Вы находитесь на странице: 1из 109

T H E U N I V E R S I T Y O F T U L S A

THE GRADUATE SCHOOL


A MECHANISTIC MODEL FOR LIQUID
HYDROCYCLONES (LHC)
by
Juan Carlos Caldentey
A Thesis Submitted in Partial Fulfillment of
the Requirements for the Degree of Master of Science
in the Discipline of Petroleum Engineering
The Graduate School
The University of Tulsa
2000
iii
ABSTRACT
Caldentey, Juan Carlos (Master of Science in Petroleum Engineering)
A Mechanistic Model for Liquid Hydrocyclones (LHC)
(98 pp. Chapter V)
Directed by Professor Ovadia Shoham and Professor Ram S. Mohan
(180 words)
Hydrocyclones provide economical and effective means for liquid-liquid
separation in the petroleum as well as other industries. This study is focused on the
deoiling of produced water utilizing a liquid hydrocyclone, LHC.
A simple mechanistic model is developed for the LHC. The model is capable of
predicting the hydrodynamic flow field of the continuous phase within the LHC. The
separation efficiency is determined based on droplet trajectories, and the inlet-underflow
pressure drop is predicted using an energy balance analysis.
The predictions of the proposed model are compared with elaborate published
experimental data sets. Good agreement is obtained between the model predictions and
the experimental data with respect to both separation efficiency and pressure drop. The
underflow separation efficiency is predicted with an average relative absolute error of
4%, while the pressure drop is predicted with an average relative absolute error of 11%.
iv
A user friendly computer code is developed in Excel-Visual Basic platform based
on the proposed model. The code provides easy access to the input data and very fast
output, and can be used for design of LHCs by the industry.
v
ACKNOWLEDGEMENTS
I would like to thank my co-advisors Dr. Ovadia Shoham and Dr. Ram Mohan for
their continued support, guidance and for the freedom they gave me to work
independently, which allowed me to explore several alternatives during my research. I
wish to acknowledge Dr. Charles Petty of Michigan State University for his valuable
assistance in the recompilation of literature.
I want to express gratitude to my colleagues within the TUSTP group, from whom
I learned invaluable knowledge. I am especially grateful to Luis Gomez and Carlos
Oropeza for the many helpful suggestions and assistance. Also, Ferhat Erdal, Shoubo
Wang and Carlos Gomez with whom I held many helpful discussions, and to Judy Teal
whose collaboration made this project a reality.
The research was made possible by the financial support of the TUSTP member
companies.
It is also important to acknowledge the Petroleum Engineering Staff of The
University of Tulsa, outstanding full time professors who share their time and experience
with the alumni and make this Department one of the top in the nation.
Finally, I would like to thank my family and friends who are the source of my
inspiration and motivation. This work is dedicated to my beloved wife, Ana, who not
only encouraged me to pursue my Masters studies but also helped me complete this
thesis.
vi
TABLE OF CONTENTS
TITLE PAGE i
APPROVAL PAGE ii
ABSTRACT iii
ACKNOWLEDGEMENTS v
TABLE OF CONTENTS vi
LIST OF FIGURES viii
LIST OF TABLES xi
CHAPTER I
INTRODUCTION 1
1.1 Motivation and Objective 1
1.2 LHC Hydrodynamic Flow Behavior 2
1.3 LHC Geometry 4
1.4 Thesis Structure 6
CHAPTER II
LITERATURE REVIEW 8
2.1 Solid Hydrocyclones 9
2.2 Liquid Hydrocyclones, LHC 13
2.2.1 LHC Modeling 14
2.2.2 Field Applications 15
2.2.3 Experimental Studies 16
2.2.4 Velocity Field measurements 19
CHAPTER III
LHC MECHANISTIC MODEL 23
3.1 Overview 23
3.2 Swirl Intensity 26
3.3 Velocity Field 29
vii
3.3.1 Tangential Velocity 29
3.3.2 Axial Velocity 31
3.3.3 Radial Velocity 33
3.4 Droplet Trajectories 34
3.5 Separation Efficiency 37
3.6 Pressure Drop 40
3.7 LHC Mechanistic Model Code 42
CHAPTER IV
RESULTS AND DISCUSSION 44
4.1 Swirl Intensity Prediction 44
4.1.1 Experimental Data Sets 44
4.1.2 Results 47
4.1.3 Discussion 50
4.2 Velocity Field Prediction 50
4.2.1 Experimental Data Sets 50
4.2.2 Tangential and Axial Velocity Results 51
4.2.3 Discussion 51
4.3 Droplet Trajectory Prediction 67
4.4 Separation Efficiency Prediction 69
4.4.1 Experimental Data Sets 69
4.4.2 Migration Probability and Underflow Purity Results 70
4.4.3 Discussion 77
4.5 Pressure Drop Prediction 78
4.5.1 Experimental Data Sets 78
4.5.2 Results 79
4.5.3 Discussion 81
CHAPTER V
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 82
5.1 Summary and Conclusions 82
5.2 Recommendations 85
NOMENCLATURE 88
REFERENCES 91
viii
LIST OF FIGURES
Figure 1.1 LHC Hydrodynamic Flow Behavior ............................................................... 3
Figure 1.2 Colman and Thews Hydrocyclone Design..................................................... 5
Figure 1.3 LHC Inlet Design ........................................................................................... 6
Figure 2.1 Tangential Velocity Diagram........................................................................ 11
Figure 2.2 Axial Velocity Profile From Colman (1984)................................................. 21
Figure 3.1 LHC Mechanistic Model Structure ............................................................... 24
Figure 3.2 LHC Characteristic Diameter ....................................................................... 28
Figure 3.3 Rankine Vortex ............................................................................................ 30
Figure 3.4 Axial Velocity Diagram................................................................................ 32
Figure 3.5 Droplet Velocities ........................................................................................ 34
Figure 3.6 Forces Acting on a Droplet........................................................................... 35
Figure 3.7 Droplet Trajectory and Migration Probability ............................................... 38
Figure 3.8 Migration Probability Curve......................................................................... 39
Figure 3.9 LHC Mechanistic Model Code .......................................................................43
Figure 4.1 Colmans Designs (1981) ............................................................................. 45
Figure 4.2 Swirl Intensity Prediction - Case 1................................................................ 48
Figure 4.3 Swirl Intensity Prediction - Case 2................................................................ 48
Figure 4.4 Swirl Intensity Prediction - Case 3................................................................ 49
Figure 4.5 Swirl Intensity Prediction - Cases 4 and 5..................................................... 49
ix
Figure 4.6 Tangential Velocity Prediction - Case 1........................................................ 53
Figure 4.7 Tangential Velocity Prediction - Case 2........................................................ 56
Figure 4.8 Tangential Velocity Prediction - Case 3........................................................ 56
Figure 4.9 Tangential Velocity Prediction - Case 4........................................................ 57
Figure 4.10 Tangential Velocity Prediction - Case 5...................................................... 58
Figure 4.11 Axial Velocity Prediction - Case 1.............................................................. 59
Figure 4.12 Axial Velocity Prediction - Case 2.............................................................. 62
Figure 4.13 Axial Velocity Prediction - Case 3.............................................................. 62
Figure 4.14 Axial Velocity Prediction - Case 4.............................................................. 63
Figure 4.15 Axial Velocity Prediction - Case 5.............................................................. 64
Figure 4.16 Axial Velocity Prediction - Case 6.............................................................. 65
Figure 4.17 Predicted Droplets Trajectories Case 7..................................................... 67
Figure 4.18 Trajectories of a 15 Microns Droplet Case 7........................................... 68
Figure 4.19 Migration Probability Curve - Case 7 ......................................................... 71
Figure 4.20 Underflow Purity,
u
- Case 7..................................................................... 71
Figure 4.21 Migration Probability Curve - Case 8 ......................................................... 72
Figure 4.22 Underflow Purity,
u
- Case 8..................................................................... 72
Figure 4.23 Droplet Size Distributions for Kuwait Oil (Colman et al., 1980) ................. 73
Figure 4.24 Droplet Size Distribution for Forties Oil (Colman et al., 1980) ................... 74
Figure 4.25 Migration Probability Curve Cases 16, 18 and 20 .................................... 75
Figure 4.26 Migration Probability Curve Case 23....................................................... 76
Figure 4.27 Migration Probability Curve Case 24....................................................... 76
Figure 4.28 Comparison of Model Underflow Purity and Experimental Data Set .......... 77
x
Figure 4.29 Pressure Drop Prediction Case 25............................................................ 79
Figure 4.30 Pressure Drop Prediction Case 26............................................................ 80
Figure 4.31 Comparison Between Pressure Drop Model and All Experimental Data ..... 81
Figure 5.1 Hypothetical Swirl Intensity Decay .............................................................. 87
xi
LIST OF TABLES
Table 3-1 Drag Coefficient Constants............................................................................ 36
Table 4-1 Geometrical Parameters of Colmans Designs (1981).................................... 46
Table 4-2 Operational Conditions of Colmans Designs (1981) .................................... 46
Table 4-3 Geometrical Parameters of Hargreaves (1990)............................................... 47
Table 4-4 Geometrical Parameters, Wolbert et al. (1995) .............................................. 70
Table 4-5 Underflow Purity Results Cases 7 to 24......................................................... 75
Table 4-6 Geometrical Parameters, Young et al. (1990) ................................................ 78
Table 4-7 Pressure Drop Cases 27 to 35 ..................................................................... 80
1
CHAPTER I
INTRODUCTION
1.1 Motivation and Objective
The petroleum industry has traditionally relied on conventional gravity based
vessels to separate multiphase flow. They are bulky, heavy, expensive and have large
residence time. The growth of the offshore oil industry, where platform costs to
accommodate these separation facilities are critical, has provided the incentive for the
development of compact separation technology. Hydrocyclones have emerged as an
economical and effective alternative for produced water deoiling and other applications.
The hydrocyclone is inexpensive, simple in design with no moving parts, easy to install
and operate, and has low maintenance cost.
In the past, hydrocyclones have been used to separate solid/liquid, gas/liquid and
liquid/liquid mixtures. For the liquid/liquid case, both dewatering and deoiling have been
used in the oil industry. This study focuses only on the latter case, using the liquid
hydrocyclones (LHC) to remove dispersed oil from a water continuous stream.
In general, oil is produced with significant amount of water and gas. Typically, a
set of conventional gravity based vessels are used to separate most of the multiphase
mixture. The small amount of oil remaining in the water stream, after the primary
separation, has to be reduced to a legally allowable minimum level for offshore disposal.
Hydrocyclones have been used successfully to achieve this environmental regulation.
2
There is a large quantity of literature available on the LHC, including
experimental data sets and computational fluid dynamic simulations. However, no simple
and overall mechanistic model has been developed to date for the LHC. The objective of
the current work is to develop a mechanistic model for the LHC to predict the separation
efficiency and the flow capacity (pressure drop flow rate relationship).
The developed model will allow the performance prediction for a given geometry
and operating conditions, that can be utilized for LHC design. Also, it will permit the
design of alternative geometries under similar conditions for optimization purposes.
1.2 LHC Hydrodynamic Flow Behavior
The hydrocyclone, as shown in Figure 1.1, utilizes the centrifugal force to
separate the dispersed phase from the continuous fluid. The swirling motion is produced
by the tangential injection of pressurized fluid into the cyclone body. The flow pattern
consists of a spiral within another spiral moving in the same circular direction (Seyda and
Petty, 1991). There is a forced vortex in the region close to the LHC axis and a free-like
vortex in the outer region. The outer vortex moves downward to the underflow outlet
while the inner vortex flows in reverse direction to the overflow outlet. Moreover, there
are some recirculation zones associated with the high swirl intensity at the inlet. These
zones, with a long residence time and very low axial velocity, have been found to be
diminished as the flow enters the low angle tapered section (see Figure 1.1).
An explanation of the characteristic reverse flow in the LHC is well described by
Hargreaves (1990). With high swirl at the inlet, the pressure is high near the wall region
and very low toward the center. As a result of the pressure gradient profile across the
3
diameter, decreasing with downstream position, the pressure at the downstream end of
the core is greater than at the upstream, causing flow reversal.
As the fluid moves to the underflow outlet, the narrowing diameter increases the
fluid angular velocity and the centrifugal force. It is due to this force and the difference in
density between the oil and the water, that the oil moves to the center where it is caught
by the reverse flow and separated flowing into the overflow outlet. Instead, if the
dispersed phase is the heaviest, like solid particles, it will migrate to the wall and exit
through the underflow.
Figure 1.1 LHC Hydrodynamic Flow Behavior
The amount of fluid going through the different outlets differs with heavy and
light dispersion. That means that for these two different separation cases, two different
4
geometries are necessary (Seyda and Petty, 1991). In the deoiling case, usually between 1
to 10 percent of the feed flow rate goes to the overflow.
Another phenomenon that may occur in a hydrocyclone is the formation of a gas
core. As Thew (1986) explained, dissolved gas may come out of solution because of the
pressure drop, migrating very fast to the axis, and eventually emerging through the
overflow outlet. A significant amount of gas can be tolerated but excessive amounts will
disturb the vortex. An experimental study on this topic is found in Smyth and Thew
(1996).
1.3 LHC Geometry
The deoiling LHC consists of a set of cylindrical and conical sections. Colman
and Thews (1988) design has four sections, as shown in Figure 1.1. The inlet chamber
and the reducing section are designed to achieve the higher tangential acceleration of the
fluid, reducing the pressure drop and the shear stress to an acceptable level. The latter has
to be minimized to avoid droplet breakup leading to reduction in separation efficiency.
