along their trajectories. In
addition, the volume fraction of the particles in the tank is of the order of 10−4. The turbulent
coagulation is well known to be proportional to this volume fraction, so it can be ignored
under the present conditions. Also, the coagulation due to differential settling can be ignored
due to the relatively low settling velocities resulting by the low densities of the flocs.
The settling velocity hindering is insignificant for these levels of solids volume fraction as it
can be shown by employing the corresponding theories (Berres et al. 2005). Moreover, Lyn
et al. 1992, based on model observations, concluded that for conditions of relatively small
particles concentrations in sedimentation tanks, the flocs coalescence do not affect the flow
field and the effects on the concentration field and the removal efficiency may be of
secondary importance. Finally, drops in the size range relevant to primary separators do not
suffer breakage (Wilkinson et al. 2000 ).
The final system of particle conservation equations is a linear one, so the superposition
principle can be invoked to estimate the total settling efficiency. The inlet particle size range
is divided in classes with the medium size of each class assumed as its characteristic (pivot).
Then independent simulations are conducted for monodisperse particles in the feed using
every time the individual pivot sizes. The overall settling efficiency can be found by adding
appropriately the efficiency for each particle size.
Tracks are computed by integrating the drag, gravitational and inertial forces acting on
particles in a Lagrangian frame of reference. The dispersion of particles due to turbulence is
modeled using a stochastic discrete-particle approach. The trajectory equations for individual
particles are integrated using the instantaneous fluid velocity along the particle path during
the integration. By computing the trajectory in this manner for a sufficient number of
- 62 -
representative particles, the random effects of turbulence on particle dispersion may be

accounted for.
6.2.1.1 Continuity and Momentum Equations
The two-dimensional (2D), axisymmetric flow of an incompressible fluid is governed by the
continuity and the Navier- Stokes equations. Employing the Reynolds-averaging procedure,
the turbulent stresses appearing in the transport equations are made proportional to the mean
velocity gradients according to the Boussinesq eddy-viscosity hypothesis. The continuity, and
momentum equations expressed in the coordinate system read, respectively
∂
(U ) + ∂ (V ) = 0 …………..…………………………………………………………………………6.1
∂X ∂Y
∂U ∂U ∂U 1 ∂P ∂ ⎛ ∂U ⎞ ∂ ⎛ ∂U ⎞
+U +V =− + ⎜⎜ μ ⎟⎟ + ⎜⎜ μ ( ⎟
⎝ eff ∂X ⎠ ⎝ eff ∂Y ⎟⎠…………………….………………6.2
∂t ∂X ∂Y ρ ∂X ∂X ∂Y
∂V ∂V ∂V∂V 1 ∂P ∂ ⎛ ⎞ ∂ ⎛ ∂V ⎞
ρ p − ρw
+U +V =− + ⎜⎜ μ ⎟⎟ + ⎜⎜ μ ( ⎟⎟ + g …6.3
∂t ∂X ∂Y ρ ∂Y ∂X ⎝ eff ∂X ⎠ ∂Y ⎝ eff ∂Y ⎠ ρw
where U and V denote the time-averaged velocities in the X and Y direction, respectively; P
= static pressure; ρw = density of clear water that serves as reference density; μ eff = effective
viscosity; and ρp = density of the particle.
6.2.1.2 k-ε Turbulence Model
The k-ε eddy-viscosity model [see, e.g., Rodi (1993)] determines the isotropic eddy viscosity
μ t as a function of the turbulent kinetic energy k and its dissipation rate ε by
μ eff = μ + μt
k2
μt =C ………………………………………………………………….…………….………………6.4
μ ε
The distributions of k and ε are determined from the following model transport equations:
∂ k ∂ k ∂ k ∂ ⎛⎜ μ eff ∂ k ⎞⎟ ∂ ⎛⎜ μ eff ∂ k ⎞⎟
+U +V = + + P + p − ε ) ………………6.5
∂t ∂X ∂ Y ∂X ⎜ σ k ∂X ⎟ ∂ Y ⎜ σ k ∂ Y ⎟ ⎝ ⎠
k b
⎝ ⎠
⎛μ ⎞ ⎛μ ⎞
∂ε ∂ε ∂ε ∂ ⎜ eff ∂ε ⎟ ∂ ⎜ eff ∂ε ⎟ ε ε 2 ………….……6.6
+U +V = ⎜⎜ σ ⎟⎟ + ⎜⎜ σ ⎟⎟ + C P − C
∂t ∂X ∂Y ∂X ∂X ∂Y ∂Y 1 k k 2 k
⎝ ε ⎠ ⎝ ε ⎠
Where
⎧ ⎡
⎛ ∂U ⎞
2
⎛ ∂ V ⎞ ⎤⎥ ⎛ ∂ U
2
∂ V ⎞ 2 ⎫⎪
P = μ eff ⎪⎨ 2 ⎢ ⎜ ⎟ +⎜ ⎟ +⎜ + ⎟ ⎬
k ⎪ ⎢⎝ ∂ X ⎠ ⎝ ∂ Y ⎠ ⎥⎦ ⎝ ∂ Y ∂X ⎠ ⎪
⎩ ⎣ ⎭
represents the rate of production of turbulent kinetic energy resulting from the interaction of
the turbulent stresses and velocity gradients, and
∂(ρ − ρ )
g μ p eff w
P = ……………………………………………………………….……6.7
b ρw σ ∂Y
c
represents the rate of production due to buoyancy effects. The empirical constants are given
the standard values suggested by Rodi (1993); Cμ = 0.09, C1 = 1.44, C2 = 1.92, σk = 1.0, and
σε = 1.3. The value of the empirical constant σc associated with the buoyancy source term was
found to depend on the flow situation considered. Test calculations have shown that the value
of σc depends on whether Pb is a source term (i.e., in unstably stratified flows) or a sink term
- 63 -
(i.e., in stably stratified flows). Rodi (1987, 1993) suggested that σc is in the range of 48-52
for stable stratification that prevails in ST and tends toward zero for unstable stratification.
6.2.1.3 Energy Equation
The following energy equation can be used to model the temperature field (Zhou et al.,
1994; Ekama et al., 1997):
⎛ ⎞
∂T
ρ ( ∂ T + u j ∂ T = ∂ ⎜⎜ λ ⎟
− ρ u T ' ⎟ ………………………………………………….……6.8
∂t ∂ x j ∂ xi ⎜ ∂x i ⎟
⎝ i ⎠
where T and T' are respectively the mean and fluctuating component of the temperature, λ is
molecular diffusivity.
The temperature effects are commonly included in the reference density and kinetic viscosity
of the water by means of equations of state. The following expressions represent examples of
such equations:
ρref =[999.8396+18.224944*T-0.00792221*T2-55.4486*10-6 * T3+14.97562*10-8 * T4-
39.32952*10-11 * T5 +(0.802-0.002* T)*TDS]/[1+0.018159725*T] ……………………….……6.9
⎡ 247.8 ⎤
⎢ ⎥ ………………………………..………….……6.10
μ = (2.414 *10 − 5 ) *10 ⎣ T + 133.15 ⎦
ref
In Equations 6.9 and 6.10 T is the water temperature in °C, ρ is the water reference density
ref
in g/L, TDS is the Total Dissolved Solids in g/L, and μ is the water dynamic viscosity in kg
ref
/m.s.
6.2.2 Particle Trajectory Calculation
In the Eulerian-Lagrangian approach, the total flow of the particulate (particles, drops, or
bubbles) phase is modeled by tracking a small number of particles through the continuum
fluid. The trajectory of a particle is calculated essentially by integrating the force balance on
the particle written in a Lagrangian reference frame for the X direction
du
p
= F (u − u p ) + g x ( ρ − ρ ) / ρ + Fx ………………………………..………….……6.11
dt D P
where FD (u-up) = drag force unit particle mass and
18 μ C D Re ……...........................................…………………………..………….……6.12
Pg =
ρ PD 2P 24
Here u = fluid phase velocity; uP = particle velocity; DP = particle diameter; ρP = particle
density; and CD = drag coefficient usually expressed in terms of Re, the particle Reynolds
number defined as
ρ D P u P − u …….......................................…………………………..……………….……6.13
Re =
μ
The drag coefficient CD is evaluated using empirical expressions depending on the Reynolds
number. In the present study, the particles are assumed to be spherical. The trajectory
equations are solved by step-wise integration over discrete time steps. Integration in time of
Eq. (6.11) yields the velocity of the particle at each point along the trajectory, with the
trajectory itself being predicted by
dX
p
=up ……………….….......................................…………………………..……………….……6.14
dt
Integration of the above equation and the force balance equation are carried out in each
coordinate direction to predict the trajectories of the discrete phase.
- 64 -
The final system of particle conservation equations is a linear one, so the superposition
principle can be invoked to estimate the total settling efficiency. The inlet particle size range
is divided in classes with the medium size of each class assumed as its characteristic (pivot).
Then independent simulations are conducted for monodisperse particles in the feed using
every time the individual pivot sizes. The overall settling efficiency can be found by adding
appropriately the efficiency for each particle size.
6.2.3 The influence of particle structure
The settling velocity of an impermeable spherical particle can be predicted from Stokes’ law.
However, the aggregates in the water not only are porous but it is well known that they have
quite irregular shapes with spatial varying porosity. The description of the aggregates as
fractal objects is the best possible one-parameter description of their complex structure, so it
has been extensively used in the literature. The well-known fractal dimension, D, is a
quantitative measure of how primary particles occupy the floc interior space. But the settling
velocity of the aggregate depends on its structure both through its effective density and its
drag coefficient. These variables must be independently estimated for the aggregate shapes
instead of the settling velocity because settling velocity cannot be directly entered to the CFD
code. As regards the drag coefficient of fractal aggregates, ample information can be found in
the literature; from simulation of the flow inside reconstructed flocs using the Fluent code
(Chu et al. 2005) to purely empirical relations. According to Gmachowski 2005, the ratio of
the resistance experienced by a floc to that of an equivalent solid sphere (feff) can be
expressed as follows:
2
f eff ⎛ D⎞
= 1.56 − ⎜ 1.728 − ⎟ − 0.288 ….......................................………………………………….……6.15
⎝ 2⎠
It is found that aggregates generated in water treatment processes exhibit a fractal dimension
ranging between 2.2 and 2.6 (Gmachowski 2005 and Lee et al.1996), so the resistance
coefficient, feff, varies between 0.85 and 0.95. This estimation agrees also with the theoretical
results for the drag coefficient of fractal aggregates given by Vanni 2000 who solved the
problem in the limit of zero Reynolds number.
Flow field computation
Assume a value of apparent density

Correct apparent
density value
Compute sedimentation efficiency for each

particle size
Drive the total sedimentation efficiency and

compare with the experimental one
No match
Match
End
Figure 6.2 Flow chart of computation sequence.
- 65 -
Contrary to the drag coefficient, the effective density cannot be estimated at all. Even if the
structure (and the porosity) of the aggregate is known, the intrinsic density of the primary
particles is not known. As the settling velocity is not so sensitive to feff (the sensitivity with
respect to effective density is much larger, since the difference between effective and water
density determines the settling velocity), the resistance coefficient was fixed at 0.90 and the
apparent density was then estimated as 1066 kg m-3 by requiring the final computed settling
effectiveness to coincide with the measured settling effectiveness of 90%. This is a typical
effective density value for the aggregates met in water treatment applications (Deininger et
al.1998). The flow chart of this computations sequence is presented in Figure 6.2.
6.2.4 Simulation
To limit computational power requirements, the rectangular settling tank was modelled in 2D.
The major assumption in the development of the model is that the flow field is the same for
all positions; therefore, a 2D geometry can be used to properly simulate the general features
of the hydrodynamic processes in the tank. As a first step, a mesh was generated across the
sedimentation tank. A grid dependency study was performed to eliminate errors due to the
coarseness of the grid and also to determine the best compromise between simulation
accuracy, numerical stability, convergence, and computational time. In addition, the mesh
density was chosen such that the grid was finest where velocity gradients are expected to be
largest. The selected grid was comprised of 147,814 quadrilateral elements. Two other grids
(one finer with 318,850 elements and one coarser with 16,175 elements) were also used to
determine the effect of the overall grid resolution on predictions. While the predictions
obtained using the coarse grid were found to be different from those resulting from the
selected one, the difference between the predictions made by the selected and fine grids were
insignificant. As a result, the solutions from the grid of 147,814 quadrilateral elements were
considered to be grid independent.
Figure 6.3 Effect of particles number on the number of iterations required to achieve a
converged solution in Holic and Hrinova STs.
The segregated solution algorithm was selected. The k-ε turbulence model was used to
account for turbulence, since this model is meant to describe better low Reynolds numbers
flows such as the one inside our sedimentation tank (Wilcox 1998). The used discretisation
schemes were the simple for the pressure, the PISO for the pressure-velocity coupling and the
- 66 -
second order upwind for the momentum, the turbulence energy and the specific dissipation.
Adams and Rodi 1990 pointed out that for real settling tanks the walls can be considered as
being smooth due the prevailing low velocities and the correspondingly large viscous layer.
Consequently, the standard wall functions as proposed by Launder and Spalding 1974 were
used. The water free surface was modelled as a fixed surface; this plane of symmetry was
characterized by zero normal gradients for all variables.
As a first step, the fluid mechanics problem was solved in the absence of particles to find the
steady state flow-field. The converged solution was defined as the solution for which the
normalized residual for all variables was less than 10-6. In addition, the convergence was
checked from the outflow rate calculated at each iteration of the run. The convergence was
achieved when the flow rate calculated to exit the tank no longer changed.
Then the particles, whose density and size could be assigned at will, are released from the
inlet and are tracked along their trajectories. The particles reaching the bottom were deemed
trapped whereas the rest were considered escaped. Particle tracking is fast and 20,000
particles could be tracked in less than 10 min once the flow field had been computed.
The number of particles was selected after many trials in order to combine the solution
accuracy with short computing time for convergence. Figure 6.3 reveals that, as the number
of particles increased from 2,000 to 20,000, the number of iterations needed for the model to
converge decreased, whereas an even higher number would not yield any significant
improvement. Therefore, a number of 20,000 particles was selected as a suitable one. The
converged solution was defined as the solution for which the normalized residual for all
variables was less than 10-6. In addition, the convergence was checked from the particle
number balance calculated at each iteration of the run. The convergence was achieved when
the percentage of particles calculated to exit the tank no longer changed.
Table 6.1 Classes of particles used to account for the total suspended solids in the
STs in Holic and Hrinova STs.
Range of particle Mean particle size
Class Mass fraction
size (μm) (μm)
1 10-30 20 0.025
2 30-70 50 0.027
3 70-90 80 0.039
4 90-150 120 0.066
5 150-190 170 0.095
6 190-210 200 0.115
7 210-290 250 0.126
8 290-410 350 0.124
9 410-490 450 0.113
10 490-610 550 0.101
11 610-690 650 0.077
12 690-810 750 0.057
13 810-890 850 0.040
The settling tank was simulated for a specific set of conditions used in the Hrinova and Holic
treatment plants for which the particle size distribution at the inlet and outlet and the total
settling efficiency has been experimentally measured. The inlet was specified as a plug flow
of water at 0.035 m s-1 and 0.065 m s-1 respectively, whereas the inlet turbulence intensity
was set at 4.5%. The outlet was specified as a constant pressure outlet with a turbulence
intensity of 6.0%. The water flow rate was 0.06 m3 s-1 and 0.08 m3 s-1 respectively.
For simulation purposes, the range of the suspended solids was divided into thirteen distinct
classes of particles based on the discretization of the measured size distribution. The number
of classes was selected in order to combine the solution accuracy with short computing time.
- 67 -
Two other numbers, 6 and 15, were tested. While the predictions obtained using 6 classes of
particles were found to be different from those resulting from the 13 classes, the difference
between the predictions made by the 13 and the 15 classes were insignificant. Therefore, a
number of 13 classes were selected as a suitable one. Within each class the particle diameter
is assumed to be constant (Table 6.1). As it can be seen in Table 6.1, the range of particle size
is narrower for classes that are expected to have lower settling rates.
The procedure used to determine the overall settling effectiveness, n, was based on the
calculation of the percent of solids settled for each particle size class, ni:
n
( )
∑ c i ni
i =1 …………………........................................…………………………..……………….……6.16
n=
n
∑ ci
i =1
The settling efficiency for each particle size class was calculated after the conclusion of the
thirteen different sedimentation simulations for the standard and the modified tank,
respectively. In each run, only one particle size class was taken into account; all injected
particles were considered to have the same diameter corresponding to the so called pivot
particle size and assumed to be the average of the lower and upper diameters of the class.
The effectiveness of particle settling is estimated as the percentage of solids settled over the
rate of solids introduced from the inlet.
In this way, to predict the overall percent solids removal efficiency one needs to know only
the particle size distribution in the influent. With this knowledge and the percentages ni
calculated from the 13 simulations, the sedimentation efficiency could be calculated for any
different particle size distribution in the influent.
Figure 6.4 A flocculation channel
6.3 Simulation of Existing Sedimentation Tanks