The tapered section is where most of the separation is achieved. The low angle of this
segment keeps the swirl intensity with high residence time. An integrated part of the
design is a long tail pipe cylindrical section in which the smallest droplets migrate to the
reversed core at the axis. This configuration gives a very stable small diameter reversed
flow core, utilizing a very small overflow port.
5
Figure 1.1 Colman and Thews Hydrocyclone Design
Young et al. (1990) achieved similar results to Colman-Thews LHC, in terms of
separation efficiency, with a different hydrocyclone configuration. Three sections were
used instead of four. The reducing section was eliminated and the angle of the tapered
section was changed from 1.5 to 6. Later, Young et al. (1993) developed a new LHC
design, which resulted in an improvement in the separation performance. The principal
modification of the enhanced design was a small change in the tail pipe section. A minute
angle conical section was used rather than the cylindrical pipe.
Another important parameter in the LHC geometry is the inlet design (Figure 1.2).
Rectangular and circular, single and twin inlets have been most frequently used by
different researchers. The main goal is to inject the fluid with higher tangential velocity
avoiding the rupture of the droplets. The twin inlets have been thought to maintain better
symmetry and for this reason maintain a more stable reverse core (Colman et al., 1984;
6
Thew et al., 1984). Good results have also been achieved with the involute single inlet
design.
Figure 1.2 LHC Inlet Design
The last element of the LHC is the overflow outlet. This is a very small diameter
orifice that plays a major role in the split ratio, defined as the relationship between the
overflow rate and the inlet flow rate. Most of the commercial LHC permit changing the
diameter of this orifice depending on the range of operating conditions.
1.4 Thesis Structure
Current chapter is a brief preface to the study. It begins with a statement of the
incentive to develop this project from the oil industry point of view. It is followed by the
objective and the scope of the thesis. Then, the principles of operation of a hydrocyclone
is discussed, focusing in the hydrodynamic behavior. The last section contains the typical
geometry of a deoiling hydrocyclone including two different patent designs.
7
The second chapter is a review of some works pertinent to solid and liquid
hydrocyclones. A general review of solid hydrocyclones is presented in the first section
while a more detailed review is done for LHC. The LHC section is divided into
theoretical, experimental and applications studies. The last topic of the LHC literature is
related to the velocity field measurements and CFD simulations.
Chapter III consists of description of the developed LHC mechanistic model. This
chapter is divided into sections presented in the same order as the calculations that the
model follows. The first topic is the swirl intensity, an important parameter for defining
the velocity field which is the next subject. The velocity field allows the calculation of
the droplet trajectories which define the separation efficiency. These two are discussed in
the following sections. The LHC pressure drop - flow rate relationship is covered next
and the last topic of the chapter is related to the developed LHC mechanistic model code.
The accuracy of the mechanistic model is evaluated in Chapter IV through
comparison with available data from other researchers. The results include the swirl
intensity, the velocity profile, the separation efficiency and the pressure drop as well. The
conclusions and suggestions for further work are covered in Chapter V.
8
CHAPTER II
LITERATURE REVIEW
For many years hydrocyclones have been used in different industries such as pulp
and paper production, food processing, chemical industries, power generation,
metalworking, and oil and mining industries. Both solid-liquid and liquid-liquid
separation are possible with this technology. Most of the available literature on
hydrocyclones is related to solid-liquid separation. Since the 1980s, liquid-liquid
separation has become popular due to the relevant application area in the oil industry.
This work is focused on liquid-liquid separation, specifically for a lighter
dispersed phase. However, a brief review of solid hydrocyclones is imperative for
understanding the principle of operation of this device and the evolution of the different
models. It is important to stress up-front what are the main differences between these two
types of separation processes.
The density difference is much smaller for liquid-liquid mixtures, making
the separation more difficult and creating the necessity of operating with
higher centrifugal forces.
The solid particles can be considered rigid unlike the liquid droplets which
deform with the interaction of external forces. If high shear stress is
present, this may cause droplet break up, reducing the probability of the
smaller droplets to be separated. In the opposite case, if two droplets get
close enough, a coalescence effect can occur, whereby the larger droplets
9
can be separated more easily.
As mentioned in the previous chapter, the amount of fluid going through
the different outlets differs with either heavy or light dispersion. For solid-
liquid separation, more than 90% of the fluid exits from the top of the
hydrocyclone while a similar quantity goes to the underflow outlet in the
liquid-liquid case. This characteristic may suggests that the velocity field
of the continuous flow differs for the two different cases.
Because of the centrifugal force, the solid particles move outward until
they reach the wall and fall to the underflow outlet. Therefore, the
boundary layer is an important zone for this case and should be considered
in any modeling (Bloor et al., 1980). In the LHC for lighter dispersed
phase, more attention has to be centered on the region away from the wall
where the separation occurs.
More information about the differences between solid and liquid hydrocyclones
can be found in Thew (1986).
Two textbooks that condense pioneering works in hydrocylones and fundamental
theories, including experimental data, design, and performance aspects, are Bradley
(1965) and Svarovsky (1984). Both refer in most of the chapters to solid hydrocyclones
with only a small section in liquid-liquid separation and other application areas.
2.1 Solid Hydrocyclones
The hydrocyclone was introduced after World War II by the Dutch State Mines as
a new tool to separate dispersed solid material from a liquid of lower density (Rietema,
1961). Although widely used nowadays, the selection and design of hydrocyclones are
10
still empirical and experience based. Even though quite a few hydrocyclone models are
available, the validity of these models for practical applications has still not been
established (Kraipech et al., 2000). A thorough review of the different available models
can be found in Chakraborti and Miller (1992) and Kraipech et al. (2000).
The models can be divided into empirical and semi-empirical, analytical solutions
and numerical modeling (Chakraborti and Miller, 1992). The empirical approaches are
based on correlations of the key parameters, considering the separator as a black box. The
semi-empirical approach is focused on the prediction of the velocity field in the main
flow using existing data as a major support. The analytical and numerical solutions solve
the non-linear Navier-Stokes Equation. The first one is a mathematical solution, which is
achieved neglecting some of the terms of the momentum balance equation. The
numerical solution uses the power of computational fluid dynamics to develop a
numerical simulation of the flow. As Svarovsky (1996) comments, it seems that the
analytical flow models have been abandoned in favor of numerical simulations.
Based on the experimental data taken by Kelsall (1952) using an optical method,
many researchers have attempted to correlate the velocity field inside the hydrocyclone,
especially the tangential velocity. It can be determined using the following relationship
(Kelsall, 1952, see also Bradley and Pulling, 1959):
Constant
n
Wr (2.1)
This implies that the tangential velocity (W) increases as the radius (r) decreases
for positive values of the empirical exponent (n). The exponent, n is usually between 0.5
and 0.9 (Svarovsky, 1984) in the outer vortex, while in the core region it is close to -1
(see Figure 2.1). If n = 1 a free vortex is obtained where a complete conservation of
11
angular momentum is implied or no viscous effect is considered. However, if n = -1 a
forced vortex or a solid body rotation type is expected. Also, Kelsall's results are an
evidence of the low dependence of the tangential velocity on the axial position.
Figure 2.1 Tangential Velocity Diagram
Analytical flow models have been pursued by Bloor and Ingham for many years
(Bloor and Ingham, 1973, 1984 and 1987 and Bloor, 1987). The momentum and
conservation of mass equations are mathematically solved for an incompressible and
inviscid fluid using the stream function concept in an axi-symmetric flow. Kang (1984)
and Kang and Hayatdavoudi (1985) follow this approach. But unlike the Bloor and
Ingham's model, a cylindrical coordinate system was used instead of spherical. In this
work, it was assumed that the velocities do not depend on the axial position. Kang
considered that the axial and radial velocity obtained from this inviscid model can be
applied without serious error. However, the addition of the turbulence effect had to be
12
included for the tangential component. A constant eddy viscosity was considered to
account for the turbulence fluctuation following a procedure similar to Rietema (1961).
Presently, numerical simulations or CFD are used widely to investigate flow
hydrodynamics. As expressed by Hubred et al. (2000), the solution of the Navier Stokes
Equations for simple or complex geometry for non-turbulent flow is feasible nowadays.
But current computational resources are unable to attain the instantaneous velocity and
pressure fields at large Reynolds numbers even for simple geometries. The reason is that
traditional turbulence models, such as k-, are not suitable for this complex flow
behavior. On the other hand, more realistic and complicated turbulence models increase
the computational times to inconvenient limits.
The flow inside hydrocyclones has been numerically simulated by Rhodes et al.
(1987). A commercial computer code, PHOENICS, was used to solve the required partial
differential equations which govern the flow. Prandtl mixing-length model was used to
account for the viscous momentum transfer effect. In further work, Hsieh and Rajamani
(1991) (see also Rajamani and Hsieh, 1988; Rajamani and Devulapalli, 1994) used a
modified Prandtl mixing-length model with a stream function-vorticity version of the
equation of motion. Good agreement with experimental data was observed in this study.
The authors mention that the key for success is choosing the appropriate turbulence
model and numerical solution scheme. In 1997, He et al. used a fully three dimensional
model with a cylindrical coordinate system and curvilinear grid for the calculation of the
flow field. A modified k- turbulence model was proved to achieve good results.
In most of the work reviewed in the previous paragraph, excluding Rhodes et al.
(1987), the models were evaluated through comparison with laser-doppler anemometry
13
(LDA) data. LDA has many advantages over other techniques. It is not as tedious as the
optical technique and does not cause flow distortion like the Pitot tubes (Chakraborti and
Miller, 1992). Many researchers have used this technique to measure the velocity field
and the turbulence intensities (Dabir, 1983; Fanglu and Wenzhen, 1987; Jirun et al.,
1990; Fraser and Abdullah, 1995).
2.2 Liquid Hydrocyclones, LHC
The modern renaissance in deoiling hydrocyclones was instigated by Martin
Thew and Derek Colman at the University of Southampton (Young et al., 1990). Works
published in the First International Conferences on Hydrocyclones (Colman et al., 1980;
Colman and Thew, 1980; Thew et al., 1980) were the motivators that generated the
application of this technology in the field a few years later. Today, the deoiling
hydrocyclones have become the standard world-wide equipment for cleaning produced
water offshore and have extensive use on-shore (Thew and Smyth, 1997).
A thorough review of LHC is found in Thew and Smyth (1997). As mentioned in
this work, in spite of the steady advances in analytical theories, and more recently CFD
simulations, hydrocyclones have principally been designed based on experimentation. A
comprehensive review on LHC is given below in four sections. Starting from the
modeling, field applications and experimental studies and finally, the velocity field
measurements and CFD simulations. These topics are treated separately because of their
significance to the present study.
14
2.2.1 LHC Modeling
From extensive experimental tests, Colman and Thew (1983) developed some
correlations to predict the migration probability curve, which defines the separation
efficiency for a particular droplet size in a similar way that the grade efficiency does for
solid particles (see 2.1.6 Grade efficiency in Svarovsky, 1984). Later it was found that the
optimized Stokes Number vs. Reynolds Number correlation used in this work was
erroneous (Nezhati in Thew and Smyth, 1997). However, relevant conclusions can be
extracted from this study, such as that the separation efficiency is independent of the split
ratio in the range 0.5 to 10%.
Seyda and Petty (1991) evaluated the separation potential of the cylindrical tail
pipe section. A semi-empirical model to predict the velocity field in a cylindrical
chamber was developed to calculate the particle trajectories, and hence, the grade
efficiency. In the model, the axial velocity was assumed to be independent of the axial
location and a constant eddy viscosity was considered. The theoretical results showed an
optimum split ratio, as opposed to previously reported results, and an increment in the
efficiency when the feed flow rate was increased.
Estimation of LHC efficiency based on a droplet trajectory was the target of
Wolbert et al. (1995) work. The velocity distribution in the tapered section of Colman
and Thew's design was modeled. This was achieved using a modified Helmholtz law for
the tangential velocity, a polynomial correlation for the axial component, and the
continuity equation and wall condition (Kelsall, 1952) for the radial velocity. The
importance of the tail pipe section to the LHC separation efficiency was confirmed by
comparing the model with experimental results.
15
An extension of Bloor and Ingham (1973) model (see section 2.1 Solid
Hydrocyclones) for LHC was elaborated by Moraes et al. (1996). The modification takes
into account the difference in the split ratio for liquid and solid hydrocyclones. Although,
this model is sophisticated, results shown by the authors, where no reverse flow is
achieved in the parallel section, disagree with existing data.
2.2.2 Field Applications
Field trials of deoiling units began in 1983-84, with the first permanent
installation in the North Sea and Bass Strait, Tasmania, in 1985. By the end of 1985, a
North Sea installation of 42 units in parallel was handling nearly 15 m
3
/min (135,860
bpd) successfully (Thew and Smyth, 1997). The field tests conducted by Serck Baker
have demonstrated that a single deoiling unit can maintain effluent oil content below 300
ppm in spite of the large fluctuations in oil content of the inflow up to 2000 ppm. A
comparison of this field data with laboratory measurements showed that two or three
units of hydrocyclones in series can provide substantial improvements (Colman et al.,
1984).
Meldrum (1988) discussed the operational performance of the four-in-one
hydrocyclone concept on the Murchison platform. It was found that the separation
efficiency falls for low flow rates as well as for high flow rates. This was attributed to the
low swirl generated for the lower flow rate limit and due to droplet break up in the higher
flow rate case. Meldrum also found that the efficiency increases as the split ratio
increases until it gets to a point where it remains constant.
16
Usually, hydrocyclones have been used, where adequate system feed pressure for
satisfactory operation is present. Flanigan et al. (1992) revealed successful field trials
with a low shear progressive cavity pump that overcame this limitation.
LHC have been successfully applied not only offshore but also in standard
oilfields. Stroder and Wolfenberger (1994) showed how this technology can be applied to
high water cut electric submergible pump (ESP) wells, as a much more economical
alternative instead of expanding the conventional water separation facilities. Also, good
results in application of hydrocyclones for heavy oil treatment were achieved by Hashmi
et al. (1996). A two stage hydrocyclones system accomplished similar performance to
that of the free water knockout (FWKO) vessels.
The oil industry has realized the benefits of downhole separation. High water cut
wells are produced re-injecting the water and pumping the oil to the surface. Field trials
and description of this application using LHC with an ESP system are found in Bowers et
al. (1996). As it is expressed by the authors, this technology has the potential to become
as significant a revolution in oilfield production as the LHC itself was to the oil industry
in the 1980's.
2.2.3 Experimental Studies
Before reviewing the experimental studies, let us consider the definition of
underflow efficiency as given below (Young, 1990):
Feed the at esented Pr Oil
Discarded Oil
(2.2)
Utilizing continuity equation:
u u i i o o
Q k Q k Q k (2.3)
17
where k is the concentration, Q is the flowrate and the subscripts o, i, and u are for the
overflow, inlet and underflow streams.
Rewriting Equation (2.2), yields:
i i
u u
i i
o o
Q k
Q k
1
Q k
Q k
(2.4)
Since very small amount of flow is taken out of the overflow, Q
u
/Q
i
is almost
equal to one. This is the basis used by many authors and also in the current study when
considering oil/water separation. This efficiency is called underflow purity, defined as
follows:
i
u
u
k
k
1 (2.5)
If k
u
tends to 0,
u
becomes 1. On the other hand, if k
u
is equal to k
i
,
u
is 0. In the
latter case, the hydrocyclone splits the flow without achieving any separation.
Colman et al. (1980) examined oil/water separation efficiency of a series of LHC.
In this work, the performance criterion used was the underflow purity,
u
. Important
observations can be made from this study. The
u
is independent of the k
i
within a range
of 100 to 1000 ppm. The authors concluded that this is a sign of no interaction between
the oil droplets. Also, constant values of
u
were found for different split ratios. However,
for split ratio values less than 2.5% it was found that
u
begins to decrease. It was also
observed that under these conditions, breakdown of the flow structure finally leads to a
complete loss of separation.
Nezhati and Thew (1987) investigated the effect of variation in the inlet area and
other parameters, such as temperature, on the performance of the LHC. The principal
18
objective of this work was to see the variation of dimensionless groups, namely, Stokes,
Euler and Hydrocyclone Number with the flow conditions. Relevant conclusions from the
experimental results are that the separation increases as the cylindrical length (tail pipe
section) is increased and that the pressure drop increases following a simple exponential
relationship with flowrate, given by
n
i
Q P (2.6)
As mentioned in section 1.3 LHC Geometry, Young et al. (1990) searched
through a broad set of experiments for optimum dimensions of the LHC. Similar to
Nezhati and collaborators, Young found that the separation efficiency increases with the
underflow length until it gets to a point where no additional separation occurs. Contrary
to this, as the inlet chamber length is increased the separation efficiency reduces. Other
variables studied were the angle of the conical section, the overflow outlet diameter, the
underflow diameter and the oil properties. In this work, it is noted that the oil droplet size
distribution at the inlet of the LHC has the greatest impact on the separation, namely, that
the bigger the droplets are, the better the separation will be.
In 1991, Weispfennig and Petty explored the flow structure in a LHC using a
visualization technique (laser induced fluorescence). Different types of inlets were
studied including an annular entry. A parameter that measures the strength of the swirling
flow, the Swirl Number, was used to characterize most of the results. This is defined as
the ratio of the axial flux of the axial component of angular momentum to the axial flux
of the tangential component of angular momentum. Vortex instability and recirculation
zones were strongly dependent on the Swirl Number and a characteristic Reynolds
Number.
19
The performance of a small hydrocyclone was summarized by Ali et al. (1994).
Deoiling hydrocyclones of 10 mm-diameter have achieved high performance with cut
size as low as 4 microns. The cut size (d
50
) is the particle size which has a 50% chance of
being separated.
Experiments have also been carried out in dewatering hydrocyclones where the
dispersed phase is water and the continuous phase is oil (Smyth et al., 1980, 1984; Smyth
and Thew, 1987; Smyth, 1988; Young, 1993; Sinker and Thew, 1996). Due to the high
viscosity of the oil, this type of separation is more difficult than the deoiling case but with
an adequate geometry good results may be achieved.
2.2.4 Velocity Field Measurements
Flow field measurements within hydrocyclones can be obtained using
photographic, optical, Pitot tubes and Laser Doppler Anemometry (LDA) (Chakraborti
and Miller, 1992). LDA has been the preferred method in the last two decades because it
permits high speed data acquisition and also because it is a non intrusive technique that
consequently does not cause any flow perturbation.
Thew et al. (1980), used a Residence Time Distribution (RTD) technique to
complement the information gathered from LDA measurements. The authors used a
tracer method in zones where the LDA is limited, i.e. near the boundary wall. Later,
Thew et al. (1984) showed results using the same technique in an improved hydrocyclone
design. The effect of changes in operational parameters, such as split ratio and inlet flow
rate, on the RTD were studied. Good separation efficiency was achieved using split ratios
down to 1%, as opposed to previously reported results.
20
Colman (1981) measured the velocity field in four different hydrocyclone
configurations. Axial and tangential velocities at different axial locations were acquired
using LDA technique. Detailed information of the flow structure was used to improve the
LHC geometry. The observed tangential velocity was a combination of a semi-free vortex
in the outer region and a forced vortex in the core region. In most of the cases, the axial
velocity measured has a reverse flow region in the LHC axis which is inside the forced
vortex region. Both axial and tangential velocities are reduced close to the cylindrical
section wall as the flow moves downward, due to the frictional losses. The reverse flow
begins to decrease as the swirl intensity decays.
Analyzing the LDA axial velocities (Figure 2.1), Colman et al. (1984) concluded
that 85% of the fluid that exits the overflow outlet comes from the reverse flow contained
in the tapered and tail pipe section, while the rest (15%) is made up from the radially
inward moving fluid at the top wall boundary layer. This effect is known as a short circuit
where part of the feed flow rate goes directly to the reject orifice.
Similar results to Colmans (1981) were obtained by Hargreaves (1990) in a
single LHC. Several flow rates and split ratios were explored, measuring the velocity and
the turbulence quantities. It was confirmed that the reverse flow is a body rotation type
and the magnitude of the axial velocity was found to be four times greater than that of the
mean axial velocity in the cylindrical section, and six times more than in the tapered
region. In this study a CFD simulation flow field was compared with LDA
measurements. The turbulence model used in the numerical solution was the Four
Equation Algebraic Stress Model. A review of this modeling and CFD applied to
deoiling hydrocyclones is found in Hargreaves and Silvester (1990).
21
Figure 2.1 Axial Velocity Profile From Colman (1984)
22
The rankine vortex behavior of the tangential velocity was confirmed by
Weispfennig et al. (1995). LDA measurements in the cylindrical section of the
hydrocyclone show how the angular momentum flux ratio decreases with the axial
distance. A flow control was used in order to increase the angular momentum. Parks and
Petty (1995) solved a numerical model using a constant eddy viscosity Boussinesq
approximation to predict the angular momentum distribution in the LHC.
After a critical analysis of the extensive literature available on LHC, it becomes
evident that researchers have directed great efforts toward understanding not only the
separation mechanisms but also the highly complex velocity and pressure field within
hydrocyclone. A large number of experiments have permitted the design of successful
configurations for specific industrial applications. However, it is the authors impression
that an optimum design can be achieved for each range of operational conditions. A
robust mechanistic model, in which the swirling motion and the collateral mechanisms
that influence the LHC efficiency are predicted, is developed in this study as described in
chapter III. This model has the potential to become an excellent tool to predict
performance of existing design, or even to design optimum LHC over a broad range of
operational conditions.
23
CHAPTER III
LHC MECHANISTIC MODEL
3.1 Overview
The mechanistic model is an intermediate approach between the empirical
approach and the exact solution. In this approach a simplified physical model is built,
which attempts to describe closely the nature of the physical phenomenon. This physical
model is then expressed mathematically to provide a tool for prediction and design
purposes. The closer the physical model is to the real phenomenon, the better the
mathematical model is, as well as its prediction. One must remember that a mechanistic
model is not a rigorous solution as the physical model is approximated by taking into
consideration the most important processes, neglecting other less important effects that
can complicate the problem without considerably adding to the accuracy of the solution.
The present work focuses on the development of a simple mechanistic model for
deoiling hydrocyclones, which captures the nature of the hydrodynamic flow behavior in
the LHC. The model is capable of predicting the separation efficiency and the pressure
dropflow rate relationship. In order to obtain these desired output, the LHC Mechanistic
Model needs as input variables, the geometry of the LHC and the operational conditions
such as fluid properties, flow rate and feed droplet size distribution.
The structure of the model can be seen in Figure 3.1. The first step of the model is
the prediction of a parameter known as swirl intensity. With the swirl intensity
24
the velocity field of the continuous phase is determined within the LHC, in terms of axial,
tangential and radial velocities. The model then predicts the separation efficiency and the
pressure drop from the inlet to the underflow outlet of the LHC.
Migration Probability Underflow Purity
Separation
Efficiency
Droplet Trajectories Pressure Drop
Velocity Field
Swirl Intensity
Figure 3.1 LHC Mechanistic Model Structure
The separation efficiency is computed based on droplet trajectories of the
dispersed phase and can be expressed in two modes: the migration probability curve and
the underflow purity (see section 2.2.3 Experimental Studies). The former yields the
separation probability for a specific droplet size, while the latter gives the ratio of the oil
concentration at the clean stream to the one at the inlet, for a given feed droplet size
25
distribution. All these concepts will be explained in more detail in the subsequent
sections.
There are some key parameters that must be considered in the LHC Model. The
angle of the tapered section is one of them. As the angle is increased, the tangential
velocity will be increased and with this the centrifugal force. This may suggest that better
separation can be achieved. However, as the angle is increased the resulting smaller cross
sectional area increases the axial velocity of the fluid which will give less residence time
for the droplets to separate. Thus, there is a compromise between the centrifugal force
and the residence time. The model considers this relationship with the prediction of the
swirl intensity, which is a parameter that defines the strength of the swirling motion
compared to the mean axial velocity.
Another geometrical parameter to be considered is the inlet configuration. The
swirl intensity, as well as the velocity field, is strongly dependent on how the swirling
motion is promoted at the inlet. In the present model the two most commonly used inlet
configurations are included, the involute single inlet and the twin inlets (see Figure 1.2).
The model does not consider the shape of the inlet at all, as rectangular and circular inlets
are treated in the same manner. Only the cross sectional area is considered crucial in the
calculations.
As mentioned before, some effects have to be neglected in order to arrive at a
sufficiently simple solution. The main assumptions that reduce the numerical effort are
listed as follows: 1) axisymmetric flow is considered where there is no variation in the
tangential component; 2) no deformation or interactions within the droplets, namely, no
coalescence or droplet break up occur; 3) no presence of a gas core and 4) no turbulence
26
effect on the droplet trajectory. Some of these assumptions as well as others will be
explained in greater detail in later sections.
3.2 Swirl Intensity
Diverse definitions of swirl intensity have been used by researchers in different
fields. The importance of this parameter, characterizing the swirling flow in liquid-liquid
hydrocyclones, is recognized by Weispfenning and Petty (1991) (see section 2.2.3
Experimental Studies) and also by Thew and Smith (1997). In both cases the swirl
number concept was used. For deoilers the swirl number is commonly within the range of
8-10 which is high enough to achieve a good separation, but low enough to avoid droplet
break up and vortex core instability (Thew and Smyth, 1997).
In the current model the swirl intensity, , is defined as the ratio of the rate of
tangential to total momentum flux at a specific axial location, given by (Mantilla, 1998
and Chang and Dhir, 1994):
avz
2 2
z c
R
0
c
U R
uwrdr 2
z