6.3.1 Hrinova Treatment Plant
The Hrinova water treatment plant uses lime, (NH4)2SO4 and Fe2(SO4)3 to flocculate the iron
before entering the sedimentation tanks. A special feature of the activated sludge system was
the continuous dosing of the potatoes solution at the ending of STs to improve the settling
properties and to adjusted pH value. The tanks have wedge shaped sludge hoppers below the
inlet. The sludge is withdrawn hydraulically from the narrowest part of the hopper. A
flocculation channel is positioned on top of the sedimentation tank at the inlet (Figure 6.4).
The inlet consists of a number of small holes (ports) in the last channel of the flocculator.
- 68 -
Then the water flow through a perforated inlet baffle. The purpose of this baffle is to
distribute the flow uniformly over the entire cross-sectional area of the tank. Each of the
openings has the same size, and they distributed uniformly across the baffle. There are
openings towards the bottom of the baffle, because this could increase scour of the deposited
sludge.
Figure 6.5 The inlet of Hrinova Settling Tank
Figure 6.6 Rectangular settling tanks at the Hrinova Water Treatment Plant.
The clean water is withdrawn from the end of the tank through vertical weir (Figure 6.5). It
was recently observed that significant amounts of sludge are not removed from the
sedimentation tank during desludging as it is washed out of the hopper into the horizontal
section of the tank. This requires frequent decommissioning of the tank to remove the sludge
buildup.
- 69 -
The current study not only focuses on the hydrodynamics of the flow using a relatively
efficient computer simulation model, but also improves the understanding of both flow and
sediment characteristics.
There are 4 rectangular tanks at the Hrinova Water Treatment Plant (Figure 6.6). The length
of the existing tank = 30 m, the tank width = 4.5 m, the average water depth at the tank
midstream part = 3.5 m (Figure 6.7). Each tank’s influent flow is introduced through a weir
and a perforated inlet baffle. In each tank there is one effluent weir (0.70*0.50 m), which
goes along the longitude direction and is distributed in a range of 4.5 meters from the side
wall (Figure 6.7).
Figure 6.7: Diagram of The horizontal settling tank of Hrinova WTP.
Figure 6.8 The outlet weir of Hrinova Settling Tank
Figure 6.9 Picture of effluent from ST

As shown in Figures 6.8, and 6.9 the effluent from rectangular ST at Hrinova WTP,
contained high effluent SS concentrations.
The problem not only prevented the plant from satisfying the discharge standard but also
significantly increased the cost of the planned downstream filtration tanks. A high effluent
- 70 -
turbidity in the ST effluent may require construction of more effluent filters between the ST
and the disinfection tanks in order to lower the overall construction and maintenance cost.
Figure 6.10 Impacts of Diurnal Flow Variations on Settling Tank Effluent
Figure 6.11 Diurnal Flow Variations in Settling Tank Effluent
Figure 6.10 and 6.11 show a typical monitoring profile of ST effluent SS in 24 hours, which
was provided by the project owner. In the figure, the ST effluent quality is with respect to the
system operation time. The field data show the very clear impact of ST flow variations on the
effluent SS concentration. The effluent SS trend indicates that the ST effluent SS was
- 71 -
between 10 to 15 mg/L during the daily low flow period and 15 to 25 mg/L during the daily
high flow period.
It had been found in most of the current process/ST operations that the light flocculant solids
were consistently blowing out in substantial amounts around the existing weir system – even
during the daily low flow period as shown in Figure 6.10. There were obvious flocculant
solids observed on the rest of ST surface area and in the area around the existing effluent weir
(Figure 6.12), which are located at the very downstream end of the ST and cover only
approximately 1/10 of the total ST surface area as shown on Figure 6.7.
Figure 6.12 Picture of Hrinova Horizontal Sedimentation Tank shows the flocculant
solids observed on the rest of ST.
During site visits, the following major features of hydraulic performance have been observed
in the horizontal sedimentation tanks:
1. There is impingement between the tank water surface water and upward influent flows.
2. After the impingement with the water surface, the tank influent dives toward the tank floor
within a distance ranging from 1 to 2 meters away the influent wall due to the density current
formed.
3. The density interface between tank influent jets and ambient clean water in the surface
layers can be visually observed.
Massive amounts of light flocculant solids are consistently blowing out around the existing
weir system - even during the daily low flow period.
Two-dimensional model is being set up and examined in order to simulate the complicated 2-
D tank geometries of real settling tank with the help of the remarkable development in the
computer technology. Especially it can describe more practically inlet ports, outlet weirs on
the water surface of ST.
A number of achievements in simulating the ST have been made by introducing 2-D model.
Figure 6.13 shows the grid generation of the 3-D ST where inlet and weir parts are indicated
as bold lines. The preliminary calculation results are summarized below (e.g. Figure 6.14).
- 72 -
• Strength of surface return flow decreased and waterfall phenomena disappeared on

the inlet section due to the high inlet velocity (increased by two order of magnitude;
high Froude number)
• Formation of small recirculation flow under the vertical top weirs
• High solid concentration in the fore part of baffle
• Small entrainment of ambient fluid to the inlet zone under the baffle lip
• Upward flow near the baffle in the inlet zone
The 2-D model should be further investigated to deal with strong buoyancy effects and the
formulation of mathematical models, which requires long calculation with the difficulty of
robust convergence. For this reason an appropriate turbulent model is required, combined
with a computationally effective solution method, which allows a fast calculation of the
coupled flow and suspended solid concentration fields.
Figure 6.13 3-D grid generation of a ST.
Figure 6.14 Velocity vector on XY area.
The removal efficiency in settling tanks depends on the physical characteristics of the
suspended solids as well as on the flow field and the mixing regime in the tank. Therefore the
determination of flow and mixing characteristics is essential for the prediction of the tank
efficiency. Figure 6.15 presents the predicted streamlines for the existing tank. The displayed
simulations made with solids present refer to particles of class size 4 (Table 6.1). The
influent, after impinging on the standard flow control baffle at point A, is deflected
downwards to the tank bottom. The flow splits at point B on the bottom of the tank,
producing a recirculation eddy at C. Generally, the flow pattern is characterized by a large
recirculation region spanning a large part of the tank from top to bottom where the sludge
gathers before leaving the tank. These regions have a substantial impact on the
hydrodynamics and the efficiency of the sedimentation tank. The same behavior was
- 73 -
observed by Stamou 1991 in his flow velocity predictions in a settling tank using a
curvature-modified k-ε model.
Figure 6.15 Calculated stream lines for existing sedimentation tank.
The above-mentioned observations are in agreement with findings of Zhou and

McCorquodale 1992, who studied numerically the velocity and solids distribution in a
clarifier. According to another numerical work (Deininger et al. 1998), in secondary
clarifiers there is a circular current showing: 1) forward flow velocities in the zone close to
the tank bottom, 2) backward flow velocities in the upper zone of the tank, 3) higher forward
flow velocities in the inlet than in the rim region, 4) higher backward flow velocities in the
inlet than in the outlet region, 5) vertical currents downwards to the tank bottom in the inlet
region, and 6) vertical currents upwards to the water level in the outlet region. For the case of
secondary clarifiers with high suspended solids concentration, a density current exists due to
a higher density of the incoming suspension. This current sinks toward the sludge blanket
right after leaving the inlet structure and flows towards the tank rim. As a result, backward
velocities are induced in the upper water zone following the continuity equation.
6.3.2 Holic Water Treatment Plant
A rectangular sedimentation tank in Slovakia water treatment plants was selected to
demonstrate the response of rectangular tanks to different internal geometries. This case is
based on the Holic settling tanks describe by Ghawi and Kris (2007 a), and Ghawi and
Kris (2007 c). This tank was selected because performance data are available for model
calibration, and because it represents a marginal performance case. Table 5.3 shows the main
tank dimensions and loadings. Figure 6.16-6.20 shows an idealized profile of the Holic
Settling Tank.
Figure 6.16 Schematic representation of the Holic sedimentation tank in 3D.
- 74 -
Figure 6.17 Picture for Holic settling tank
Figure 6.18 Picture of the Hopper
Figure 6.19 Outlet weir
- 75 -
Figure 6.20 Outlet

weir
Figure 6.21 shows the velocity profiles of the existing tanks for a flow rate of 80 l/s and an
inlet concentration of 50 mg/l (~75 NTU). High velocities are present at the inlet (0.065 m/s).
The flow is further accelerated towards the bottom of the hopper due to the density
differences as well as the wedge shape of the hopper. The strong bottom current is balanced
by a surface return current inside the hopper. The velocities near the effluent weir are very
low.
Figure 6.22 Solids concentration profile (g/l) for existing tank

The solids concentration profile is shown in Figure 6.22. Note the high concentration
downstream of the sludge hopper. The sludge that is supposed to settle in the hopper is
washed out of the hopper into the flat section of the tank. Over time a significant amount of
- 76 -
sludge accumulates. Figure 6.23 clearly shows the location and the extent of the accumulated
settled sludge downstream of the sludge hopper.
Figure 6.23 Solids concentration profile (g/l) for existing tank
central axis) in Existing Tank
As shown in Figure 6.24 (a) and (b), the predicted hydraulic regime typically consists of the
upward inlet jet; the influent density waterfall, a bottom density current and a strong surface
reverse flow in the absence of proper baffling. For a case with a thick sludge blanket, the
simulated velocity field showed that the bottom density current deflects upward while near
the tank bottom a strong reverse sludge flow appears. According to both the field
observations and the modelling of the existing process, each of the following reasons (or
combination of them) may cause the ST problems, i.e. the flocculant solids blowing out:
1. The location of the existing weir (distributed in a range of 1 meter at the very
downstream end of the ST) cause very strong upward currents, which could be one of
the major reasons that the flocculant solids were blowing out around the effluent area.
2. The strong upward flow is not only related to the small area the effluent flow passes
through but also to the rebound effect between the ST bottom density current and the
- 77 -
downstream wall. The “rebound” phenomenon has been observed and reported by
many operators as well as field investigators, especially in ST with small amounts of
sludge inventory. A reasonable amount of sludge inventory can help dissipate the
kinetic energy of the bottom density current.
3. In the existing operation, the bottom density current must be fairly strong due to the
lack of proper baffling and the shortage of sludge inventory in the tank.
6.4 Simple Improvements to the Existing Sedimentation Tank
6.4.1 Hrinova Water Treatment Plant
Sedimentation is the removal by settling of particles or flocculated particles which have a
relative density higher than water. The process takes place in a tank where water stands still
or flows very slowly.
Field data collected from the laboratories during the last 3 years was used to analyze the tank
behavior and to enhance the performance of the settling tanks at the Hrinova Water
Treatment Plant. Several tank modifications including flocculation baffle, energy dissipation
baffles, perforated baffles and inboard effluent launders, were recommend based on their
field investigation.
In this study, a recently developed 2-dimensional settling tank model was applied to evaluate
the proposed design modifications of the settling tanks. The model contains a set of
conservation equations for momentum and solids transport.
Fig. 6.25 The model simulated velocities and solids concentration (g/l) distributions for the
original and modifications tanks (in a vertical section along tank central axis) (a. original tank,
b. Perf. Baffle, c. Perf. Baffle and Inboard Launder, and d. Inboard Launder.
Figure 6.25 illustrate the model simulated velocities and solids distributions for the
modifications tanks. The modelled Effluent Suspended Solids (ESS) is summarized in Table
6.2.
- 78 -
These simulations demonstrate that the inboard placement of the launder in a rectangular ST
is a major factor affecting the tank performance (Figure 6.25 and 6.26). The role of the inlet
baffle is important in two aspects: 1) This baffle reduces the re-entrainment of already settled
liquid into the inlet zone, and 2) It provides a zone for flocculation. The perforated baffle
interrupts the bottom density current and helps to redistribute the flow over the tank depth.
However this type of baffle may create its own bottom current which then partially negates
its positive effect. For these trials the best combination was an inboard launder with a
perforated baffle.
Table 6.2 Performance data for modelled settling tank
Perf. Baf. and
Perf. Baffle Inboard
Data Original Inboard Launder
(1) Launder (2)
effluent suspended solids
(ESS) mg/l 25 10 6 3
(1) Perforated baffle distance from inlet = 16m; gap above bed = 0.5 m; height above bed = 1.8 m; porosity =
55%
(2) Length of launder = 12 m.
The modified tank allows flocs to settle at much short distances from the right-hand corner of
the tank. This diminishes the overall settling efficiency of the tank. On the whole, the
simulation results demonstrate quantitatively the drastic effect of particle velocity on
sedimentation effectiveness. Higher settling velocities lead to more effective sedimentation.
However, even small differences in particle settling velocity can cause large changes in the
percent of settled particles.
Figure 6.26 Schematic representation of the Hrinova sedimentation tank with

modification.
At a plant flow rate of 200 l/s and inlet concentration of 50 mg/l (~75 NTU), we were only
able to achieve outlet concentration within the range of 8–11 mg/l (~10 NTU), which was
well above our target (< 5 NTUs) (Figure 6.27). Furthermore we observed qualitatively
sludge formation within the tank due to flocs being broken-up and resuspended into the mass
of liquid in the tank. As a result there was no floc settling in the tank contrary to what has
been anticipated.
Further adjustments needed to be made to the operating and design parameters in order to be
able create sludge within the tank and to observe floc settling in the tank. In order to optimize
the parameters mentioned, we explored the influent velocities, and other configuration, all of
which are adjustable either through the use of an automated process control program or by
manually adjusting the set parameters, which result in a viable suspended sludge layer.
- 79 -
Figure 6.27 A Plot of Flocculent Suspended mg/l vs. Time with The Initial Setup of The
Settling Tank.
Influent Velocity
Another important characteristic that we considered in optimizing the flocculation process in
the plant was the effects of influent velocity to shear and residence times of the particles in
the settling tank. The velocity of the existing tanks for a flow rate of 50 l/s and an inlet
concentration of 50 mg/l (~75 NTU). High velocities are present at the inlet (0.035 m/s). The
flow is further accelerated towards the bottom of the hopper due to the density differences as
well as the wedge shape of the hopper. The strong bottom current is balanced by a surface
return current inside the hopper. The velocities near the overflow weir are low (0.0017 m/s).
These data calculated from laboratory of Hrinova water treatment plant.
At this velocity we were only able to achieve outlet water turbidity between 15 – 30 NTUs
which does not meet our requirement (< 5 NTUs).
The main cause of the high velocities in the exiting tanks is the potential energy at the inlet
(due to the density differences and the height difference between the inlet and the sludge
hopper) that is converted into kinetic energy as the particles settle in the sludge hopper. This
upsets the settled and settling particles in the hopper. The high velocities are further
exacerbated by the presence of the flocculation channels that causes a restriction at the outlet
of the hopper and the wedge shape of the hopper.
Figure 6.28 A Plot of Turbidity vs. Time Showing the Effects of Decreasing Inlet
Velocity and Adjusting the Tank Configurations.
- 80 -
This is because at a higher velocity the particle’s residence time within the tank is shorter (1.5
hour based on calculations from CFD model) since particles are forced out of the tank much
faster. This will affect the sludge formation since the amount of time allowed for the particles
to be suspended in the tank to form flocs is decreased. Furthermore, a higher velocity will
induce shear at the inlet in tank. Shear might cause flocs at the bottom of the tank to deform
and as a result decrease the efficiency of the flocculation process. Therefore, higher velocity
decreases the residence time of the particle in the tank and increases the shear at the tank
inlet. In order to verify this finding, we simulated a CFD model with the same parameters and
tank dimensions as in the previous CFD except for the plant flow rate. We decreased the
plant flow rate by 2/3 to 35 l/s (for one tank) to reduce the inlet velocity and to monitor the
effects of shear and residence time to the flocculation process.
Results from the CFD confirm that we were able to reduce the outlet water turbidity to less
than 6 NTUs (at inlet turbidity of 50 NTUs) at a lower flow rate and inlet velocity (Figure
6.28). The plant had run fairly stable and the turbidity fluctuated between 6 NTUs and 12
NTUs. By decreasing the flow rate of the plant (35 l/s), we were able to increase the
residence time of the particles in the tank to 2 hours (1.5 hours of residence time at flow rate
of 50 l/s). This allowed longer particle interactions to form flocs and sediment to the bottom
of the tank.
6.4.1.1 Behavior of Tanks with Modifications in Hrinova ST
Total of the all alternatives tested by modelling are presented in Table 6.3. The relationship
between the effluent SS and the hydraulic loading is summarised in Table 6.3 for the existing
ST and ones with three different modification combinations. The predicted ESS in Table 6.3
and Figure 6.29 indicates that the average ESS can be significantly reduced by improving the
tank hydraulic efficiency. The comparison of model predictions with the subsequent field
data indicates that the significantly improvement of ST performance was obtained by using
the minor modifications based on the 2-D computer modelling.
Table 6.3 Performance data for modelled settling tank
Q= 50 l/s Q= 70 l/s Q= 80 l/s Q= 80 l/s
Influent conc.= Influent conc.= Influent conc.= 50 Influent conc.=
40 mg/l 40 mg/l mg/l 75 mg/l
Predicted average effluent concentration
Existing tank 20 30 40 50
Modification 1 12 11 30 22
Modification 2 6 8 12 13
Modification 1 and 2 4 6 7 9
(1) Perforated baffle distance from inlet = 16m; gap above bed = 0.5 m; height above bed = 1.8 m; porosity =
55%. (2) Length of launder = 12 m.
Figure 6.29 Comparisons of Solids Distributions between Existing and Modified Tanks
- 81 -
6.4.2 Holic Water Treatment Plant

6.4.2.1 Proposed Tank Modifications
The proposed modifications include: energy dissipation baffles, a flocculation zone,
perforated and non-perforated baffles, modifications to the launders and modifications of
sludge withdrawal facility. In the original project, total 5 of proposed modifications and
combinations of them were tested. The following major modifications adopted in final tank
construction are presented as:
¾ Modification 1 Inlet flocculation baffle, the distance from tank influent to the baffle = 6
m and the baffle depth = 2.5 m;
¾ Modification 2 A perforated baffle between bay A and B with slot space of 54% of flow
cross section area;
¾ Modification 3 A perforated baffle between bay B and C with slot space of 66% of flow
cross section area;
¾ Modification 4 a conventional baffle between bay A and B with baffle depth of 1.75 m
below the surface.
¾ Modification 5 Removing existing surrounding effluent weir and adding 4 to 5 new
launders in the bay C (Figure 6.30 A and B). All effluent launders are aligned with the
tank longitudinal direction. The launders extend from the end wall to the perforated
baffle between B and C.
(A) Four Launders
(B) Six Launders

Figure 6.30 Tank with baffle and launder Modifications(A and B)
- 82 -
(A)
(B)
central axis) in Tank with Modification 1, 2, 3, and 5
The predicted flow and solids fields in the tank with modification 1, 2, 3 and 5 are presented
in Figure 6.31 (a) and (b). The flow pattern shows that influent density waterfall and surface
reverse flow were significantly reduced by the 3 baffles. The flocculation baffle eliminates
most of the entrainment flow from the surface clear water layers into the influent density flow
thus, both the surface return flow along the entire tank surface and the bottom density current
are substantially reduced [see Figs 6.24(a) and 6.31(a)].
The distribution of the sludge blanket among the 3 Bays has been significantly changed by
using two perforated baffles [see Figs. 6.24 (b) and 31 (b)]. In the flocculation zone relatively
minor solids compression takes place in the local sludge blanket. The highest sludge blanket
occurs in Bay A due to the high resistance of the perforated baffle A/B. The lowest sludge
blanket appears in Bay C. The difference of the sludge blanket level between Bay B and C is
relatively small due to the lower resistance of perforated baffle B/C. The predicted flow
pattern and solids field in Figures 6.31(a) and 6.31 (b) show that solids, spill gently over the
baffle slots, at a lower velocity and potential energy head then that in the upstream Bays.
Figure 6.32 includes a distribution port in which the fluid divide uniformly and flow into the
tank. Some of return flow goes outward through top outlets and others form circulating flow
in the settling zone as shown in Figure 6.32. Return flow and upward flow are not clearly
divided, which enlarge the settling zone so that SS have many chances to settle down and
then the removal efficiency reaches over 96%.
Figure 6.32 Flow pattern as the change of inlet configuration Streamline contour
(Uin=0.23 cm/sec, Cin=300 ppm)
- 83 -
6.4 .2.1 Behavior of Tanks with Modifications in Holic ST

Because of the kinetic energy dissipation in the bottom density current due to the perforated
baffles, greater solids compression occurs in Bay B and C. The solids distributions on the
surface layer in the tank with modification 1, 2, 3, and 5 are presented in Figs 6.33. The
simulated solids distributions indicate that the relocated effluent launders can avoid the
sludge blowing out at the two downstream corners. Extending launder length more evenly
distributes the effluent flow.
Table 6.4 Summary of Loading and Effluent Concentration in Tank

Q= 50 l/s Q= 70 l/s Q= 80 l/s Q= 80 l/s
Influent conc.= Influent conc.= Influent conc.= 50 Influent conc.=
50 mg/l 50 mg/l mg/l 75 mg/l
Predicted average effluent concentration
Existing tank 28 34 50 50
Modification 1, 2, 10 12 33 22
and 3
Modification 1, 3, 7 9 25 14
and 5
Modification 1, 2, 3, 5 6 11 9
and 5
The model results in Table 6.4 show that the greatest effectiveness in reducing the effluent
solids using baffles alone was obtained when the tank is operated in stage 1 (the sludge
blanket was lower than 40% of the water depth). For stage 2 operation since the effluent
concentration was primarily controlled by the sludge blanket, the baffle’s efficiency was
dramatically reduced. For most cases the effluent concentration in the tanks with
modifications 1, 2, 3 and 5 can be reduced to less than half of that in the current tank for the
same loading. The results in the tanks with combinations to flocculation baffle and various
inter-bay baffles are also included in Table 6.4. The performance of the tanks with the
intermediate baffles was sensitive to the sludge blanket stage. Exceptionally good
improvement in the tank performance was predicted for both solid (conventional) and
perforated baffles for Stage 1 or low sludge blanket operation; for Stage 1, both perforated
and solid baffles gave about 50% reduction in ESS. However, under high solids loading, i.e.,
Stage 2 or high blanket operation, the baffles became less effective and produced a short
circuiting that in some cases degraded the effluent SS. The perforated baffle appeared to
marginally better than conventional baffle in Stage 1 and significantly better in Stage 2 since
the perforated baffles keep more sludge at upstream of the tank.
The all scenarios cases developed in the study not only provide the project owner a solution
with the best effectiveness but also let them have a choice requiring a very limited cost while
providing reasonable performance enhancement.
The relationship between the effluent concentration and the hydraulic loading is summarized
in Table 6.4 for the existing tank and tanks with 5 different modification combinations. The
predicted effluent concentration in Table 6.4 indicates that the average effluent concentration
can be significantly reduced by improving the tank hydraulic efficiency.
Figure 6.33 indicates that the average effluent concentration can be significantly reduced by
improving the tank hydraulic efficiency. The results demonstrate the effect of use the effluent
launders on the effluent quality. The tank effluent with the highest solids concentration
occurs at the downstream where the upward current, which is focused from both directions, is
much stronger than elsewhere. This interesting phenomenon also was observed by the tank
operator at Holic WTP and they describe is as “Sludge blows out”.
- 84 -
Figure 6.33 Comparisons of Solids Distributions between Existing and Modified Tanks
(1, 2, 3, and 5)
6.5 Proposals Improvement in Holic and Hrinova WTPs

6.5.1 Solids Distribution
It is essential to present first the computed variation in settling efficiency with respect to
particle size. Fig. 6.34 presents the predicted percents of solids settled for different tested
particle size classes (in Holic and Hrinova WTPs). As it can be inferred, the theoretical
settling efficiency tends to nonzero (in fact, relatively large) values as the particles size tends
to zero. This is due to the combined effect of convection (fluid velocity towards the bottom of
the tank, turbulent diffusivity which is independent of the particle size and the perfect sink
boundary condition). In practice it is expected that the settling efficiency decreases as particle
size decreases going to a zero (or close to zero) value for Brownian particles. Although this
inconsistency is not exhibited in the case studied here due to the relatively large particle sizes
of the feed, it must be considered for the shake of completeness of the simulation procedure.
The easiest way to accommodate the realistic behavior of a decreasing settling efficiency as
particles size decreases is by incorporating a particle size dependent trapping probability in
the Lagrangian code. This probability should depend on the interparticle (particle deposits)
interactions, turbulent diffusivity and gravity. As the particles size decreases the effect of
gravity decreases leading the probability from unity to the inverse of the stability ratio well
known to the studies of small particle deposition from turbulent flows (Kostoglou and
Karabelas 1992). Nevertheless, despite the aforementioned improvement, the present CFD
model provides a good overall description of the system behavior.
The percents presented in Fig. 6.34 result in an overall settling efficiency of 88.4 and 98.6%
for the existing and the modified tank, respectively at Holic WTP and an overall settling
efficiency of 84.4 and 98.8% for the existing and the modified tank, respectively at Hrinova
WTP. As it can be seen, the model predicts highly distinct concentration for different classes
of particle; lower removal rate for the smallest and higher removal rate for the heaviest
particles.
Therefore, the improvement in the overall efficiency of solids removal is only achieved by
improving the settling of particles with lower settling velocities (classes 1-5). This
observation is similar to that obtained by Huggins et al. 2005, who used a CFD model to
evaluate the impact of potential raceway design modifications on the in-raceway settling of
solids.
Although the increase in the overall effectiveness seems small it corresponds to an estimated
reduction in the solids exiting the tanks of approximately 520 kg d-1 or to a reduction of about
85% of the solids in classes 1-13 that exit the tank. These values are greater than those
reported by other researchers. Huggins et al. 2005, who tested a number of potential raceway
design modifications, noticed that by adding a baffle the overall percent solids removal
efficiency increased from 81.8 to 91.1% resulting in a reduction of the effluent solids of
- 85 -
approximately 51%. Crosby 1984 used an additional baffle at mid-radius extending from
the floor upwards to mid-depth and observed a reduction of 38% in effluent concentration.
According to Huggins et al. 2005, the particle settling velocity has a significant impact on
the settling efficiency for a given raceway design. These authors argued that an important
consideration in trying to improve the settling of particles is to reduce the mass fraction of
solids with settling velocities below 0.01 m s-1. However, since influent solids load is usually
uncontrollable one should focus instead on the design of a proper baffle, which will improve
solids settling by forcing them to reach fast the bottom of the tank. Fig. 6.35 shows contours
of velocity for the modified tank in two water treatment plants. It must be noted that in this
and following figures only results of simulations with small particles (less than 150 μm) are
presented since for larger particles settling is satisfactory even with the standard tank. Indeed,
in Fig. 6.36 one can see that particle settling velocity increases for such small particles when
using the above modifications.
(A) Holic WTP
(B) Hrinova WTP

Figure 6.34 Predicated percents of solids settled for each particle size class
- 86 -
The effect of modifications is also displayed in Figs. 6.36, 6.37 and 6.38 that show flocs
concentration along the tank bottom. In Figs. 6.36, 6.37 and 6.38 the zero position of the
horizontal axis is set at the right-hand end of the tank bottom. Clearly, the modified tank
allows flocs to settle at much short distances from the right-hand corner of the tank. This
diminishes the overall settling efficiency of the tank. On the whole, the simulation results
demonstrate quantitatively the drastic effect of particle velocity on sedimentation
effectiveness. Higher settling velocities lead to more effective sedimentation. However, even
small differences in particle settling velocity can cause large changes in the percent of settled
particles.
Figure 6.35 Contours velocity (m/s) for the existing and modified settling tank in
Hrinova ST.
Figure 6.36 Flocs concentration (kg m-3) for the standard and the modified tank for
particle class size 2 in Hrinova and Holic STs.
- 87 -
Figure 6.37 Flocs concentration (kg. m-3) for the Standard and the Modified Tank for
inboard launder and a perforated baffle modified for particle class size 3 at Hrinova ST.
Figure 6.38 Flocs concentration (kg.m−3) for the standard and the modified tank for
particle class size 4 in Hrinova and Holic STs.
(a) Concentration field of SS
(b) Streamline contour

Figure 6.39 Calculation results in Hrinova WTP.
- 88 -
6.5.2 Comparison of Settling Models

The flow field is significantly changed even by small density differences. Individual settling
properties of SS are also critical for the accurate prediction of removal efficiency and solid
distribution. SS distribution in the settling tank is calculated from settling models. Results of
solid distribution with DPM are shown in Figure 6.39 at Hrinova WTP. The DPM predicts
high distinct concentration for different classes of particle, which estimates lower removal
rate for the slowest particles and higher removal rate for the heaviest particles. However, it
yields a removal efficiency of 98.4% for the average solid distribution. Figure 6.40 shows the
calculation results by using the DPM at Holic WTP, in which large highly-settleable SS are
removed from the system before the small ones and stratified obliquely to the flow direction
(Takacs et al., 1991). From comparing the streamlines shown in figure 6.39 (b) and figure
6.40 (b) it is understood that a little strong upward flow is formed in the withdrawal zone for
Holic WTP but the most of flow pattern is similar throughout the ST. Therefore the SS is
removed in 99.3 % of the efficiency that is higher than Hrinova WTP. Comparison of two
different settling tanks shows that solid distribution is mainly affected by settling properties
of individual particles and the flow pattern in the outlet region is affected by the coupling of
momentum and density. Hence, the value of settling velocity can be a key parameter to obtain
reliable predictions of the removal rate.
(a) Concentration field of SS
(b) Streamline contour