(3.1)
where u and w are the axial and tangential velocities of the continuous fluid, respectively,
r is the radial position,
c
is the continuous phase density and R is the radius. U
av
is the
bulk axial velocity and the subscript z is for a given axial position.
Several published data sets on cylindrical cyclones indicate that the swirl intensity
decays exponentially with the axial position (Mantilla, 1998). Mantilla also developed a
modification of Chang and Dhir (1994) correlation to account for fluid properties and
inlet effects. In the current study, based on analysis of experimental data sets, a modified
27
correlation of Mantillas model is developed. The modified correlation takes into account
the semi-angle, , of a tapered section, resulting in:
) ) tan( 2 . 1 1 ( I
M
M
48 . 1
15 . 0
93 . 0
2
T
t
+

,
_


( )
1
]
1

+
,
_

,
_

,
_

,
_

12 . 0
7 . 0
16 . 0
z
35 . 0
4
T
t
) tan( 2 1
Dc
z
Re
1
I
M
M
2
1
EXP (3.2)
This correlation was developed using experimental data for small semi-angles, ,
from 0 to 0.75. However, a good prediction has also been obtained for 3 case (see
section 4.5 Pressure Drop Prediction). Due to this limitation and lack of experimental
data for larger angles, this equation is mainly valid for the tapered and tail pipe sections
of Colman and Thews LHC Design (Figure 1.1).
In the above equation Dc, also shown in Figure 3.1, is the characteristic diameter
of the LHC, measured where the angle changes from the reducing section to the tapered
section in the Colman and Thews Design, and at the top diameter of the 3 tapered
section of the Youngs Design (see section 1.3 LHC Geometry); z is the axial position
starting from Dc.
T
t
M
M
is the ratio of the momentum flux at the inlet slot to the axial momentum flux
at the characteristic diameter position, calculated as follows:
is
c
c c
is c
avc
is
T
t
A
A
A / m
A / m
U m
V m
M
M


&
&
&
&
(3.3)
where V
is
is the velocity at the inlet, U
avc
is the average axial velocity at Dc, m& is the
mass flow rate, A
c
is the cross sectional area at Dc and A
is
is the inlet cross sectional area.
28
Figure 3.1 LHC Characteristic Diameter
The Reynolds number is defined in the same way as for pipe flow with the
caution that it refers to a given axial position, yielding:
c
z avz c
z
D U
Re

(3.4)
where
c
is the viscosity of the continuous fluid.
The inlet factor, I, which was modified from Mantilla (1998), is defined as:
( )
n EXP 1 I (3.5)
where n = 1.5 for twin inlets and n = 1 for involute single inlet.
29
The LHC Mechanistic Model considers only the separation occurring at the
tapered and the tail pipe sections. This is a good assumption for the following reasons:
1) Several researchers have reported that most of the separation is achieved in the low
angle tapered section. 2) It can be expected that the biggest droplets that may separate
close to the inlet section will be separated anyway in the consecutive sections of the LHC
and 3) the length of the inlet and reducing sections is usually less than 10% of the total
length of the LHC.
3.3 Velocity Field
The swirl intensity is related, by definition, to the local axial and tangential
velocities. Therefore, it is assumed that once the swirl intensity is predicted for a specific
axial location, it can be used to predict the velocity profiles (Mantilla, 1998). Both
tangential and axial velocities are calculated following a similar procedure as proposed
by Mantilla (1998). The radial velocity, which is the smallest in magnitude, is computed
considering the continuity equation and the wall effect.
3.3.1 Tangential Velocity
It has been experimentally confirmed that the tangential velocity is a combination
of forced vortex near the hydrocyclone axis and free-like vortex in the outer wall region,
neglecting the effect of the wall boundary layer (Figure 3.1). This type of behavior is
known as a Rankine Vortex.
30
Figure 3.1 Rankine Vortex
Algifri et al. (1988) proposed the following equation for the tangential velocity
profile:

'

1
]
1

,
_

,
_

2
c
c
m
avc
R
r
B EXP 1
R
r
T
U
w
(3.6)
where w is the local tangential velocity, which is normalized with the average axial
velocity, U
avc
, at the characteristic diameter; r is the radial location and R
c
is the radius at
the characteristic location.
T
m
represents the maximum momentum of the tangential velocity at the section
and B determines the radial location at which the maximum tangential velocity occurs.
The following expressions were obtained by curve-fitting several sets of the experimental
data.
31

m
T (3.7)
Involute Single Inlet:
7 . 1
7 . 55 B

(3.8)
Twin Inlets:
35 . 2
8 . 245 B

(3.9)
It can be seen that the above equations are only functions of the swirl intensity, .
Thus, for a given axial position, the tangential velocity is only function of the radial
location and the swirl intensity.
3.3.2 Axial Velocity
In swirling flow the tangential motion gives rise to centrifugal forces which in
turn tend to move the fluid toward the outer region (Algifri 1988). Such a radial shift of
the fluid will result in a reduction of the axial velocity near the axis, and when the swirl
intensity is sufficiently high, reverse flows can occur near the axis. This phenomenon
causes a characteristic reverse flow in the LHC axis, which allows the separation of the
different density fluids.
A typical axial velocity profile for LHC is illustrated in Figure 3.1. Here, the
positive values represent the downward flow near the wall, which is the main flow
direction, and the negatives values represent the upward reverse flow near the LHC axis.
The reverse radius, r
rev
, is the radial position where the axial velocity is equal to zero.
32
Figure 3.1 Axial Velocity Diagram
To predict the axial velocity profile, a third-order polynomial equation is used
with the proper boundary conditions. The general form is as follows:
4 3
2
2
3
1
a r a r a r a ) r ( u + + + (3.10)
where a
1
, a
2
, a
3
and a
4
are constants. The boundary conditions considered are:
1. 0
dr
) R r ( du
z

the velocity is maximum at the wall,


2. 0 ) r r ( u
rev
zero velocity at the location of reverse flow, r
rev
,
3. 0
dr
) 0 r ( du

the velocity is symmetric about the LHC axis and


4.
2
z
c
R
0
avz c
R U rdr ) r ( u 2
z

Mass Conservation.
Substituting the boundary conditions in Equation (3.10), yields the axial velocity
profile, which is a function of the swirl intensity, only:
33
1
C
7 . 0
R
r
C
3
R
r
C
2
U
u
2
z
3
z avz
+ +

,
_

,
_

(3.11)
7 . 0
R
r
2 3
R
r
C
z
rev
2
z
rev

,
_

,
_

,
_

(3.12)
358 . 0
z
rev
293 . 0
R
r
(3.13)
Several assumptions are implicit in these equations. First, an axisymetric
geometry is imposed. Then, the effects of the boundary layer are neglected, and finally
the mass conservation balance does not consider the split ratio. The last assumption can
be considered a good approximation for small values of split ratios used in the LHC,
usually less than 10%.
3.3.3 Radial Velocity
The radial velocity, v, of the continuous phase is very small, and has been
neglected in many studies. In our case, in order to track the position of the droplets in
cylindrical and conical sections, the continuity equation and wall conditions suggested by
Kelsall (1952) and Wolbert, (1995) are used for the radial velocity profile, yielding:
) tan( u
R
r
v
z
(3.14)
The radial velocity is a function of the axial velocity and geometrical parameters.
In the particular case of cylindrical sections, where tan( ) = 0, the radial velocity, v, is
equal to 0.
34
3.4 Droplet Trajectories
The droplet trajectory model is developed using a Lagrangian approach in which
single droplets are traced in a continuous liquid phase. The droplet trajectory model
utilizes the flow field presented in the previous section.
Figure 3.1 presents the physical model. A droplet is shown at two different time
instances, t and t + dt. The droplet moves radially with a velocity V
r
and axially with V
z
.
It is assumed that in the tangential direction the droplet velocity is the same as the
continuous fluid velocity, as no force acts on the droplet in this direction. Therefore, the
trajectory of the droplet is presented only in two dimensions, namely r and z.
Figure 3.1 Droplet Velocities
During a differential time dt, the droplet moves at velocity V
r
= dr/dt in the radial
direction and V
z
= dz/dt in the axial direction. Combining these two equations and
solving for the axial distance yields the governing equation for the droplet displacement:
dr
V
V
z
V
V
dt
dr
dt
dz
dr
dz
r
z
r
z
(3.15)
35
Neglecting the axial buoyancy force (no-slip condition), the droplet axial velocity,
V
z
becomes the axial velocity of the fluid, u. This simplification is reasonable when the
acceleration due to the centrifugal force in the radial direction is thousand times larger
than the aceleration of gravity. Due to this, the LHC is not sensitive to external
movements and it can be installed either horizontally or vertically.
The droplet velocity in the radial direction is equal to the fluid radial velocity, v,
plus the slip velocity, V
sr
. Rearranging Equation (3.15) yields the total trajectory of the
droplet, namely:
r
V v
u
z
2
1
r r
r r
sr

,
_

(3.16)
The only unknown parameter in Equation (3.16) is the slip velocity, which can be
solved from a force balance on the droplet in the radial direction, as shown Figure 3.2.
Figure 3.2 Forces Acting on a Droplet
Assuming a local equilibrium momentum yields:
4
d
V C
2
1
6
d
r
w
) (
2
2
sr c D
3 2
d c

(3.17)
where the left side of the equation is the centripetal force, and the right side is the drag
force. Solving for the radial slip velocity, results in:
36
2
1
D
2
c
d c
sr
C
d
r
w
3
4
V

,
_

,
_


(3.18)
where d is the droplet diameter;
d
is the density of the dispersed phase and C
D
is the
drag coefficient calculated using the following relationship (Morsi and Alexander, 1971
and Hargreaves, 1990):
2
d
3
d
2
1 D
Re
b
Re
b
b C + +
(3.19)
The coefficients b are dependent on the Reynolds Number of the droplets,
defined as:
c
sr c
D
V d
Re

(3.20)
The values for the b coefficients as function of the range of Re
D
are shown in
the Table 3-1.
Table 3-1 Drag Coefficient Constants
Range b
1
b
2
b
3
Re
D
< 0.1 0 24 0
0.1 < Re
D
< 1 3.69 22.73 0.0903
1 < Re
D
< 10 1.222 29.1667 -3.8889
10 < Re
D
< 100 0.6167 46.5 -116.67
Finally, a numerical integration of Equation (3.16) determines the axial location
of the droplet as a function of the radial position. The trajectory of a droplet of a given
37
size is mainly a function of the LHC velocity field and the physical properties of the
dispersed and continuous phases.
3.5 Separation Efficiency
The separation efficiency of the LHC can be determined based on the droplet
trajectory analysis presented above. Starting from the cross sectional area corresponding
to the LHC characteristic diameter, it is possible to follow the trajectory of a specific
droplet, and determine if it is either able to reach the reverse flow region and be
separated, or if it reaches the LHC underflow outlet, dragged by the continuous fluid and
carried under.
As illustrated by Figure 3.1, the droplet that starts its trajectory from the wall
(r = Rc) is not separated, but rather carried under. However, if the starting location is at
r < Rc, the chance of this droplet to be separated increases. When the starting point of the
droplet trajectory is the critical radius, r
crit
, the droplet reaches the reverse radius, r
rev
, and
is carried up by the reverse flow and is separated.
Therefore, assuming homogeneous distribution of the droplets, the efficiency for a
droplet of a given diameter, (d), can be expressed by the ratio of the area for which the
droplet is separated, defined by r
crit
, over the total area for flow. This assumption has also
been applied by other researchers (Seyda and Petty, 1991; Wolbert et al., 1995 and
Moraes et al., 1996).
38
Figure 3.1 Droplet Trajectory and Migration Probability
As proposed by Moraes et al. (1996), the efficiency is given by:

'

< <


c crit
c crit rev 2
rev
2
c
2
rev
2
crit
rev crit
R r if , 1
R r r if ,
r R
r r
r r if , 0
) d (
(3.21)
Repeating this procedure for different droplet sizes, the migration probability
curve is obtained (Figure 3.2). This function has an S shape and represents the
separation efficiency, (d), vs. the droplet diameter, d. It can be seen that small droplets
39
have an efficiency very close to zero and as the droplet size is increased, (d) increases
sharply until it reaches d
100
, which is the smallest droplet size with a 100% probability to
be separated.
Figure 3.2 Migration Probability Curve
The migration probability curve is the characteristic curve of a particular LHC for
a given flow rate and fluid properties. This curve is independent of the feed droplet size
distribution and it is used in many cases to compare the separation of a given LHC
configuration.
Using the information derived from the migration probability curve and the feed
droplet size distribution, the underflow purity,
u
, can be determined as follows:


i i
i
i i
u
V
V ) d (
(3.22)
where
u
is expressed in %, and V
i
is the percentage volumetric fraction of the oil
droplets of diameter d
i
.
The underflow purity is the parameter that quantifies the LHC capacity to separate
the dispersed phase from the continuous one (see section 2.2.3 Experimental Studies).
40
3.6 Pressure Drop
The pressure drop from the inlet to the underflow outlet is calculated using a
modification of the Bernoullis Equation:
( ) L sin g h h U
2
1
P V
2
1
P
c f cf c
2
U c U
2
is c is
+ + + + + (3.23)
where
c
is the density of the continuous phase; P
is
and P
u
are the inlet and outlet
pressures respectively; V
is
is the average inlet velocity and U
u
is the underflow average
axial velocity; L is the hydrocyclone length, is the angle of the LHC axis with the
horizontal; h
cf
corresponds to the centrifugal force losses and h
f
is the frictional losses.
The frictional losses are calculated similar to pipe flow:

2
) z ( V
) z ( D
z
) z ( f ) z ( h
2
R
f

(3.24)
where f is the friction factor and V
R
is the resultant velocity.
In the case of conical sections, all parameters in Equation (3.24) change with the
axial position, z. The conical section is divided into m segments and assuming
cylindrical geometry in each segment, the frictional losses can be considered as the sum
of the losses in all the m segments, as follows.
)
2
z
) 1 n 2 (( at V
2
D D
z
2
) z ( f
h
2
R
m
1 n
n 1 n
) conical ( f


(3.25)
The resultant velocity, V
R
, is calculated as the vector sum of the average axial and
tangential velocities, The annular downward flow region is only considered, as presented
in the following set of equations:
41
2
Z
2
Z
2
R
W U ) z ( V + (3.26)

2
0
R
r
2
0
R
r
z z
rev
z
rev
rdrd
Wrdrd
W
(3.27)
For simplification purposes, the average axial velocity in Equation (3.26), Uz, is
calculated assuming plug flow, namely, Uz is equal to the total flow rate over the annular
area from the wall to the reverse radius, r
rev
. The Moody friction factor is calculated using
Halls Correlation (Hall, 1957).