Fig. 6.40 Calculation results in Holic WTP.
6.6 Temperature Effect

6.6.1 Temperature Measurements
The inflow to the ST could be distinguished from the colder ambient water by using
temperature as a tracer. The inlet baffle for each tank caused the inflows to enter initially as
an axial submerged buoyant jet.
Experimental data indicated that surface heat loss between the tank inlet and outlet caused
temperature losses ranging from 0.5oC to 1.2oC. The inlet tank water temperature varied from
1oC to 4.3oC.
- 89 -
Figure 6.41 Temperatures for Hrinova sedimentation tanks on 13/02/06 near the surface
(0.3 m depth) at distance 1 m, 9 m, and 29 m from the inlet of the tank.
Figure 6.42 Temperatures for Hrinova sedimentation tanks on 13/02/06 (2.5 m below
the surface) at distance 1 m, 9 m, and 29 m from the inlet of the tank.
The first study was performed on the Hrinova ST on 13/02/06. Most water temperatures
ranged between 2.6oC and 2.2oC. For about 1.5 hours of the 2 hour study period, temperatures
at a depth of 2.5 m and at a distance of 9 m were consistently about 0.5oC greater than in
other areas of the tank.
- 90 -
These data show that the inflow temperature was often warmer than that of the water on the
tank surface. The inlet baffle caused the inflow to enter as a density underflow. The buoyancy
of the inflow became more dominant as solids settled and momentum diminished. Figures
6.41 and 6.42 demonstrate clearly the warmer bottom temperatures and cooler surface
temperatures by comparing temperature as a function of time and position at depths of 0.3 m
and 2.5 m, respectively. The temperatures were always warmest at the position of 9 m, and
coolest near the outer walls of the tank. The effect of summer temperatures on settling
velocity are discussed in the next section
6.6.2 Settling Velocity Correction Factor
The traditional discrete settling model proposed by the Stoke’s law suggests that the settling
velocity of discrete particles depends indirectly on the temperature of the fluid since the
settling velocity is inversely proportional to the kinematic viscosity of the liquid. A similar
relationship is presented in the compression rate model proposed by Kinnear (2002) in
which the settling velocity is inversely proportion to the dynamic viscosity of the fluid. The
model proposed by Stoke’s law and Kinnear (2002) indicate that the settling velocity of
discrete particles and the compression rate of the sludge are influenced by the temperature of
the mixture.
Table 6.5 Settling velocity and dynamic viscosities for summer and winter temperature.
CFD Calculated at summer temperature
SS mg/l Settling velocity Inlet temperature Outlet temperature Dynamic viscosity
o o
m/h C C kg/m.s
60 1.5 27.5 27.5 8.5e-04
50 1.7 27.5 27.5 8.6e-04
25 1.83 26 26 8.7e-04
15 2.52 25.4 25.4 8.8e-04
CFD Calculated at cooled temperature
60 0.95 8 9.2 1.3e-03
50 1.05 6.6 6.8 1.35e-03
25 1.9 7.8 8.8 1.29e-03
15 2.7 7 8.9 1.30e-03
In order to define a correction factor for the settling velocities based on temperature
difference, the temperature effect on the zone settling velocity has to be determined.
Table 6.5 presents the settling velocities calculated by CFD at two different conditions: (a)
zone settling velocity calculated at summer (field conditions) temperature, and (b) zone
settling velocity calculated at a cooled temperature. This table also shows the temperature of
the sample at the beginning and at the end of the calculation, suspended solids concentration,
and the respective dynamic viscosity of the mixture calculated using CFD model.
Assuming the viscosity as the only variable in the models proposed by Stoke’s law and
Kinnear (2002), the settling velocity of discrete part and the compression rate of activated
sludges can be expresses as:
K
VS = …………..……………………………………..………………………………………………6.17
μ
where K is a constant independent of temperature but dependent on all other parameters
affecting settling.
For a fixed value of K in Equation 6.17, and two different temperatures T1 and T2, Equation
6.17 can be rearranged as:
VST1 μ T1 = VST2 μ T2 = K …………..………………………………………………………………………6.18
- 91 -
where VsT1 and VsT2 are the settling velocities at temperatures T1 and T2 respectively, and μT1
and μT2 are the dynamic viscosities of the mixtures at temperatures T and T respectively.
1 2
Figure 6.43 displays graphically the value of the relationship VsT2 / VsT1 and μT2 /μT1 for the
data presented in Table 6.5 at temperatures Ts (summer temperature) and Tw (winter
temperature).
From Figure 6.43 can be observed that the numerical values of the ratios VsT2 / VsT1 and μT2
/μT1 are very close, suggesting that an easy correction in the zone settling velocity for
different temperatures can be made with a correction factor based on the dynamic viscosity of
the water at the two temperatures. Figure 6.44 shows an extended data set indicating the
relationships between the ratios VsT2 / VsT1 and μT2 /μT1.
Figure 6.43 Ratios of VsT2 / VsT1 and μT2 /μT1 for Different suspended solid (SS) concentrations.
Fitting a straight line to the data point presented in Figure 6.44 and using Equation 6.10 can
find a correction factor for the settling velocities based on temperature
⎛ ⎡⎢⎣ T 1247 .8 ⎤
⎥ ⎞
⎜ 10 ⎟
+133.15 ⎦
VST2 = VST1 ⎜ ⎡ 247.8 ⎤ ⎟ …………..………………………………………………………………………6.19

⎜ 10 ⎢⎣ T 2 +133.15 ⎥⎦ ⎟
⎝ ⎠
Figure 6.44 Effect of Temperature on Settling Velocity.
- 92 -
Equation 6.19 can be applied to correct the settling velocities for difference in temperatures
in whichever of the four types of sedimentation, i.e., unflocculated discrete settling, and
flocculated discrete settling. Even though equation 6.19 can be used for a sensitivity analysis
on the performance of the model for different seasons, e.g. summer and winter, there is no
evidence that the settling properties can be accurately extrapolated from one season to
another.
The effects of the influent temperature variations on the hydrodynamics and the performance
of the settling tank were evaluated by simulating the Hrinova ST with an influent temperature
difference of ± 1ºC, and two different inlet concentrations, i.e. 30 and 75 mg/L. The low
concentration was selected in order to simulate loading conditions similar to those found in
Hrinova and Holic WTP. The simulations were set up for summer and winter conditions.
Tables 6.6 shows the general data used in the simulations. Table 6.6 also shows the peak ESS
found during the simulations. The values presented in Table 6.6 represent two specific sets of
data for the City Hrinova, and Holic. These values were obtained from the WTP laboratories,
the winter data corresponds to Winter/2006 and 2007, and the summer data corresponds to
Summer/2006 and 2007.
The values of the peak ESS presented in Table 6.6 indicate that the influent temperature
variations have an important effect on the performance of the SST.
Table 6.6 Predicted ESS values for Different Temperature Variations

Influent temperature Peak ESS
Heat exchange
variation (mg/l)
+1 OC Summer 36.24
-1 OC Summer 16.25
0 OC Summer 16.05
0 OC Winter 33.65
+1 OC Summer 23.90
-1 OC Summer 41.19
0 OC Summer 41.00
0 OC Winter 21.19

Distribution A) time 120 min, B) 220 min. at ΔT= +1ºC, SS= 75 mg/L
- 93 -
Figure 6.45 shows the temperature changes in the ST with the warmer influent for the
summer conditions. This figure shows that the surface heat exchange warms up the surface
water. Since this is a stable stratification gradient there is little mixing between the surface
and the inner layers. This figure also shows the incoming water plume traveling near the
bottom until it rises close to the end wall. A study presented by Wells and LaLiberte (1998)
showed similar results to those presented herein.
The case with the cooler influent shows a rise of the ESS when the cooler water enters the
settling zone, but the rise is much smaller than the case with the warmer influent. In this case
the cooler influent makes the inflow even denser which strengthen the density current. This
effect is dissipated as the cooler water keeps entering and the difference in temperatures
decreases. The generation of a stable stratification of the vertical density gradients seems to
suppress the turbulence in the vertical direction and to damp the diffusion of the suspended
solids.
The simulations with the lower SS (30 mg/L) and the warmer influent (ΔT= +1ºC) shows a
different flow pattern. In this case the ST shows a rising buoyant plume that changes the
direction of the density current, i.e. previous to the change in temperature the density current
rotates counterclockwise and then it changes to a clockwise rotation. The warmer influent
impacts the baffle and is deflected downward. Immediately after passing below the baffle the
flow shows a strong rising plume which reaches the surface and develops a surface density
current. The surface density current travels at the surface, impacts the end wall and is
deflected downward, and then is recirculated as an underflow current. As the temperature in
the tank becomes more uniform, the counter flow becomes weaker, eventually returning to
the counterclockwise density current dominated by the suspended solids. This flow pattern
can be observed in Figure 6.46. Similar results to results to those presented herein have been
presented by McCorquodle (1987), Godo, and McCorquodle (1991) in studies conducted
with physical models and full scale facilities and also by Zhou et al. (1994) in numerical
simulations.

Distribution A) 120 min, B) 220 min. at ΔT= +1ºC, SS= 30 mg/L
In Figure 6.48, there is a stable density interface between the cooler fluid at the bottom and
the warmer fluid at the top. Little mixing is observed between these two layer, and the cooler
water is withdrawn from the tank through the hopper. A similar stable-sharp interface is
observed at the top of the tank between the surface water and the cooler water below it.
- 94 -
The results presented in Figures 6.45 and 6.46 indicate that the performance of the ST
strongly decreases under the influence of the surface cooling process presented during winter
conditions. These results agree with the findings of Larsen (1977), Wells and LaLiberte
(1998) and Kinnear (2004) who found that the removal efficiency of settling tanks may vary
over the year with a minimum during the winter season.
When the surface water becomes cooler it develops an unstable stratification density gradient
promoting the mixing with the lower layers. As the cooler-denser water coming from the
surface penetrates the tank, the counter clockwise density current carries the denser water
towards the inlet region, where it impacts the baffle and is deflected downward. This denser
plume mixes with the warmer influent, warming up and travelling with the bottom density
current. This cycle creates a radial density gradient that promotes positive vorticity in the
settling zone reinforcing the strength of the density current and increasing the suspended
solids carrying capacity of the upward current. Close to the ST outlet, the current towards the
effluent weirs obligates the surface denser water to pass below the baffle creating a density
gradient in the opposite direction to the main gradient developed in the settling zone, this
condition creates a small eddy that rotates in the opposite direction to the main density
current. This eddy is apparently counteracting the upward flow of suspended solids, but its
effect is too local and small to avoid the high ESS at the outlet.
for an Influent Temperature Equal to 26.5ºC
Figure 6.47 shows the unstable temperature stratification that develops under the surface
cooling process. The effect presented herein might have been exaggerated for the relative
warm influent for “winter” conditions, i.e., 26.5ºC. To evaluate the conditions for a cooler
influent, the test was repeated using a constant influent temperature equal to 11.0ºC. Figure
6.48 shows the temperature field pattern for this influent temperature; this figure shows the
same pattern observed in Figure 6.47. In this case the peak ESS was 27.15 mg/L, 10 mg/L
higher than the “summer” conditions, which supports the statements about the effects of the
surface cooling process presented in the previous paragraphs.
for an Influent Temperature Equal to 11.0ºC.
- 95 -
6.7 Water Treatment by Enhanced Coagulation Operational Status and

Optimization Issues
6.7.1 Correlations of Single Variable versus Effluent Fe
Figure 6.49 presents the CFD model result for system Effluent iron versus ESS concentration.
The results indicate somewhat of a correlation between effluent Fe+3 and ESS. It can be
observed that as the ESS concentration of process increases, the effluent Fe+3 concentrations
has the similar increasing trend. However, the wide spread of data points means that more
parameters other than ESS concentration could affect the system effluent Fe+3.
Figure 6.50 presents the system ESS concentration with respect to Coagulant Dosage (mg/L).
The Hrinova water treatment plant uses lime, (NH4)2SO4 and Fe2(SO4)3 to flocculate the Fe+3
before entering the sedimentation tanks. The coagulant dosage is in the range of 7 to 8 mg/L
in the water treatment plant. There is no strong correlation between coagulant dosage and
sedimentation ESS concentration in this dosage range. Obviously, the Settling Tank (ST) ESS
concentration is affected by more parameters, such as ST operation conditions, ST hydraulic
efficiency and solids settling properties, etc. In the tested range of coagulant dosage (3~5
mg/L), coagulant dosage may have some impact on solids settling properties. However, the
direct impact of coagulant dosage on ST effluent SS concentration cannot be observed in the
experimental field dosage range. Therefore the CFD model has been used to study this
impact.
Figure 6.49 Correlation between effluent Fe and ESS concentration.
Figure 6.50 Correlation between ESS and Coagulant dosage
- 96 -
Figure 6.51 presents the correlation between effluent iron and coagulant dosage. The high
correlation between effluent Fe+3 and coagulant dosage can be observed. The model results
indicate, as expected, that as coagulant dosage increases the effluent Fe+3 decreases. The
fairly wide data spread indicates the effluent Fe+3 must be affected by some other parameters
too, probably including form and concentration of Influent iron, alkalinity and pH.
Figure 6.51 Correlation between effluent Fe and flocculation chemical dosage
6.7.2 Correlations of Two Variables versus Effluent Fe

The effluent Irons may be related to both coagulant dosage and process influent Irons
(amount of irons entering into system). Figure 6.51 presents the correlation between the Irons
removal [(Fe- Feo)/ Fe] and dosage, where Fe and Feo present the Influent and effluent Irons
concentrations, respectively. It can be concluded that the higher dosage gives higher Irons
removal percentage. Nevertheless, the very large variation spread of the data in Figure 6.52
means that more factors should be simultaneously included in the analysis of Iron removal.
Figure 6.52 Correlation between Fe removal and chemical dosage
- 97 -
6.7.3 Correlations of Three Variables Versus Effluent Fe

This study assumes that the final effluent irons of system could be directly affected by
following three major variables as Coagulant dosage; settling tank ESS concentration; and ST
Influent Irons concentration. It implies that the ST flow condition, hydraulic efficiency,
Influent SS concentration and solids settling properties, etc. could impact the ST effluent SS
thus, indirectly impact the process effluent Fe+3. To reduce the number of variables, two
dimensionless variables have been assumed: the fraction Irons removed (removal ratio) and
the relative coagulant dosage. The relative coagulant dosage is defined as Dosage/Effluent
SS, based on the concept that the higher dosage normally gives lower effluent irons and
higher effluent SS always generates the reverse impact on effluent Irons.
If there is a relationship among the four variables and a certain variable change could be
found in the four-dimensional system it should make the same set of data (presented in
previous Figures 6.49 to 6.52) converge to a line (or a curve).
In this study a composite variable (Dosage/Removal Ratio) has been found which allow as all
of data (collected by Hrinova WTP in 2006 and 2007) to converge to an almost perfect line
for the Irons removal as shown in Figures 6.53.
Figure 6.53 Correlative relationships between relative dosage and dosage/Fe removal
ratio revealed in this study
6.7.4 Coagulant Selections
Figure 6.53 shows the convergent linear relationship between a relative coagulant dosage and
the Dosage/Removal Ratio, which is given as [Relative Dosage – Fe removal] divided by Fe
removal. The test data clearly indicates that both (NH4)2SO4 and Fe2(SO4)3 have almost
identical tangential rate and intercept distance in Figure 6.53. According to the data analysis,
it can be concluded that both coagulants used in the CFD model have very similar
performance in terms of irons removal efficiency. It also appears that the iron salt may have
given a somewhat better performance for suspended solids removal. To obtain a higher irons
removal efficiency it is more important to properly choose a coagulant dosage rather than to
select which coagulant between the (NH4)2SO4 and Fe2(SO4)3 by Fe removal.
6.7.5 Irons Removal Model

As shown in Figures 6.53 and 6.54, a linear equation between Relative Dosage and
Dosage/Removal Ratio has been obtained by using a simple linear regression, which is:
- 98 -
[Dosage/ STESS – (Fe- Feo)/ Fe]/[( Fe- Feo)/ Fe] = 1.0992 (Dosage/ STESS) – 0.5855
………………………………………………………………………………………………………………….6.20
Where Dosage = Coagulant dosage for either (NH4)2SO4 and Fe2(SO4)3

Fe = System Influent Irons
Feo = ST effluent Irons
STESS = Settling tank effluent SS concentration
The dimensions of the model (formula) developed above are balanced. Therefore, there are
no unit specifications for any variables in the model as long as the same units are used.
As shown in Figure 6.54, the final model for Irons removal can be written as:
(Fe- Feo)/ Fe = Dosage /(0.4145* STESS + 1.0992 Dosage) ………………………………….6.21
Figure 6.54 Impact of Settling Tank Effluent SS and Flocculation Chemical Dosage on
Irons Removal Efficiency [(Fe – Feo)/Fe = Dosage /(0.4145*SS + 1.0992*Dosage)]
6.7.6 Applications of Fe Removal Model

The model has confirmed a meaningful principle for Irons removal in a ST process, i.e. the
Irons removal is largely dependent on the solids removal. For example, if a ST effluent SS
could be controlled around 2 mg/L, the dosage would be only 3.5 mg/L for a given Irons
removal of 80%. However, if the effluent SS went up to 60 mg/L the dosage requirement
would be higher than 12 mg/L to obtain the same Irons removal. On other hand, if an
economical dosage of 3.5 mg/L were to be used, a ST producing an effluent SS concentration
of 2 mg/L would be expected to have an Irons removal of 80%. However, for the same
dosage, the Irons removal percentage in an operation with effluent SS of 6 mg/L will
dramatically reduce to about 56%. Bottom line: for a given desired level of Irons removal, the
lower the ESS the better.
The Irons removal model here combined with a CFD based process/ST model could be a very
powerful tool to optimize a ST design/operation with respect to following aspects:
1. Revealing the mechanism of Irons removal
2. Determining optimized operational strategies, which can used to achieve a best cost-
effective Irons removal;
3. Providing a ST design with a higher hydraulic efficiency, which can provide better solids
removal thus, potentially significantly minimizing the demanding of coagulant dosage.
- 99 -
For example: the computer modeling results (see Figure 6.29) illustrate that if the ST effluent
SS concentration could be controlled at a low level (1~6 mg/L) due to the ST design
optimization, the coagulant dosage variations have very limited impact on Irons removal
percentage. In this case, the economical coagulant dosage (4 mg/L) should be used to
significantly save usage of coagulant. The design optimization also reduces the burden of
sludge disposal and minimizes the impact of solids disposal on the environment while
maintaining an adequate Irons removal efficiency. Using the principle the existing design
dosage could be reduced more than 50% in the plant operation (approximately 170 kg/per
day).
On another hand, if the ST effluent SS went up to a high level (50~60 mg/L) due to some
uncertainties, the increase of coagulant dosage (up to 10 mg/L) could be an effective way
temporally applied to maintain the Irons removal around 75%.
As shown in Figure 6.55, in the operations with an influent SS of 100 mg/L, the ST capacity
approximately increases from 1.7 to 4.3 m3/m2/h for a recommended SS discharge standard
of 8 mg/L due to application of the optimized influent baffles. The modification gives about a
39% increase of ST capacity. For the effluent SS of 8 mg/L, as used in the base design, the
application of the modification is able to increase the ST capacity from 3.0 to 5.0 m3/m2/h,
providing about a 25% increase in ST capacity.
Figure 6.55 Settling tank flow capacity in base design and optimized alternative
The operation of Hrinova would need about 170 kg chemicals everyday for the design dosage
of 7 mg/L. According to the results of the Irons removal model, for a given Irons removal
percentage, the lower solids effluent concentration in the ST could substantially reduce the
coagulant dosage demanding in a routine operation.
As shown in Figure 6.55, for the Surface Over Flow Rate (SOR) of 3.2 m3/m2/h and influent
SS of 100 mg/L in the optimized ST the effluent SS concentration is approximately 3 mg/L,
which satisfies the SS effluent standard. However, if this SOR value were selected as the ST
design SOR, the coagulant dosage would be 4 mg/L for the given Irons removal of 80%. To
reduce the operational cost, the design SOR could be selected as 3.5 m3/m2/h in order to
control the effluent SS below than 5 mg/L. In this case, the coagulant dosage needs only 3.5
mg/L. The cost for coagulant and solids disposal could be dramatically reduced.
6.8 Modelling the Scraper Mechanism
Despite the fact that solids removal mechanisms are a major consideration in the design of
ST, the function of the scraper in sludge removal is still under debate. The settling tanks at
Hrinova and Holic operate with scrapers (Figures 6.56, 6.57). Figure 6.58 shows the proposal
scraper to modified sludge removal in Hrinova and Holit WTP. Traditionally, it has been
- 100 -
accredited with the function of transporting the settled sludge towards the hopper (e.g.
Albertson and Okey, 1992); but, lately such function has been questioned. Some researches
have found out that typical scraper velocities are not conveying the solids, but are merely
resuspending it. In this matter, McCorquodale (2004) believes more research is required in
order to determine the effectiveness of scrapers. An own conclusion about this topic is
expected to be obtained with this research.
Here are some simulation results: These are the concentration contour at different moment.
The scrape is moving toward the inlet at speed of 0.008m/s. The density current hit the
moving scrape and headed toward the effluent.
Figure 6.56 Picture shows Hrinova ST scraper
Figure 6.57 Picture shows Holic ST scraper
Figure 6.58 ST scraper for Holic and Hrinova
- 101 -
Figure 6.59 Effect of scraper on solids concentration profiles (Volume Fraction)
- 102 -
Consequently, the gravitational (and laminar) flow along the bottom, which may go up to 8-
15 mm/s near the sump, is blocked for 40 minutes of scraper passage. This is clearly seen in
Figure 6.59. The scraper blade thus constrains the bottom flow discharge by counteracting the
gravitational force. Near the floor the velocity increases with height in the shear flow region,
but is obviously limited by the scraper’s velocity.
The sludge is obviously expected to gravitationally accelerate after the scraper has passed the
modelled transect in the time period 40-50 minutes. The solids blanket height should
therefore decrease again due to enhanced sludge removal. Two-dimensional simulations of
the settling tank, however, show this is only the case for water viscosity (Figure 6.59).
In accordance to these fundamental calculations, Figure 6.59 allows to conclude that scrape is
important in the settling process and play a big role in changing the flow field. Dynamic
mesh approach can simulate the scrape movement and its actual effect. Moving wall velocity
approach is an averaged approximation of the effect of the scrape. Averaged on many runs,
this approach may have the similar effect. But for the process in one particular run of scrape
movement, it is not capable. Momentum source/sink approach only considers the force
balance between scrape and fluid (sludge). Disregarding the boundary conditions on the
scrape makes it not correct for scrape modelling.
Table 6.7- The conventional design procedure according to van Duuren (1997) with its
corresponding assumptions and deviations from real tanks.
Typical design procedure
Step Calculate Assumption Deviation from assumption
Step 1 Theoretical settling velocity, Spherical particles Flocs are not spherical
based on Stoke’s viscosity law Laminar flow Flow in tank is turbulent
Downward movement Flow is primarily horizontal
Step 2 Calculate theoretical tank Vertical plug flow Flow is not uniform
surface area Uniform (average) tank Tank behaviour complex and location
behaviour specific
Step 3 Adjust tank area for non-ideal Assume dispersion is the Dispersion is not the reason for non-
behaviour (dispersion) cause for non-ideal behaviour ideal behaviour.
Assume a level of dispersion Dispersion dependent on solid and
hydraulic loading.
Can only be determined after the tank is
built, unless tank is a copy of an
existing tank.
Very difficult to determine a priori.
Step 4 Select L/W ratio and calculate Selected L/W ratio is optimal
tank width and length
Step 5 Select tank depth Selected tank depth is optimal Cannot be determined a priori with
Step 6 Select tank weir overflow rate, Selected tank weir load and conventional approach.
calculate weir length and weir position is optimal
choose position
Step 7 Add sludge removal device Selected sludge removal
location and rate is optimal
6.9 Design procedures and guidelines

The design procedures are necessarily based on many assumptions, not normally stated as
shortcomings and limitations during the design process. To demonstrate the implications of
these assumptions and the way in which these assumption deviate from real tanks, a typical
design procedure is given in Table 6.7.
- 103 -
Many improvements have been made to the above procedure. One such example is the
procedure proposed by Kawamura (2000). Given the flow rate, the floc type, settling rate
and the expected settled turbidity, the design process as listed in Table 6.8 can be applied.
The approach of Kawamura (2000) has already addressed some deficiencies not addressed
by Van Duuren (1997). His approach already shows a more balanced approach taking into
account some of the intricacies of a sedimentation tank, such as the tank inlet. There are,
however, a few areas where his method uses parameters unrelated to the performance of
sedimentation tanks such as the Re and Fr numbers.
The biggest shortcomings of both the conventional processes discussed above are:
¾ The use of average values.
¾ The neglect of specific guidelines for tank inlets.
¾ Inappropriate use of non-dimensional parameters, such Re and Fr, as guidelines
for tank stability and performance.
Table 6.8 Adjusted design procedure according to kawamaura (2000) with its
corresponding assumptions and implications.
Improve design procedure
Step Calculate Assumption Deviation from assumption
Step 1 Choose number of tanks
Step 2 Use empirical settling velocity to Vertical plug flow Flow is not uniform
calculate theoretical tank area Uniform tank behaviour Tank behaviour complex
(include a safety factor)
Step 3 Select tank depth Selected tank depth is May not be optimal
optimal
Step 4 Select L/W ratio and calculate tank Selected L/W ratio is optimal May not be optimal
width and length (check ratios and
adjust)
Step 5 Calculate Re and Fr. Add Re is an important criteria Fr Re & Fr not of primary
longitudinal baffles if Re > 18 000 is an important criteria importance due to dominating
and Fr < 10-5 density effects
Step 6 Design inlet ports and baffle wall, Selected inlet configuration is Specific configuration may not be
if required, based on port area, optimal optimal although headloss and
headloss and port velocity port velocity are within guidelines
Step 7 Select tank weir overflow rate, Selected tank weir load and May not be optimal
calculate weir length and choose weir position is optimal
position
Step 8 Add a sludge removal device location and rate is optimal May not be optimal
Selected sludge removal
The CFD based model does not replace the conventional approach, but can most certainly
enhance the conventional design process. CFD does not have to resort to crude assumptions
relating to the flow, the geometry and non-dimensional parameters. CFD can directly assess
the tank performance. Some of the deficiencies of the conventional approach can be
overcome as suggested in Table 6.8. It should be noted that some of the limitations
mentioned in the previous table are carried into the CFD model. Of particular note is the
settling velocity. Modellers should be aware of the inherent limitations when using the
settling velocity concept. In order to assess the extent to which the settling velocity and
sludge density can distort the CFD solution a sensitivity analysis can be performed. This can
assist the modeller to determine what impact on the model results if the settling velocity was
for instance over-predicted or under predicted.
The sensitivity analysis does not overcome these limitations, but it can provide the modeler
with a better ‘feel’ for the relative importance of one modelling parameter in relation to the
- 104 -
complete CFD model. In this dissertation we tried to improve design procedure as show
in Table 6.9.
Table 6.9 - Proposed CFD enhanced design procedure.
Improved design procedure
Step description
Step 1-8 As per table 6.8 for initial sizing.
Step 9 Measurement of settling velocity and sludge density
Step 10 Set up of computational grid
Step 11 Simulate tank
Step 12 Evaluate results and check for evidence of the following:
- short circuiting
- high velocities zones
- high overflow concentration
- poor sludge removal
Step 13 If none of the above is present, tank size can be reduced to reduce capital cost.
If problems are evident, adjust the design by adjusting the:
- inlet
- position of sludge withdrawal
- position of overflow launders
Also consider using perforated, porous and deflecting baffles
Step 14 Repeat until satisfactory tank geometry is obtained and check final geometry for various
process changes such as density, concentration and inflow rate.
Step 15 Asses the influence of the settling velocity and sludge density input parameters .
6.10 Validation of the Model

After the calibration of the model a validation process was carried out. The validation process
involves comparing the model response to actual measured data. The model was validated
using measured data from the Hrinova and Holic WTP. Up till now, the CFD model of the
settling tank under study has been developed and described. To investigate the accuracy of
the CFD model, however, simulation results should be confronted with measurement data.
Depending on which part of the model is to be validated different data sets may be used.
Firstly, the simulated solids concentrations should be validated with measured profiles. If
unsteady simulations are performed, also some measurements in time should be conducted at
e.g. inlet, outlet, underflow and some other measurement points located inside the settler.
Secondly, to validate velocity profiles several possibilities exist. In this work, solids
concentrations, and velocity profiles were injection were utilized for validation.
Table 6.10 Comparison of model predictions with field data
Operation
Effluent concentration (mg/l) and improvement
conditions
Ave. flocculent Baffle and
Ave. flow baffle
concentration No Modifications launder
l/s Modifications
mg/l Modifications
Field data
February- 35 50 15 - -
April, 2007
Model
34 48 14 6 5
predictions
6.10.1 Holic WTP

The average values of effluent concentration of improvement are presented in Table 6.10.
The tank operation conditions in the data collection periods (February-April, 2007) and the
CFD model predictions are very close as shown in Table 6.10. The comparison of model
predictions with the subsequent field data indicates that the significantly improvement of tank
- 105 -
performance was obtained by using the minor modifications based on the 2-D computer
modelling.
Figure 6.60 Validation of the ESS Simulated by the Model
After the development of the hydrodynamic model, and turbulence model, the ST model was
tested. The ESS predicted by the model was tested during seven days (from a 10 day period)
showing a very good agreement with the field data. Figure 6.60 shows the comparison of the
model prediction and the field data during the aforementioned seven days trial.
The first validation consisted of comparing simulations with velocity profiles measured at 7
different distances, i.e. 5, 10, 15, 20, 25 and 30 m
Figure 6.61 Comparison between predicted and measured Velocity Profiles.
- 106 -
Densimetric Froude number similarity coupled with appropriately scaled physical properties
should give very similar velocity within the tank and in the two dimensional simulation the
results are identical.
Comparisons with experimental velocity data taken at the center plane of the tank at times
2500s, 4500s and 6500s are given in Figure 6.61.
Considering Figure 6.61 it can be observed that;
1) The sludge bed level is rising throughout the simulation. It is at approximately the correct
level at 4500s for the first four stations but fails at the later stations to reach the experimental
bed height at any time. This indicates that the settling velocity may be high, causing
suspended solids to be deposited in the first half of the main body of the tank.
2) The peak velocity and general shape of the velocity profile are well predicted at the 5m
station. After that, the predicted peak velocity gradually exceeds the experimental results,
although the depth of the current is well predicted at the 10m, 15m and 20m stations. At the
last station, where the experimental velocity indicates the density current had been dissipated,
the simulation indicates that it still persists.
6.10.2 Hrinova WTP