'

1
]
1

,
_


+
3 / 1
6
4
) z Re(
10
) z ( D
10 x 2 1 0055 . 0 ) z ( f (3.28)
where is the pipe roughness and Re is the Reynolds Number, calculated based on the
resultant velocity computed in Equation (3.26).
The centrifugal losses are the most important one in Equation (3.23), and account
for most of the total pressure drop in the LHC. They are calculated using the following
expression:
( )

u
rev
R
r
2
u
cf
dr
r
) r ( nW
h (3.29)
where W
u
is calculated from Equation (3.27) at the underflow outlet and the centrifugal
force correction factor, n = 2 for twin inlets, and n = 3.2 for involute single inlet.
The centrifugal force correction factor compensates for the use of Bernoullis
Equation under a high rotational flow condition. Its meaning is similar to the kinetic
energy coefficient used to compensate for the non-uniformity of the velocity profile in
42
pipe flow (Munson et al., 1994). Rigorously, the Bernoullis Equation is valid for a
streamline and the summation of the pressure, the hydrostatic and the kinetic terms can
only be considered constant in all the flow field if the vorticity is equal to zero.
3.7 LHC Mechanistic Model Code
In order to validate and compare the model with published experimental data, a
computer code was built, which includes the equations shown in this chapter. The
program was developed in Visual Basic for Excel Application. Excel/VBA platform can
provide great advantages such as user-friendly interface forms and easiness to manipulate
the output data.
Figure 3.9 presents the multipage form used in the computer code where the user
can interact with the program. All the input such as geometry, operating conditions, fluid
properties and feed droplet size distribution are located in this form as separate folders.
Buttons to run the program, as well as save and open input cases are also included. All
the results of the program are presented in the worksheets of the Excel Application.
The code uses mainly two different numerical methods to obtain the results. The
tangential velocity, given by Equation (3.27), is solved using the Trapezoidal Rule, and
for the droplet trajectory, a fourth-order Runge-Kuttta method is used to solve Equation
(3.16). Also, a commercial program (Mathematica 4.0) was used to verify the resulting
numerical values given by the computer code.
43
Figure 3.9 LHC Mechanistic Model Code
44
CHAPTER IV
RESULTS AND DISCUSSION
This chapter presents a comparison between 35 published experimental data sets
and the prediction of the LHC mechanistic model developed in the present study. The
outline is similar to the previous chapter, starting from the swirl intensity prediction and
ending with the pressure drop prediction. In each section, the source of the experimental
data is described followed by the results and discussion. The only section that differs
from this structure is the droplet trajectory, where only the model predictions are shown.
4.1 Swirl Intensity Prediction
4.1.1 Experimental Data Sets
The swirl intensity, which is the ratio of the local tangential momentum flux to
the total momentum flux, can be obtained from the numerical integration of Equation
(3.1). The numerical method employed for this purpose was the trapezoidal rule. The
experimental data sets used to compare with the swirl intensity predicted by the model,
are described next:
45
Colmans (1981) work, where the flow field was measured using a
Laser Doppler Anemometer. In this study four different hydrocyclone
designs were used. However, only designs II, III and IV, as named in
the original work, are used here. The configurations of these
hydrocyclones are shown in Figure 4.1 and the geometrical and
operational conditions of each study case are detailed in Table 4-1 and
Table 4-2, respectively.
Figure 4.1 Colmans Designs (1981)
46
Table 4-1 Geometrical Parameters of Colmans Designs (1981)
Case Design Dc(mm) L As
1

2
D
2
L
2
Ds Ls Du
1 II 58 30Dc 0.125 - 90 0.5Dc 22Dc - - 0.14Dc
2 III 30 20Dc 0.0625 10 10 0.5Dc 21Dc 2Dc 3Dc 0.14Dc
3 IV 30 - 0.0625 10 0.67 0.5Dc 21Dc 2Dc 3Dc 0.14Dc
Table 4-2 Operational Conditions of Colmans Designs (1981)
Case Design Dc(mm) Flowrate (lpm) T (C) F (%)
1 II 58 175 25 10
2 III 30 60 25 10
3 IV 30 60 25 10
where F(%) is the split ratio.
Hargreaves data (1990) were taken with a LDA in a similar LHC
configuration as that of Colmans Design IV, but with a single involute
inlet instead of the twin inlets. The cross sectional area of the inlet is 644
mm
2
. Figure 4.1 is used as a reference for the rest of the geometrical
parameters expressed in Table 4-3. The flow rate used in these cases is
180 lpm and two different split ratios, F, were tested. Case 4 with F =
10%, which corresponds to Du = 0.117 Dc, and the Case 5 with F = 1%,
which corresponds to Du = 0.04 Dc.
47
Table 4-3 Geometrical Parameters of Hargreaves (1990)
Case Design Dc(mm)
1

2
D
2
L
2
Ds Ls
4 and 5 IV 60 10 0.6365 0.5Dc 15Dc 1.5Dc 1.67Dc
The amount of experimental data, presented for each case in the next section,
depends on the availability of axial and tangential velocity measurements at specific axial
locations published by the above mentioned authors.
4.1.2 Results
The experimental data shown in Figure 4.1 to Figure 4.4 correspond to cases 1, 2,
3, 4 and 5 mentioned in the previous section. All the data used from design II and III
(Figure 4.1) correspond to the cylindrical sections with Dc diameter, while the data
shown for design IV correspond to the small angle tapered section.
The results display the swirl intensity versus the dimensionless axial position,
where z is the axial distance from the characteristic diameter (see Figure 3.1).
48
0
1
2
3
4
5
6
0 5 10 15 20 25 30 35
z / Dc
S
w
i
r
l

I
n
t
e
n
s
i
t
y
,


Experimental Data
LHC Mechanistic Model
Figure 4.1 Swirl Intensity Prediction - Case 1
0
0.5
1
1.5
2
2.5
3
3.5
4
0 5 10 15 20 25
z / Dc
S
w
i
r
l

I
n
t
e
n
s
i
t
y
,

Experimental Data
LHC Mechanistic Model
Figure 4.2 Swirl Intensity Prediction - Case 2
49
0
1
2
3
4
5
6
0 5 10 15 20 25 30 35 40 45
z / Dc
S
w
i
r
l

I
n
t
e
n
s
i
t
y
,

Experimental Data
LHC Mechanistic Model
Figure 4.3 Swirl Intensity Prediction - Case 3
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 5 10 15 20 25 30 35 40
z / Dc
S
w
i
r
l

I
n
t
e
n
s
i
t
y
,

Experimental Data F=10% (Case 4)


Experimental Data F=1% (Case 5)
LHC Mechanistic Model
Figure 4.4 Swirl Intensity Prediction - Cases 4 and 5
50
4.1.3 Discussion
It has been experimentally proved by several researchers that the swirl intensity
decays exponentially with axial position in cylindrical pipes due to the wall frictional
losses (Chang and Dhir, 1994 and Mantilla, 1998). The results in Figure 4.1 show that the
swirl intensity in a cylindrical hydrocyclone predicted by the modified model of Mantilla
(1998) agrees very accurately with the experimental data.
The model also shows good agreement for the low angle conical sections (Case 3,
4 and 5). However, there are not sufficient experimental data to ensure the reliability of
the modified correlation to predict the swirl intensity as a function of a variety of angles.
In this sense, a question that remains open is to what extent can the correlation predict the
swirl intensity for larger angles of the conical sections.
4.2 Velocity Field Prediction
4.2.1 Experimental Data Sets
The velocity field predicted by the proposed mechanistic model is compared with
the same experimental data sets used for the swirl intensity prediction, namely, cases 1, 2,
3, 4 and 5. The only new sets of data incorporated are from Case 6 (Hargreaves, 1990),
which is for the same geometry of Case 5 but with a change of the flow rate from 180
lpm to 150 lpm. In the latter case only the axial velocity was published for the lower
angle conical section.
51
4.2.2 Tangential and Axial Velocity Results
Figure 4.1 to Figure 4.5 present the comparison between the data and model
prediction for the tangential velocity at different axial positions. The y-axis of each chart
corresponds to the axis of the LHC, and the x-axis represents the radial position. The
units used originally were conserved, namely, millimeters per second for the tangential
velocity, and millimeters for the radial position.
Next, the axial velocities predicted by the model are compared with the
experimental data at different axial locations in Figure 4.6 to Figure 4.11. In all the charts
the positive values of axial velocities correspond to downward flow, which is the
direction of the main flow, while the negative values represent the reverse flow.
Figure 4.11 shows the axial velocity in the entire cross sectional area of LHC. In
the rest of the figures, the axial velocity is only shown from the center line to the wall of
the LHC.
4.2.3 Discussion
In general, the model predictions are close to the experimental data. The closer
the prediction of the velocities is to the real phenomenon, the better will be the prediction
of the separation efficiency and the pressure drop.
Tangential Velocity
The experimental data of case 1 (Design II) cannot be predicted accurately by an
axy-symmetric model. It seems that if the data are shifted to the right, as if to get a zero
velocity value in the LHC axis, the model predictions will get much closer to the
experimental data. The non-symmetry of this case compared with the others is one of the
52
reasons that this design had an inferior separation performance to the Design III and IV
(Colman, 1981) Cases. Except for Case 1, the model predicts with acceptable accuracy
the tangential velocity at the wall, the peak velocity and the radius where it occurs. The
experimental data and the model display a Rankine Vortex shape, namely, a combination
of forced vortex near to the LHC axis and a free like vortex in the outer region.
It can also be seen from Case 1 that the experimental data and the model
predictions follow the same tendency as the fluid moves downward. The tangential
velocity at the wall decreases and the peak velocity value increases approaching the LHC
axis.
Axial Velocity
The mechanistic model performance is excellent with respect to the axial velocity,
in the downward flow region, and not so good in the reverse flow. Considering the
calculations that the model follows to compute the separation efficiency, the prediction of
the reverse flow velocity profile is not so important. What is really important is the
prediction of the radius of zero velocity since beyond this point the droplet is assumed to
be separated.
Further, it can be observed how the model and the experimental data show a
reduction in the reverse radius as the axial position increases and the swirl intensity
decreases.
53
z / Dc = 9
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 10.5
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 12
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.1 Tangential Velocity Prediction - Case 1
54
z / Dc = 15
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 18
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 21
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.1 Tangential Velocity Prediction - Case 1 (Contd.)
55
z / Dc = 24
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 27
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 28
0
2000
4000
6000
8000
10000
12000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.1 Tangential Velocity Prediction - Case 1 (Contd.)
56
z / Dc = 10.5
0
4000
8000
12000
16000
0 4 8 12 16
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.2 Tangential Velocity Prediction - Case 2
z / Dc = 10.5
0
4000
8000
12000
16000
0 4 8 12
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.3 Tangential Velocity Prediction - Case 3
57
z / Dc = 3.75
F = 10%
0
2000
4000
6000
8000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 7.5
F = 10%
0
2000
4000
6000
8000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 15
F = 10%
0
2000
4000
6000
8000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.4 Tangential Velocity Prediction - Case 4
58
z / Dc = 3.75
F = 1%
0
2000
4000
6000
8000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 7.5
F = 1%
0
2000
4000
6000
8000
0 5 10 15 20 25 30
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 15
F = 1%
0
2000
4000
6000
8000
0 5 10 15 20 25
Radius (mm)
T
a
n
g
e
n
t
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.5 Tangential Velocity Prediction - Case 5
59
z / Dc = 10.5
-6500
0
6500
-30 0 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 12
-6500
0
6500
-30 0 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 15
-6500
0
6500
-30 0 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.6 Axial Velocity Prediction - Case 1
60
z / Dc = 18
-6500
0
6500
-30 0 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 21
-6500
0
6500
-30 0 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 24
-6500
0
6500
-30 0 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.6 Axial Velocity Prediction - Case 1 (Contd.)
61
z / Dc = 27
-6500
0
6500
-30 0 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 28.5
-6500
0
6500
-30 0 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.6 Axial Velocity Prediction - Case 1 (Contd.)
62
z / Dc = 10.5
-7000
0
7000
0 4 8 12 16
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.7 Axial Velocity Prediction - Case 2
z / Dc = 10.5
-8000
0
8000
0 4 8 12
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.8 Axial Velocity Prediction - Case 3
63
z / Dc = 3.75
F = 10%
-6000
0
6000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 7.5
F = 10%
-6000
0
6000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 15
F = 10%
-6000
0
6000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.9 Axial Velocity Prediction - Case 4
64
z / Dc = 3.75
F = 1%
-6000
0
6000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 7.5
F = 1%
-6000
0
6000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 15
F = 1%
-6000
0
6000
0 5 10 15 20 25
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.10 Axial Velocity Prediction - Case 5
65
z / Dc = 3.75
-5000
0
5000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 7.5
-5000
0
5000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.11 Axial Velocity Prediction - Case 6
66
z / Dc = 11.25
-5000
0
5000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
z / Dc = 15
-5000
0
5000
0 5 10 15 20 25 30
Radius (mm)
A
x
i
a
l