Solids concentration profiles were investigated on June and August 2006. Samples were
taken hourly at the following settling tank locations,
• Inlet,
• Effluent, and
• 7 Station inside the tank
Solids concentration profiling was conducted between 1:30 pm and 7:30 pm when the most
stable inlet flow rates occurred. Confrontation of measured profiles and data from the
mounted acoustic solids blanket depth sensor revealed that the latter blanket depth
corresponded to solids concentrations of between 15-25 mg/l (see Figure 6.62). Hence, the
threshold of the blanket sensor set by experience corresponds rather well with the prevailing
effluent quality standard. Consequently, the solids blanket depth in time may be utilised for
validation as well.
Figure 6.62 Detail of measured and simulated solids concentration profiles.
Simulations of a settling tank system with appropriate initial conditions were conducted. The
initial state was obtained by subjecting the settling tank to a 2-day dynamic inlet solids
concentration; inlet flow rate and underflow rate (see Figure 6.63). A supplementary 2-day
simulation showed that identical system states were obtained after 2 and 4 days. For each
- 107 -
validation exercise, solids concentration data were collected every 15 minutes from
simulations.
The first validation consisted of comparing simulations with solids concentration profiles
measured at 6 different distances, i.e. 5, 10, 15, 20, 25 and 30 m. Simulated profiles were
recorded every 15 minutes giving a range of profiles between which the measured solids
concentrations should be found for successful validation. Results are shown in Figure 6.64.
Without any additional calibration, an excellent agreement is found between simulations and
measurements. Only close to the bottom floor at distance of 20 m, the simulated
concentration largely deviates from the measured value. A possible cause may be a clump of
solids stuck to the bottom, not being removed by the scraper. Good agreement between
measured and simulated solids concentrations profiles is observed.
Figure 6.63 Input profiles for the 2-day pre-simulation to initialise the settling tank
model
Figure 6.64 Comparison between predicted and measured solids concentration
- 108 -
Table 6.11 Comparison of model predictions with field data

Operation
Effluent concentration (mg/l) and improvement
conditions
Ave. Baffle and
baffle
concentration No Modifications launder
Modifications
mg/l Modifications
Field Data
50 28 - -
June-August
Model
47 27 6 (+78%) 5 (+82%)
Predictions
The average values of effluent concentration of improvement are presented in Table 6.11.
The tank operation conditions in the data collection periods (June-August, 2006) and the CFD
model predictions are very close as shown in Table 6.11. The comparison of model
predictions with the subsequent field data indicates that among the major modelling results
the two crucial points related to the impact of proposed modifications on the tank
performance were precisely captured by the model simulations. They are:
• The optimized baffle and launder modifications can reduce the effluent concentration;
by more than 75%; and
• Baffle modifications dominate the positive effect on the tank performance in tank.
6.10.3 Validation of solids concentration profiles

As far as the CFD model validity is concerned, Figure 6.65 presents a comparison between
the experimentally measured and the simulated values of the floc size distribution in the
effluent of the existing tanks in Hrinova and Holic. Apparently, there is a good agreement
between measured and predicted values.
Figure 6.65 Simulated and experimental particle size distribution in the effluent of the
existing tank
6.11 Conclusions
This chapter deals with the development a specialized strategy for the simulation of the
treatment of potable water in sedimentation tanks in Slovakia. The strategy is based on the
CFD code Fluent and exploits several specific aspects of the potable water application (low
- 109 -
solids mass and volume fraction) to derive a computational tool computationally much more
efficient (due to the independent handling of flow field and different particle classes) than the
corresponding tools employed to simulate primary and secondary wastewater settling tanks.
The present code is modified based on data from a real sedimentation tanks. Then it is
applied to predict the tank performance in the existing settling tanks and the STs with
proposed modifications at the Hrinova and Holic Water Treatment Plant. The existing STs
suffered from the relatively poor hydraulic performance. The domain finally obtained was
meshed with quadrilateral elements. An improperly meshed domain is known to influence the
solution field and, therefore, different mesh sizes were investigated. For the flow field
considered, the selected size resulted in 147,814 cells and was a trade-off between numerical
accuracy and computation time.
The water temperature affects the settling velocities via the fluid viscosity. The Stokes’ Law
relationship was shown to adequately describe the effect of temperature on a selected floc.
This relationship can be applied to correct the settling velocities for difference in
temperatures in whichever of the one type of sedimentation settling processes, i.e.,
unflocculated discrete settling.
In practice, solids removal may largely depend on the removal mechanism. In this respect,
the studied settling tank operated with a scraper mechanism, and it therefore seemed essential
to properly model the scraper. However, solids mixing and/or conveyance by the scraper are
by definition 3D phenomena making a 2D modelling approach hard to realise. To account for
the scraper effect in the model, its force exerted on the solids was decomposed. Only the
radial component was considered due to the 2D modelling approach. The temporally and
spatially varying scraper’s force obtained was implemented in Fluent as a momentum source.
Although the analytically computed 2D scraper velocities corresponded well with 3D
measurements and modelling results found in literature, the scraper modelled in this way was
found to be detrimental for the tank’s operation. Due to the flow obstruction near the bottom
floor caused by the scraper’s velocity being lower than the (local) free flow, the scraper made
the blanket rise; this did not correspond to reality though.
Finally, the simulated solids concentration and velocity fields were confronted with
measurements in order to evaluate the model’s prediction power.
In general, CFD can be a powerful tool for troubleshooting problems, particularly those
associated with flow patterns in a sedimentation tank. The results of this work give an insight,
which can be used to investigate novel designs or different operating conditions, such as
temperature variation, for production-scale tanks. Of course, because of the complexity of the
processes taking place, CFD will not completely replace experimental testing and the partly
empirical nature of the design process.
Traditional techniques will continue to be used for routine design, but CFD is invaluable for
backing up this work and for investigating novel tank designs.
- 110 -
Chapter 7
Conclusions and Recommendations
The introduction of this work made clear that many factors influence the performance of
settling tanks. They may be categorised as physic-chemical and hydraulic influences. To
account for them in terms of process operation and design, mathematical models may be
utilised. In this respect, Computational Fluid Dynamics (CFD) enables the investigation of
internal processes, such as local velocities and solids concentrations, to identify process in
efficiencies and resolve them. Although these complex models demand for considerable
computational power, they may become an option for the study of process operation and
control as computer speed increases. Nowadays, they mostly find applications in the world of
settling tank design.
The main purpose of this investigation was to develop a CFD ST model capable of
simulating the major processes that control the performance of settling tanks, this goal was
achieved. The accomplished objectives of this research include: the development of a
compound settling model that includes the representation of the settling velocity for the
suspended solids usually encountered in this type of tank the inclusion of swirl effects, a
flocculation sub-model, and a temperature sub-model.
These types of sub-models have not been previously incorporated in CFD ST models. The
model was rigorously tested and validated. The validation process confirms the utilities and
accuracy of the model. An important benefit of this research is that it has contributed to a
better understanding of the processes in STs. The results presented in this research clarify
important points that have been debated by previous researchers.
This research may also open the discussion for future research and different ways for
improving the performance of existing and new STs. In summary, this research has led to
more complete understanding of the processes affecting the performance of settling tanks,
and provides a useful tool for the optimization of these corn stone units in water treatment.
The major conclusions, general and specifics, obtained from this research are:
7.1 Development of a CFD model
The development of settling tank CFD model was conducted by means of a case study. It
concerned a rectangular settling tank (Slovakia) WTP surface and ground water sources with
a scraper as solids removal mechanism.
The performance of settling tanks depends on several interrelated processes and factors that
include: hydrodynamics, settling, turbulence, flocculation, temperature changes, geometry,
loading, and the nature of the floc
A ST model has been developed to include the following factors: axisymmetric
hydrodynamics, I-Type and II-Type of settling, turbulence, flocculation with 13 classes of
particles, temperature changes, various external and internal geometries, unsteady solids and
hydraulic loading.
A field testing procedure is presented that addresses all of the settling regimes that are
encountered in a Settling Tank, i.e. non-settling particles, unflocculated primary particles,
partially flocculated particles, highly flocculated particles with discrete settling,. This
procedure is used for the calibration of the settling model.
The ST model reproduces the major features of the hydrodynamic processes and solids
distribution on STs. When the model is executed with the field derived settling characteristics
as recommended in this dissertation, it can accurately predict the effluent and recirculation
suspended solids concentration.
- 111 -
7.1.1 Modelling of the scraper

Scraper is important in the settling process and plays a big role in changing the flow field.
Dynamic mesh approach can simulate the scraper movement and its actual effect. Moving
wall velocity approach is an averaged approximation of the effect of the scraper. Averaged on
many runs, this approach may have the similar effect. Momentum source approach only
considers the force balance between scrape and fluid (sludge).
7.1.2 Particle size distributions
The ability of sedimentation the tanks to clarify water by letting suspended solids settle out as
flocculated particles depends on two aspects: (a) The water flow pattern through the tank,
which in turn is determined by the configuration of the tank and by operational parameters
(solids concentration, water flow rate and temperature). (b) The settling characteristics of the
particles as determined by their shape, size and interaction with the water through drag and
buoyancy forces. Water flow patterns dominated particle settling in determining the dynamics
and efficiency of rectangular sedimentation tanks experiments and simulations done in the
project. It was also observed that the water flow patterns were remarkably stable and robust
for any particular configuration. With regard to the water flow pattern through tanks it was
found that the most important aspect was the design of the inlets, especially their placement.
It can be seen, the model predicts highly distinct concentration for different classes of
particle; lower removal rate for the smallest and higher removal rate for the heaviest particles.
Therefore, the improvement in the overall efficiency of solids removal is only achieved by
improving the settling of particles with lower settling velocities (classes 1-5).
7.1.3 Model validation
The fairly good agreement between model predictions and field data indicates that the present
modeling has achieve a status that it can be used by design engineers to optimize the design
of new tanks and to diagnose the performance of existing tanks.
The work in this manuscript shows that CFD models can be used to optimize settling tank
design and improve performance. A calibrated CFD model can serve as a numerical
laboratory to test design concepts and considerations, and assess the benefits prior to full
design and implementation of any proposed changes.
7.1.4 Iron removal
This work deals with the development a specialized strategy for the simulation of the
treatment of potable water in sedimentation tanks. The strategy is based on the CFD and Fe
removal model. The present code is modified based on data from a real sedimentation tank.
The settling tank model has been used in this study to simulate the details of flow and solids
fields in the Hrinova water treatment plant. Using the CFD modeling results together with the
Fe removal model presented in this study, the following conclusions can be obtained as:
• The Irons removal model here combined with a CFD based process/ST model could
be a very powerful tool to optimize a ST design/operation.
• If an economical dosage of 4 mg/L were to be used, a ST producing an effluent SS
concentration of 2 mg/L would be expected to have an Irons removal of 80%
• Using the CFD modeling results together with the Irons removal model To reduce the
dosage from 8 to 4 mg/L,
• In the ST process using the optimized design and the existing ST design, the
coagulant dosage in daily operation could be reduced to 50% of the design value.
• For the ST process with both of optimized ST design and design existing tank, the
coagulant dosage in daily operation is reduced by more than 50%. The design SOR of
optimized ST is 3.5 m3/m2/h while keeping the effluent SS concentration below than 5
mg/L.
- 112 -
• The work in this manuscript shows that CFD models can be used to optimize settling
tank design and improve performance. A calibrated CFD model can serve as a
numerical laboratory to test design concepts and considerations, and assess the
benefits prior to full design and implementation of any proposed changes
7.1.5 Temperature
• Temperature data from this study indicated that ideal, uniform flow often does not
occur in sedimentation tanks when subject to winter cooling and summer heating. The
degree of non-uniformity is a complex function of inflow conditions (temperature,
suspended solids, flow rate, tank geometry, and inlet baffle design) and
meteorological conditions. Hence, design approaches for sedimentation tanks should
consider how inflow conditions and meteorological conditions (temperature) impact
tank hydraulics.
• The water temperature affects the settling velocities via the fluid viscosity. The
Stokes’ Law relationship was shown to adequately describe the effect of temperature
on a selected floc. This relationship can be applied to correct the settling velocities for
difference in temperatures in whichever of the I-types of sedimentation settling
processes, i.e., unflocculated discrete settling, flocculated discrete settling.
• The changes in temperature on STs play an important role in the performance of
settling tanks. A warmer inflow produces a transient strengthening of the density
current which results in a higher ESS; the creation of an unstable stratification density
gradient in the vertical direction magnifies this effect. A cooler inflow also results in a
strengthening of the density current, but in this case the stable vertical density
gradient seems to suppress the turbulence in the vertical direction and to damp the
diffusion of the suspended solids.
• The changes in temperature on settling tanks play an important role on the
performance of settling tanks.
7.2 Proposed modifications
The 2-D fully mass conservative ST model is applied to predict the tank performance in
the existing STs and the STs with proposed modifications at the Hrinova and Holic
Water Treatment Plants. The existing Hrinova and Holic STs suffered from the
relatively poor hydraulic performance which typically occurs in a large tank without
proper baffling. The unfavorable hydraulic regime includes strong turbulence, a high
influent potential energy and a strong density current due to excessive flow
entrainment. The baffle modifications can considerably reduce the strength of the
density flow and increase the solids detention time in the tank; the effluent quality can
be improved by more than 60% for any cases. Proper launder modifications can be
used to improve local flow pattern near the effluent weir and to re-distribute the
effluent flow along the tank longitudinal direction.
This work deals with the development a specialized strategy for the simulation of the
treatment of potable water in sedimentation tanks. The strategy is based on the CFD
code Fluent and exploits several specific aspects of the potable water application (low
solids mass and volume fraction) to derive a computational tool computationally much
more efficient (due to the independent handling of flow field and different particle
classes) than the corresponding tools employed to simulate primary and secondary
wastewater settling tanks. The present code is modified based on data from a real
sedimentation tank. Then it is used to study the effect of different baffles positions and
different baffle configurations on the performance of settling tanks. Using the CFD
modeling results together with the field work presented in this study, the following
conclusions can be obtained as:
- 113 -
The use of an inboard launder also produced a significant benefit in terms of reduced
ESS. The best result was obtained by a combination of inboard launder and a
perforated baffle.
The results show that an extended baffle forces the solids to move faster towards the
bottom of the tank and decreases the inlet recirculation zone, thus yielding significantly
enhanced sedimentation. Although the increase in the overall effectiveness by this baffle
may show only a small change, this actually reflects a reduction of the effluent solids of
estimated around 85%.
In general, CFD can be a powerful tool for troubleshooting problems, particularly those
associated with flow patterns in a sedimentation tank. The results of this work give an
insight, which can be used to investigate novel designs or different operating conditions,
such as temperature variation, for production-scale tanks. Of course, because of the
complexity of the processes taking place, CFD will not completely replace experimental
testing and the partly empirical nature of the design process.
7.3 Current design guidelines
Many design guidelines have been developed during the past decades. The guidelines are
often considered in isolation from the intricate balance of forces affecting sedimentation
tanks. Design practices need to be considered in the context of the sedimentation tank as a
whole. When the guidelines used by the conventional approach are compared to the CFD
approach (Table 6.9) the difference is astounding.
All the guidelines used in the conventional approach are based on ideal transport
assumptions. None of the guidelines can predict the removal efficiency of the tank. The CFD
approach does not have to rely on idealistic transport assumptions and can with reasonable
accuracy predict the removal efficiency.
Overall, the following conclusions can be taken in relation to this study:

1. CFD modeling was successfully used to evaluate the performance of settling tank.
2. Solid removal efficiency can be estimated by calculating solids concentration at effluent.
3. High solid removal efficiency was achieved for all cases tested.
4. Baffling inlet arrangement succeeded in controlling kinetic energy decay.
5. Improved energy dissipation due to an improved inlet configuration.
6. Reduced density currents due to an improved inlet configuration.
7. Improved sludge removal due to the inlet configuration and hopper shape.
8. Troubleshoot existing STs and related process operations
9. Evaluated ST design under the specified process conditions
10. Develop reliable retrofit alternatives with the best cost-effectiveness
In general the study demonstrated that CFD could be used in reviewing settling tank design
or performance and that the results give valuable insight into how the tanks are working. It
can be inferred that CFD could be use to evaluate settling tank designs where the tanks are
not functioning properly.
7.4 Suggestions for Further Research

Even though this research makes some important advances in the modeling of rectangular
settling tanks, the model makes use of simplifying assumptions that limit is applicability. The
following are suggestions for additional developments to address these limitations:
• The sedimentation tank model attempts to model two-phase behaviour. The modelling
of two or more interacting phases is extremely complex. The approach selected in this
case is limited in many respects. The interaction of the solid and liquid phases has
been simplified considerably.
- 114 -
• A number of factors such as the effect of the presence of the rake mechanism, particle
size agglomeration, and particle-particle interaction have been neglected or treated
approximately in the present study. With more refinement of the modeling to take
account of such factors, more accurate simulations can be obtained. Finally, the study
has brought out important features of the particle motion in settling tanks under
typical conditions; it would be interesting to verify these by conducting detailed
experimental studies.
• Such improvements in the model need to be validated against a well planned
experimental program. Velocity and concentration measurements both need to be
taken at as close a time together as possible, methods for doing this in the main body
of the tank are reasonably well established. However, measurements of concentration
and velocity in the hopper region and concentration measurements in the settled bed
also need to be made to give a comprehensive picture of the flow field in the tank.
Increasing the capabilities of the model means that the number of quantities measured
in experiment has to increase.
• The ST model neglects the effects of the Brownian motion flocculation in the general
flocculation process, partly because it leads to the aggregation of particles that fall
into the non-settleable portion. However, the effects of the Brownian motion
flocculation inside the ST needs to be further investigated.
• The effect of poorly flocculated sludge on the performance of ST should be further
evaluated. Even though it is suspected that a good representation of the discrete
settling fractions, and representative kinetic flocculation constants should yield an
accurate simulation of this type of incoming sludge, this assumption needs to be
validated.
• In Chapter 6, Equation 6.19 proposes a relationship for the correction of the settling
properties for difference in temperatures. Even though this relationship can be used to
conduct a sensitivity analysis in the performance of the model for different seasons,
e.g. summer and winter, there is no evidence that the settling properties can the model
for different seasons, e.g. summer and winter, there is no evidence that the settling
properties can be accurately extrapolated from one season to another. More research
is needed to define the effect of seasonal variations on the settling characteristics and
other sludge’s properties like the flocculation kinetic constants.
• This research concluded that, “In rectangular tanks with scrapers, the blades are not
highly effective in conveying the solids in the region near the hopper“ However, field
data should be gathered and more research should be conducted on the effect of
blades’ position in order to come up with a better design of the scraper mechanism.
- 115 -
References
1. Abdel-Gawad S. M. & McCorquodale J.A. (1985). Simulation of particle