V
e
l
o
c
i
t
y

(
m
m
/
s
e
c
)
Experimental Data
LHC Mechanistic Model
Figure 4.11 Axial Velocity Prediction - Case 6 (Contd.)
67
4.3 Droplet Trajectory Prediction
The prediction of the separation efficiency is based on the droplets trajectories.
Once the velocity field of the continuous phase is known, the slip velocity of the droplets
can be calculated and consequently, the trajectory can be determined.
Figure 4.1 and Figure 4.2 present the model prediction for the droplet trajectory
for a case study. The geometry and operational conditions selected for this analysis are
from Case 7, detailed in the experimental data sets of the Separation Efficiency
Prediction (Section 4.4.1). Only model predictions are shown for droplet trajectory due to
non-availability of data. In Figure 4.1, the path that the different droplet sizes follow
within the LHC are shown starting from the radius at Rc only. On the other hand, Figure
4.2 describes the trajectory of a 15 microns droplet only, with different starting trajectory
radial positions.
Reverse Flow
Radius
d=15 m
d=25 m
d=29 m
d=35 m
d=45 m
d=60 m
0
350
700
1050
1400
0 5 10 15
Radius (mm)
A
x
i
a
l

P
o
s
i
t
i
o
n

(
m
m
)
Figure 4.1 Predicted Droplets Trajectories Case 7
68
Reverse Flow
Radius
Droplet Size
Diameter
15 m
0
350
700
1050
1400
0 5 10 15
Radius (mm)
A
x
i
a
l

P
o
s
i
t
i
o
n

(
m
m
)
Figure 4.2 Trajectories of a 15 Microns Droplet Case 7
As can be seen from Figure 4.1, the larger droplets are separated easily, but the
smallest droplets of 25 and 15 microns diameter are not able to reach the reverse flow
region, even when their starting trajectory point is at the wall.
Figure 4.2 demonstrates the concept in which the separation efficiency is
calculated by the model. A 15 microns droplet is dragged by the fluid when it enters the
top cross sectional area near the wall. However, as the droplet is introduced closer to the
LHC axis the probability of this droplet to be separated is increased. It can be observed
that this droplet reaches the reverse flow region when it is released at a radius equal to 11
mm, which is close to the critical radius, r
crit
for this case.
If we consider that the tangential velocity increases considerably as it approaches
the LHC center, and that the axial velocity behaves in an opposite manner, once the
69
droplet is able to reach these zones of very high centrifugal forces and moderate resident
times, it migrates very fast to the center, reaching the reverse flow region and is
separated. This is the reason why the case study droplet (15m diameter) passes
suddenly from a starting radius where it cannot be separated, to one where it is easily
separated.
4.4 Separation Efficiency Prediction
The main objective of the LHC is to produce very clean water through the
underflow outlet, while minimizing the amount of water loss that exits with the oil core
through the overflow outlet. An expression that helps to evaluate the separation
performance is the underflow purity given by Equation (2.2). This variable expresses the
ratio of the difference in concentration from the feed to the underflow stream over the
feed concentration. Another common way of looking at the efficiency of the
hydrocyclone is through the migration probability curve. This kind of chart represents the
separation efficiency of a particular droplet size.
Both, the underflow purity and the migration probability curve predicted by the
model are evaluated through comparisons with published experimental data. In order to
calculate the underflow purity, the LHC model needs as an input the feed droplet size
distribution which is also shown in the results section.
4.4.1 Experimental Data Sets
The experimental data sets used for the separation efficiency comparisons are
from four sources:
70
Case 7 is from Colmans (1981) study. This case is the same as Case 3 but the
experimental temperature was 50C, which corresponds to a water viscosity of 0.55 cp.
The dispersed phase used were solid particles (polypropylene) instead of oil droplets. The
density of these particles at the operational temperature is 0.89 g/cc.
Case 8 includes the geometry and operational conditions reported by Wolbert et
al. (1995). Using Figure 4.1 as a reference, the following table gives the details of the
geometrical parameters for this case. The flow rate used in this case was 32 lpm with an
oil dispersed phase density of 0.902 g/cc. The feed droplet size distribution for Case 7
and 8 are shown in Figure 4.20 and Figure 4.4 respectively.
Table 4-1 Geometrical Parameters, Wolbert et al. (1995)
Case Design Dc(mm)
1

2
D
2
L
2
Ds Ls Di
8 IV 20 10 0.75 0.5Dc 30Dc 2Dc 2Dc 0.35Dc
where Di is the inlet diameter of one of the twin inlets (mm)
Cases 9 to 23 are part of the set of experiments published by Colman et al. (1980)
and Case 24 by Colman et al. (1984). These experimental data sets are for the same
configuration as the one shown in the Table 4-1 and the characteristic diameter and
operational conditions are reported together with the results in Table 4-1.
4.4.2 Migration Probability and Underflow Purity Results
The results of Cases 7 and 8 can be seen in Figure 4.1 to Figure 4.4. Initially the
migration probability curve is illustrated, followed by a chart that contains the underflow
purity and the experimental feed droplet size distribution, and the underflow droplet size
distribution, as calculated by the model. Next, the feed droplet size distribution of Cases
71
9 to 24 are plotted for two different oils, from Kuwait and Forties, as shown in Figure 4.5
and Figure 4.6.
0
10
20
30
40
50
60
70
80
90
100
0 8 16 24 32 40 48 56 64
Droplet Diameter (microns)
S
e
p
a
r
a
t
i
o
n

E
f
f
i
c
i
e
n
c
y

(
%
)
Experimental Data
LHC Mechanistic Model
Figure 4.1 Migration Probability Curve - Case 7
0
10
20
30
40
12.75 15.9 20 25 31.4 39.6 50 62.6 79.8 100 129
Droplet Diameter (microns)
V
o
l
u
m
e
t
r
i
c

F
r
a
c
t
i
o
n

(
%
)
Feed (Experimental) Underflow (LHC Mechanistic Model)
Experimental
u = 91%
LHC Mechanistic Model

u = 93%
Figure 4.2 Underflow Purity,
u
- Case 7
72
0
10
20
30
40
50
60
70
80
90
100
0 8 16 24 32 40 48 56 64
Droplet Diameter (microns)
S
e
p
a
r
a
t
i
o
n

E
f
f
i
c
i
e
n
c
y

(
%
)
Experimental Data
LHC Mechanistic Model
Figure 4.3 Migration Probability Curve - Case 8
0
10
20
30
40
50
1.5 5 7 10 14 20 28 40 56
Droplet Diameter (microns)
V
o
l
u
m
e
t
r
i
c

F
r
a
c
t
i
o
n

(
%
)
Feed (Experimental) Underflow (LHC Mechanistic Model)
Experimental

u = 81%
LHC Mechanistic Model

u = 76%
Figure 4.4 Underflow Purity,
u
- Case 8
73
0
5
10
15
20
25
3.5 5 6.4 8.12 10 12.75 16.25 20 25.15 31.62 39.75 50 81.23 81.23
Droplet Size (microns)
V
o
l
u
m
e
t
r
i
c

F
r
a
c
t
i
o
n

(
%
)
Mean Drop Size =17 m
Oil Density = 0.84 g/cc
0
5
10
15
20
25
3.5 5 6.4 8.12 10 12.8 16.3 20 25.2 31.6 39.8 50 63.7 81.2 100 115
Droplet Size (microns)
V
o
l
u
m
e
t
r
i
c