concentration distribution in primary clarifiers. Canadian Journal of Civil
Engineering, 12(3), pp. 454-463.
2. Abdel-Gawad S. M. & McCorquodale J.A., (1984a). Strip integral method applied
to settling tanks, Journal of Hydraulic Engineering, ASCE, 110(1), 1-17.
3. Abdel-Gawad, S. M. & McCorquodale, J. A. (1984b). Hydrodynamics of Circular
primary clarifiers. Canadian J. Civil Engr. ,11, 299-307.
4. Adams, E.W. & Rodi, W. (1990). Modelling flow and mixing in sedimentation
tanks, Journal of Hydraulic Engineering, 116(7):895-913, Jul.
5. Anderson J.D., Grundmann R., Dick E. & Degrez G. (2003). Introduction to
computational fluid dynamics. Annual Lecture Series von Karman Institute, St.
Genesius-Rode, Belgium (ISSN 0377-8312)
6. Anderson N.E. (1945). Design of final settling tanks for activated sludge. Sewage
Works Journal, 17(1), pp. 50-65.
7. APHA (1992). Standard methods for the examination of water and wastewater.
nd
Eds. Clesceri L.S, Greenberg A.E. and Eaton A.D., 2 Ed. Washington DC, USA.
8. Armbruster M., Krebs P. & Rodi W. (2001). Numerical modelling of dynamic
sludge blanket behaviour in secondary clarifiers. Wat. Sci. Tech., 42(11), pp. 173-
180.
9. Baud O. & Hager W.H. (2000). Tornado vortices in settling tanks. J. Environ. Eng.,
126(2), pp. 189-191.
10. Bretsher, U. Krebs, P. & Hager, W.H. (1992). Improvement of flow in final settling
tanks, Journal of environmental engineering, 118(3), 307-321.
11. Brouckaert C.J. & Buckley C.A. (1999). The use of computational fluid dynamics
for improving the design and operation of water and wastewater treatment
plants. Wat. Sci. Tech., 40(4-5), pp. 81-89.
12. Camp, T.R. (1945). Sedimentation and the design of settling tanks. Transactions of
the American Society of Civil Engineers, Paper No. 2285, pp. 895-958.
13. Caretto, L.S., Gosman, A.D. Patankar, S.V. & Spalding, D.B. (1972). Two
calculation procedures for steady three-dimensional flows with recirculation,
Proceedings of the 3rd International conference on numerical methods in fluid
dynamics, Paris, p 60.
14. Casonato, M. & Gallerano, F. (1990). A finite difference self-adaptive mesh
solution of flow in a sedimentation tank, Int. J. for numerical methods in fluids, 10,
697-711.
15. Celik I., Rodi W. & Stamou A.I. (1985). Prediction of hydrodynamic
characteristics of rectangular settling tanks. Proc. Int. Symp. Refined Flow
Modelling and Turbulence Measurements, Iowa City, Iowa, USA.
16. Chu C.P., Lee D.J., & Tay J.H. (2005). Floc model and intrafloc flow, Chem.
Eng.Sci. 60 565–575.
17. Crosby R.M, (1984). Hydraulic characteristics of activated sludge-secondary
clarifiers, EPA 600/2-84-131, NTIS No. PB-84-229665, U.S. EPA, Municipal and
Environmental Research Laboratories, Cincinnatti, OH.
18. Dahl, C. Larsen, T. & Petersen, O. (1994). Numerical modelling and measurement
in a test secondary settling tank, Wat. Sci. Tech. 30(2), 219-228.
19. De Clercq B., & Vanrolleghem P.A., (2002). Computational fluid dynamics in
wastewater treatment Med. Fac. Landb. Toegep. Biol. Wet. 67. 15–19.
20. De Cock W., Blom P., Vaes G. & Berlamont J. (1999). The feasibility of
flocculation in a storage sedimentation basin. Wat. Sci. Tech., 39(2), pp. 75-83.
- 116 -
21. De Vantier B.A. & Larock B.E. (1986). Modelling a recirculating density-driven
turbulent flow. Int. J. Num. Meth. Fluids, 6, pp. 241-253.
22. De Vantier, B.A. & Larock, B.E. (1987). Modelling sediment-induced density
currents in sedimentation basins, Journal of hydraulic engineering, 113(1), 80-94.
23. Deininger A., Holthausen E. & Wilderer P.A. (1998). Velocity and solids
distribution in circular secondary clarfiers: full scale measurements and
numerical modelling. Wat. Res., 32(10), pp. 2951-2958.
24. Dick, R. I. (1982) Sedimentation Since Camp, J Society of Civil Engrs, 68, 199-235.
25. Drew, D. A., & Lahey, R. T. (1979). Application of general constitutive principles
to the derivation of multidimensional two-phase flow equations. Int. J. Multiphase
Flow, 5, 243–264.
26. Ekama G.A., Barnard J.L., Günthert F.W., Krebs P., McCorquodale J.A., Parker D.S.
& Wahlberg E.J. (1997). Secondary settling tanks: theory, modelling and
operation. IAWQ Scientific and Technical Report No. 6.
27. Forbes, R.J. & dijksterhuis, E.J. (1963). A history of Science and Technology 1 –
Ancient times to the seventeenth century, Pelican Books, London, 294 p.
28. Frey, PH. et al. (1993). Hydrodynamics fields and solid particle transport in a
settling tank, Journal of Hydraulic research, 31(6), 763-776.
29. Fuchs A. & Staudinger G. (1999). Characterising the clarification of the
supernatant of activated sludges. Wat. Res., 33(11), pp. 2527-2534.
30. Ghawi A. Hadi & Jozef Kris (2007a).” Design and Optimization of Sedimentation
Tank in Slovakia with CFD Modeling”. 10th International Symposiumon Water
Management and Hydraulic Engineering 2007 with special emphasis on the impact of
hydraulic engineering construction on the environment 4 – 9 September 2007. Šibenik ,
Croatia
31. Ghawi A. Hadi (2007c). Drinking Water Service Reservoirs Design
Improvement Using CFD JUNIORSTAV 2007 Conference for Doctoral Study.
JUNIORSTAV 2007" on 24th January 2007 Czech Republic. pp. 220.
32. Ghawi A. Hadi, & Jozef Kris (2007b). A Numerical Model of Flow in
Sedimentation Tanks in Slovakia Third International PhD Symposium in
Engineering 25-26 October 2007, HungaryUniversity of Pollack Mihály Faculty
of Engineering Pécs, Hungary
33. Ghawi A. Hadi, & Jozef Kris (2007d). Improved, Modeling, Simulation and
Operational Parameters of Settling Tank. 6th International Conference of PhD
Students University of Miskolc, Hungary 12- 18 August 2007. pp. 69-75.
34. Ghawi A. Hadi, & Jozef Kris (2007e). Improved, Modeling, Simulation and
Operational Parameters of AL Wathba Settling Tank. International doctoral
seminar in Smolenice castle may 13-16, 2007 .pp.147-15
35. Gibson, M.M & Rodi, W. (1989). Simulation of free surface effects on turbulence
with a Reynolds stress model, Journal of Hydraulic Research, 27(2), 233-245
36. Gmachowski L., (2005). Aggregate structure and hydrodynamics of aggregated
systems, Colloids Surf. A: Physicochem. Eng. Aspects 255 105–110.
37. Godo, A.M. (1990). Density currents in rectangular settling tanks. M.S. thesis,
University of Windsor, Windsor, Ontario.
38. Günthert F.W. (1984). Thickening zone and sludge removal in circular final
settling tanks. Wat. Sci. Tech., 16, pp. 303-316.
39. Hazen A. (1904). On sedimentation. Trans. ASCE, 53, pp. 45-71.
40. Hirsch C. (1997). Numerical computation of internal and external flows. Vol 1,
Fundamentals of numerical discretization. John Wiley & Sons, West Sussex, UK, 515
pp.
41. Huggins D.L., Piedrahita R.H., & Rumsey T., (2005). Use of computational fluid
dynamics (CFD) for aquaculture raceway design to increase settling
effectiveness, Aquacult. Eng. 33, 167–180.
- 117 -
42. Imam E., McCorquodale J.A. & Bewtra J.K. (1983). Numerical modeling of
sedimentation tanks. J. Hydr. Engrg., 109(12), pp. 1740-1754.
43. Imam, E.,H. (1981). Numerical Modeling of Rectangular Clarifiers. Thesis
presented to the University of Windsor, Windsor Ontario, in partial fulfillment of the
requirement of the degree of Doctor of Philosophy.
44. Jasak, H. (1996). Error analysis and estimation for the finite volume method with
application to fluid flows, D.Phil. Thesis, Department of Mechanical Engineering,
Imperial College, 396 p.
45. Jayanti S. & Narayanan S., (2004). Computational study of particle-eddy
interaction in sedimentation tanks, J. Environmental Eng., 130 (1), 37-49.
46. Kalbskopf K.H. (1970). European practices in sedimentation. In: Water quality
improvement by physical and chemical processes (Water Resources Symposium
3), Eds. Gloyna E.F. and Eckenfelder W.W., University of Texas Press, Austin,
Texas, USA.
47. Kawamura S., (2000). Integrated Design and Operation of Water Treatment
Facilities, John Wiley and sons, New York 2000.
48. Kim H.S., Shin M.S., Jang D.S., Jung S.H., & Jin J.H., (2005). Study of flow
characteristics in a secondary clarifier by numerical simulation and radioisotope
tracer technique, Appl. Radiat. Isotopes 63 519–526.
49. Kinnear D. J. (2002). Biological Solids Sedimentation: A Model Incorporating
Fundamental Settling Parameters. Ph.D. Thesis, University of Utah.
50. Kinnear D.J. & Deines K. (2001). Acoustic Doppler current profiler: clarifier
th
velocity measurement. Proc. 74 Annual WEF Conference and Exposition on Water
Quality and Wastewater Treatment, 13-17 October, Atlanta, USA.
51. Kinnear, D.J., (2004). Littleton Englewood Wastewater Treatment Facility,
Personal communication.
52. Konicek Z. & Burdych J. (1988). Effects of activated sludge processes on
secondary settling tank efficiencies. Wat. Sci. Tech., 20(4-5), pp. 153-163.
53. Kostoglou M., & Karabelas A.J., Comprehensive modeling of precipitation and
fouling in turbulent pipe flow, Indus. Eng. Chem. Res. 37 (1998) 1536–1550.
54. Krebs, P. (1991) The hydraulics of final settling tanks, Wat. Sci. Tech., 23 Kyoto,
1037-1046.
55. Krebs, P. Vischer, D. & Gujer, W. (1992) Improvement of secondary clarifiers
efficiency by porous walls, Wat. Sci. Tech., 26(5-6), 1147-1156
56. Krebs, P. Vischer, D. & Gujer, W. (1995) Inlet-structure design for final clarifiers,
Journal of environmental engineering, 121(8), 558-564.
57. Kris Jozef & Ghawi A. Hadi. (2007f) Potable Water Service Reservoirs Design
Improvement Using Computational Fluid Dynamics. 4th IWA Specialist
Conference.on Efficient Use and Management Of Urban Water Supply 21-23, May,
JejuIsland, Korea. Pp. 1225-1227.
58. Kriš, J., & Ghawi, A., H. (2006). Application of Computational Fluid Dynamics
Modelling to Potable Water Service Reservoirs. In: Proceedings of International
Conference on New Trends in Water Supply, Sewerage and Solid Waste Treatment,
Žiar, September, pp. 81-92, ISBN 80-227-2469-6.
59. Lakehal D., Krebs P., Krijgsman J. & Rodi W. (1999). Computing shear flow and
sludge blanket in secondary clarifiers. J. Hydr. Engrg., 125(3), pp. 253-262.
60. Larsen P., (1977). On the hydraulics of rectangular settling basins, Report No.
1001, Department of Water Research Engineering, Lund Institute of Technology,
Lund, Sweden.
61. Launder, M. & Spalding, D. (1974). The numerical computation of turbulent
flows, Computer methods in applied mechanics and engineering, (3), 269-289.
62. Lee D.J., Chen G.W., Liao Y.C., & Hsieh C.C. (1996). On the free-settling test for
estimating activated sludge floc density, Water Res. 30 541–550.
- 118 -
63. Leonard, B.P & Mokhtari S. (1990). Beyond first-order upwinding: the ultra-
sharp alternative for non-oscillatory steady-state simulation of convection, Int.
Journal for numerical methods in engineering, 30:729-766.
64. Li D. & Ganczarczyk J. (1991). Size distribution of activated sludge flocs. Res. J.
WPCF, 63(5), pp. 806-814.
65. Lomax H., Pulliam T.H. & Zingg D.W. (2001). Fundamentals of computational
fluid dynamics. Springer-Verlag, 251 pp.
66. Lyn D.A., Stamou A., & Rodi W. (1992). Density currents and shear induced
flocculation in sedimentation tanks, J. Hydr. Eng. 118 849–867.
67. Marais, P.M. Ekama, G.A & DE Haas D.W. (2000) Comparison of the 1D idealised
flux theory and a 2D hydrodynamic model with full scale secondary settling tank
performance data, Proceedings of WISA 2000 conference, Sun City, South Africa
68. Matko T, Fawcett N, Sharp A & Stephenson T (1996). A numerical model of flow in
circular sedimentation tanks. Trans IChemE., 74(B), 197-204.
69. McCorquodale J.A., & Zhou S. (1993). Effects of hydraulic and solids loading on
clarifier performance, J. Hydr. Res. 31 461–477.
70. McCorquodale J.A., Yuen E.M., Vitasovic Z. & Samstag R. (1991). Numerical
simulation of unsteady conditions in clarifiers. Water Poll. Res. J. Canada, 26(2),
pp. 201-222.
71. McCorquodale, J.A. (1977). Temperature profiles in primary settling tanks. IRI
Report University of Windsor.
72. McCorquodale, J.A. (1987). Density currents in clarifiers. Proceedings of Hydraulic
Division, ASCE Conference,. Williamsburg, VA, pp. 56-61.
73. Narayanan B., Buhr H.O., Leveque E.G., Lee M.C. & Dielmann D. (2000). Could
your secondary clarifier sludge removal mechanism be a process bottleneck?
th
Proc. 73 Annual WEF Conference and Exposition on Water Quality and Wastewater
Treatment, 14-18 October, Anaheim, California, USA.
74. Parker D.S., Kinnear D.J. and Wahlberg E.J. (2001). Review of folklore in design
and operation of secondary clarifiers. J. Environ. Engrg., 127(6), pp. 476-484.
75. Patankar, S.V. & Spalding, D.B. (1972). A calculation procedure for heat, mass
and momentum transfer in three - dimensional parabolic flow. International
Journal of Heat and Mass Transfer.
76. Patankar, S.V. (1980). Numerical Heat Transfer and Fluid Flow, McGraw-Hill,
London.
77. Rhie, C.M. & Chow, W.L. (1983). Numerical study of the turbulent flow past an
airfoil with trailing edge separation, Journal American Institute of Aeronautics and
Astronautics, 1525-1532.
78. Rodi, W (1980). Turbulence Models and Their Application in Hydraulics, A State
of the Art Review. International Association for Hydraulic Research, Delft,
Netherlands
79. Rodi, W. (1993). Turbulence Models and their application in hydraulics – a state-
of-the-art review 3rd ed, IAHR Monograph Series, A.A. Balkema, Rotterdam.
80. Schamber D R, & Larock B E, (1981). Numerical analysis of flow in sedimentation
basins [J]. Journal of the Hydraulics Division, ASCE, 107(5): 575-591
81. Schulz, C. R. & D. A. Okun (1984). Surface Water Treatment for Communities in
Developing Countries, John Wiley & Sons.
82. Stamou A.I., Adams E.W. & Rodi W. (1989). Numerical modeling of flow and
settling in primary rectangular clarifiers. J. Hydr. Res., 27(5), pp. 665-682.
83. Stamou A.I., Latsa M. & Assimacopoulos D. (2000). Design of two-storey final
settling tanks using mathematical models. J. Hydroinformatics, 2(4), pp. 235-245.
84. Stamou, A.I & Rodi, W. (1984). Review of Experimental Studies on Sedimentation
Tanks. Sonderforschungsbereich 210 / E / 2. Universitat Karlsruhe. August
85. Stamou, A.I. (1991). On the prediction of flow and mixing in settling tanks using
a curvature modified k-ε model, Applied Mathematical Modelling, 15:351-358, Jul.
- 119 -
86. Stamou, A.I. McCorquodale & Zhou (1992). Improving the numerical modelling
of river water quality by using high order difference schemes, Wat. Res., 26(12),
1563-1570.
87. Surucu, G., & Cetin F.D. (1990). Effects of temperature, pH, and D.O.
concentration on settleability of activated sludge. Environmental Technology, 11,
pp. 205-212.
88. Szalai, L. Krebs, P. & Rodi, W. (1994). Simulation of circular clarifiers with and
without swirl, Journal of hydraulic engineering, 120(1), 4-21.
89. Takács I., Patry G.G. & Nolasco D. (1991). A dynamic model of the clarification-
thickening process. Wat. Res., 25(10), pp. 1263-1271.
90. Tchobanoglous G. and Burton F.L. (1991). Wastewater Engineering: Treatment,
rd
Disposal and Reuse. 3 Ed., Metcalf & Eddy Inc.
91. Van der Walt J.J. (1998). Is a sedimentation tank really that simple? in:
Proceedings of the WISA Biennale Conference and Exhibition, Cape Town, South
Africa.
92. Van Duuren, E D. (1997) Water purification works design, WRC report no TT
92/97, WRC, Pretoria, 422 p.
93. Van Marle C. & Kranenburg C. (1994). Effects of gravity currents in circular
secondary clarifiers. J. Environ. Eng., 120(4), pp. 943-960.
94. Van Rijn L.C. & Schaafsma A.S. (1986). Evaluation of measuring instruments for
suspended sediment. Proc. Int. Conf. Measuring Techniques of Hydraulics
Phenomena in Offshore, Coastal & Inland Waters, 9-11 April, London, UK.
95. Vanni M, (2000). Creeping flow over spherical permeable aggregates, Chem. Eng.
Sci. 55 685–698.
96. Versteeg, H. K., and W. Malalasekera, (1995) An Introduction to Computational
Fluid Dynamics: The Finite Volume Method, Longman Scientific & Technical,
Essex, UK.
97. Wells S.A. & LaLiberte D.M. (1998). Winter temperature gradients in circular
clarifiers. Wat. Environ. Res., 70(7), pp. 1274-1279.
98. Wilcox D.C. (1998). Turbulence Modeling for CFD, DCW Industries, Inc., La
Canada, California.
99. Wilkinson, D, B. Waldie, M.I. Mohamad Nor, H.Y. & Lee, (2000) Baffle plate
configurations to enhance separation in horizontal primary separators, Chem.
Eng. J. 77 221–226.
100. Wren D.G., Barkdoll B.D., Kuhnle R.A. & Derrow R.W. (2000). Field techniques for
suspended-sediment measurement. J. Hydr. Engrg., 126(2), pp. 97-103.
101. Zhou S. & McCorquodale J.A. (1992a). Mathematical modelling of a circular
clarifier. Can. J. Civ. Eng., 19, pp. 365-374.
102. Zhou S., & McCorquodale J.A. (1992). Influence of skirt radius on performance of
circular clarifiers with density stratification, Int. J. Numer. Meth. Fluids 14. 919.
103. Zhou S., & McCorquodale JA (1992b). Modelling of rectangular settling tanks.
Journal of Hydraulic Engineering, ASCE, 118(10), 1391-1405.
104. Zhou S., McCorquodale J.A. & Zhong J. (1993). Semi-implicit skew upwind
method for problems in environmental hydraulics. Int. J. Num. Meth. Fluids, 17, pp.
803-823.
105. Zhou S., McCorquodale J.A., & Godo A.M. (1994). Short circuiting and density
interface in primary clarifiers, J. Hydr. Eng. 120 1060– 1080.
106. Zhou, S., McCorquodale, J.A., & Vitasovic, Z. (1992). Influences of density on
circular clarifiers with baffles. Journal of Environmental Engineering, 118 (6), pp.
829-847.
- 120 -
Publication
1. Ghawi A. Hadi & Jozef Kris (2007a).” Design and Optimization of Sedimentation
Tank in Slovakia with CFD Modeling”. 10th International Symposium on Water
Management and Hydraulic Engineering 2007 with special emphasis on the impact
of hydraulic engineering construction on the environment 4 – 9 September 2007.
Šibenik , Croatia
2. Ghawi A. Hadi & Jozef Kris. (2008) “Improving Performance of Horizontal
Sedimentation Tanks” IWA World Water Congress and Exhibition 7–12 September
2008 Austria Center, Vienna. Accepted.
3. Ghawi A. Hadi & Jozef Kris. (2008). “Design and Optimization of Settling Tanks
Performances in Slovakia”. XX-th National, VIII-Th International Scientific and
Technical Conference Water Supply and Water Quality Water 2008. Poland 15-18
June 2008 Accepted.
4. Ghawi A. Hadi & Jozef Kris. (2008). “Study the Effect of Temperature on Settling
Velocity”. XX-th National, VIII-Th International Scientific and Technical
Conference Water Supply and Water Quality Water 2008. Poland 15-18 June 2008.
Accepted.
5. Ghawi A. Hadi (2007b). “Drinking Water Service Reservoirs Design
Improvement Using CFD” JUNIORSTAV 2007 Conference for Doctoral Study.
JUNIORSTAV 2007" on 24th January 2007 Czech Republic. pp. 220.
6. Ghawi A. Hadi, & Jozef Kris (2007c) “A Numerical Model of Flow in
Sedimentation Tanks in Slovakia” Third International PhD Symposium in
Engineering 25-26 October 2007, Hungary University of Pollack Mihály
Faculty of Engineering Pécs, Hungary
7. Ghawi A. Hadi, & Jozef Kris (2007d). “Improved, Modeling, Simulation and
Operational Parameters of Settling Tank”. 6th International Conference of PhD
Students University of Miskolc, Hungary 12- 18 August 2007. pp. 69-75.
8. Ghawi A. Hadi, & Jozef Kris (2007e). “Improved, Modeling, Simulation and
Operational Parameters of AL Wathba Settling Tank”. International doctoral
seminar in Smolenice castle may 13-16, 2007 .pp.147-15.
9. Ghawi, A. Hadi (2000). “Water Treatment Plants Design” B.Sc Thesis, University
of Al-Mustansiryia, Baghdad, Iraq.
10. Ghawi, A. Hadi (2002). “Modeling of Particulate Matter Dispersion in Street
Canyons in Baghdad City”. M.Sc. Thesis Al-Mustansiryia, Baghdad, Iraq.
11. Jozef Kris & Ghawi A. Hadi. (2008) “Study the Effect of Temperature on
Sedimentation Tanks Performance” IWA World Water Congress and Exhibition
7–12 September 2008 Austria Center, Vienna. Accepted.
12. Kriš Jozef & Ghawi A. Hadi (2007a). “CFD Investigation of Particle
Deposition and Resuspension in a Drinking Water Distribution System” Slovak
journal of civil engineering. Accepted
13. Kris Jozef & Ghawi A. Hadi. (2007b) “Potable Water Service Reservoirs Design
Improvement Using Computational Fluid Dynamics”. 4th IWA Specialist
Conference.on Efficient Use and Management Of Urban Water Supply 21-23, May,
JejuIsland, Korea. Pp. 1225-1227.
14. Kriš, J., & Ghawi, A., H. (2006). ”Application of Computational Fluid Dynamics
Modelling to Potable Water Service Reservoirs”. In: Proceedings of International
Conference on New Trends in Water Supply, Sewerage and Solid Waste Treatment,
Žiar, September, pp. 81-92, ISBN 80-227-2469-6.
- 121 -
View publication stats

The Thesis

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

The Thesis

Caricato da

Copyright:

Formati disponibili

See

A Numerical Model of Flow and Settling in

Thesis · February 2008

Ali hadi Ghawi

The user has requested enhancement of the downloaded file.

A Numerical Model of Flow and Settling in Sedimentation

Ing. Ali Hadi Ghawi

Bratislava, February 2008

SLOVAK UNIVERSITY OF TECHNOLOGY IN BRATISLAVA

A Numerical Model of Flow and Settling in Sedimentation

Doctoral: Ing. Ali Hadi Ghawi

Bratislava, February 2008

3.3 Fluid Dynamics ……………………………………………………………………………….. 26

5.8 Temperature …………………………………………………………………………………... 59

Figure Title Page

Figure 6.11 Diurnal Flow Variations in Settling Tank Effluent………………………… 71

Figure 6.39 Calculation results with Fr =0.003 in Hrinova WTP………………………. 88

Table Title Page

List of Symbols and Units

Abbreviation Description Unit

Abbreviation Description Unit

1.1 Background and Problem Definition

1.2 Numerical Investigation

Computational Fluid Dynamics (CFD) is a well established technique for numerical

Figure 1.1 Overview of the Settling Tank Project

1.3 Scope and Objectives

1.4 Dissertation Organization

Figure 2.2: lamella settling tank

2.7 Ideal sedimentation tank with a horizontal flow pattern

Figure 2.5 Types of Sedimentation

2.7.1 Classification of settling behavior

Figure 2.6 Definition diagram for particle terminal settling velocity

2.7.1.1 Models Based on Discrete Particles Settling

103 < Re < 105, CD ≈ 0.4

understanding sedimentation behavior. Furthermore, in practical sedimentation tanks, the

Figure 2.7 Sedimentation Basin Design (Ideal Basin)

According to this relationship, the slowest-settling particles, which could be expected to be

The ideal sedimentation design assumptions

2.8 Factors influencing the capacity and performance of settling tanks

It is evident from Table 3.1 that:

Where U – The average horizontal velocity in the tank (L/t)

Where R – The wetted perimeter of the tank (L)

3.2.5 Hazen number (Ha)

The Hazen number is limited for the following reasons:

3.2.7 Balance of forces

Figure 3.1 Flow Processes in a Rectangular ST (Larsen, 1997)

d) Energy equations (conservation of energy).

where ρ - fluid density

∂( ρv) ∂( ρuv) ∂( ρvv) ∂P ⎛⎜ ∂ ∂v ∂ ∂v ⎞⎟

Figure 4.1 A finite number of small control volumes (cells) by a grid

4.1 Discretization of the domain: grid generation

4.2 Discretization of the equations

4.2.1 Discretization schemes

Update properties, e.g. p

Solve numerical equations (u. v velocity)

Solve pressure-correction (continuity) equation

Solve turbulence and scalar equations

Update solution values with converged values at current time

Requested time steps completed?

Take a time step No Yes STOP

Figure 4.3 Solution Procedure

CONCEPTUAL MODEL MODEL

Figure 5.1 Process of developing CFD code.

Figure 5.2 Laboratory settling column in Hrinova WTP