F
r
a
c
t
i
o
n

(
%
)
Mean Drop Size =35 m
Oil Density = 0.84 g/cc
0
5
10
15
20
25
30
5 6 8 10 13 16 20 25 32 40 50 64 81 100 115 135
Droplet Size (microns)
V
o
l
u
m
e
t
r
i
c

F
r
a
c
t
i
o
n

(
%
)
Mean Drop Size =70 m
Oil Density = 0.84 g/cc
Figure 4.5 Droplet Size Distributions for Kuwait Oil (Colman et al., 1980)
74
0
5
10
15
20
25
6.4 8.12 10 12.75 16.25 20 25.15 31.62 39.75 50 63.73 81.23 100 115
Droplet Size (microns)
V
o
l
u
m
e
t
r
i
c

F
r
a
c
t
i
o
n

(
%
)
Mean Drop Size =41 m
Oil Density = 0.87 g/cc
Figure 4.6 Droplet Size Distribution for Forties Oil (Colman et al., 1980)
The experimental underflow purity results and the one computed by the LHC
model for cases 7 to 24 are described in Table 4-1. The migration probability curve for
the Cases 16, 18, 20, 23 and 24 are reported in Figure 4.7 to Figure 4.9 as examples.
75
Table 4-1 Underflow Purity Results Cases 7 to 24
Case Dc (mm)
Flowrate
(lpm)
Oil
Density
(g/cc)
Mean
Drop Size
( c)
Experimental
Underflow
Purity (%)
Model
Underflow
Purity (%)
7 30 60 0.89 91 93
8 20 32 0.902 81 76
9 30 60 0.87 41 88 86
10 30 40 0.84 35 78 73
11 30 50 0.84 35 82 79
12 30 60 0.84 35 84 83
13 30 70 0.84 35 88 86
14 58 160 0.84 35 72 65
15 58 190 0.84 35 74 69
16 58 220 0.84 35 78 74
17 58 250 0.84 35 81 79
18 58 220 0.84 17 43 42
19 58 250 0.84 17 47 47
20 58 220 0.84 70 96 92
21 58 250 0.84 70 97 93
22 58 220 0.87 41 80 76
23 58 250 0.87 - - -
24 51 150 0.84 - - -
0
10
20
30
40
50
60
70
80
90
100
0 8 16 24 32 40 48 56 64
Droplet Diameter (microns)
S
e
p
a
r
a
t
i
o
n

E
f
f
i
c
i
e
n
c
y

(
%
)
Experimental Data
LHC Mechanistic Model
Figure 4.7 Migration Probability Curve Cases 16, 18 and 20
76
0
10
20
30
40
50
60
70
80
90
100
0 8 16 24 32 40 48 56 64 72 80
Droplet Diameter (microns)
S
e
p
a
r
a
t
i
o
n

E
f
f
i
c
i
e
n
c
y

(
%
)
Experimental Data
LHC Mechanistic Model
Figure 4.8 Migration Probability Curve Case 23
0
10
20
30
40
50
60
70
80
90
100
0 8 16 24 32 40 48 56 64 72
Droplet Diameter (microns)
S
e
p
a
r
a
t
i
o
n

E
f
f
i
c
i
e
n
c
y

(
%
)
Experimental Data
LHC Mechanistic Model
Figure 4.9 Migration Probability Curve Case 24
77
40
50
60
70
80
90
100
40 50 60 70 80 90 100
Experimental Underflow Purity (%)
L
H
C

M
e
c
h
a
n
i
s
t
i
c

M
o
d
e
l

U
n
d
e
r
f
l
o
w

P
u
r
i
t
y

(
%
)
Average Relative Error (%) = -3.7
Average Relative Absolute Error (%)= 4
Figure 4.10 Comparison of Model Underflow Purity and Experimental Data Set
4.4.3 Discussion
Figure 4.10, incorporates all the data sets, showing how well the model predicts
the underflow purity for such broad range of conditions. The characteristic diameter of
the hydrocyclone varied from 20 to 58 mm, and the flowrate studied from 32 to 250 lpm.
The average relative absolute error is 4% and the average relative error is 3.7%, which
confirms what can be seen in the chart, that the model is consistently under predicting the
underflow purity. This difference stems from the discrepancy between the theoretical and
experimental values of the migration probability.
In most of the cases the model estimates lower separation efficiency for droplet
diameters less than 25 microns. On the other hand the biggest droplets that have
experimental efficiencies around 95%, are predicted by the model to be completely
separated. Both effects may compensate each other and give more realistic results.
78
It is interesting to note that the only case in which the model predicts higher
efficiency is Case 7, which is the case with a solid particle dispersion. This may suggest
that the under-prediction of the separation efficiency by the model may be due to the
effect of droplet coalescence.
From Table 4-1 it can be concluded that the model and the experimental data
follow the same trends:
Higher droplet diameters produced better underflow purities (Case 17,
d35,
u
=81% and Case 19, d17,
u
=43%).
The separation efficiency increases at higher flow rates (see Cases 10 to
13).
4.5 Pressure Drop Prediction
In this section the pressure drop from the inlet to the underflow outlet is
calculated by the model and compared with the experimental data described as follows.
4.5.1 Experimental Data Sets
Young et al. (1990) conducted experiments using Colman and Thews (1988)
design. The configuration that the authors used consisted of an involute inlet of area
equal to 197 mm
2
. Referring to design IV of Figure 4.1, the following table gives the
details of the geometrical parameters.
Table 4-1 Geometrical Parameters, Young et al. (1990)
Case Design Dc(mm)
1

2
D
2
L
2
Ds Ls
25 IV 35 10 0.75 0.73Dc 26Dc 2Dc 2Dc
79
The source of Case 26 is from the specification of a commercial 2 inch
hydrocyclone called MQ HYDRO-SWIRL manufactured by MPE/NATCO. This is the
only experimental data collected for a design different from Colman and Thews and is
the only case in which the model is evaluated against a LHC with a semi-angle greater
than 0.75. This 3 semi-angle LHC was designed by Young et al. (1993) and the
differences with Colman and Thews (1988) design is explained in section 1.3 LHC
Geometry.
The last experimental data set used for the pressure drop is from Colman and
Thew (1983). Cases 27 to 35 are defined with the geometry of Table 4-1 and the
experimental results are provided in Table 4-1. In this study the effect of the viscosity of
water on the pressure drop was studied by varying the temperature.
4.5.2 Results
The pressure drop comparison between the data and model predictions are given
in Figure 4.29, Figure 4.30 and Table 4-7.
0
40
80
120
160
0 40 80 120
Pressure Drop (psi)
F
l
o
w
r
a
t
e

(
l
p
m
)
Experimental Data
LHC Mechanistic Model
Figure 4.1 Pressure Drop Prediction Case 25
80
0
50
100
150
200
250
300
0 50 100 150 200 250
Pressure Drop (psi)
F
l
o
w
r
a
t
e

(
l
p
m
)
Experimental Data
LHC Mechanistic Model
Figure 4.2 Pressure Drop Prediction Case 26
Table 4-1 Pressure Drop Cases 27 to 35
Case Dc (mm)
Flowrate
(lpm)
Water
Density
(g/cc)
Water
Viscosity
(cp)
T (C)
Experimental
Pressure Drop
(psi)
LHC Mechanistc Model
Pressure Drop (Psi)
27 58 250 1.00 1.23 12 32.63 31.37
28 58 250 1.00 1.27 11 32.49 31.08
29 58 250 0.99 0.55 50 36.98 38.87
30 58 220 1.05 1.32 14 35.24 24.40
31 58 220 1.00 1.14 15 27.56 23.93
32 30 60 1.00 1.10 16 19.14 20.86
33 30 40 1.00 1.10 16 7.83 8.17
34 30 60 0.99 0.55 50 26.83 25.42
35 30 50 0.99 0.55 50 17.55 16.72
The model is now evaluated with the results of Cases 25 to 35 in Figure 4.3.
81
0
30
60
90
120
150
180
210
0 30 60 90 120 150 180 210
Experimental Pressure Drop (psi)
L
H
C

M
e
c
h
a
n
i
s
t
i
c

M
o
d
e
l

P
r
e
s
s
u
r
e

D
r
o
p

(
p
s
i
)
Average Relative Error (%) = -7.9
Average Relative Absolute Error (%)= 11.1
Figure 4.3 Comparison Between Pressure Drop Model and All Experimental Data
4.5.3 Discussion
Satisfactory results for pressure drop prediction were achieved by the model for
the Colman and Thews Design (Case 25) and in Youngs Design (Case 26).
Figure 4.3 shows the evaluation of the model for all the data gathered. The
average relative absolute error equal to 11.1% and average relative error equal to 7.9%
are evidences of good performance of the model.
The pressure drop model exhibits correct sensitivity to the fluid properties. Table
4-1 shows that the experimental data and the model predictions have the same trend with
respect to the variation of the water viscosity.
82
CHAPTER V
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
5.1 Summary and Conclusions
A simple mechanistic model is developed for the LHC. The model is capable of
predicting the hydrodynamic flow field of the continuous phase within the LHC. The
separation efficiency is determined based on droplet trajectories, and the inlet-underflow
pressure drop is predicted using an energy balance analysis. A user friendly computer
code is developed based on the proposed model. The code provides easy access to the
input data and very fast output, and can be used for the design of LHC by the industry.
The prediction of the proposed model are compared with elaborated published
experimental data sets. Good agreement is obtained between the model predictions and
the experimental data with respect to both separation efficiency and pressure drop.
A summary of the tasks performed during this study and the most important
conclusions are described as follows.
A set of correlations are developed to predict the hydrodynamic flow
behavior of the continuous phase in the LHC. The swirl intensity, which is
the ratio of tangential momentum flux to the average axial momentum
flux, can be predicted using a modified form of Mantilla (1998) model,
incorporating the semi-angle of a conical section and adjusting the inlet
factor for LHC geometry. Good agreement with experimental data is
83
observed for a small angle range, from 0 to 0.75. These are the range of
values used in the Colman and Thew (1988) Design.
It is confirmed that the swirl intensity defines the velocity field within the
LHC. The tangential velocity exhibited a forced vortex near the axis and
free-like vortex in the outer region. This behavior and the order of
magnitude are well predicted by the model, utilizing some of the
parameters of the Rankine Vortex Equation used by Mantilla (1998) and
Algifri et al (1988).
The axial velocity, which shows a reverse flow in the core region, is
predicted by a third order polynomial equation, as suggested by Mantilla
(1998). A modification of the relationship between the reverse flow radius
and the swirl intensity is proposed. The prediction of the downward flow
by the model is excellent as opposed to the reverse upward flow. No
attempt to correct this is done in this study mainly because the critical
parameter considered by the model is only the downward flow, where the
separation is achieved.
The radial velocity is predicted using the continuity equation and wall
conditions suggested by Wolbert et al. (1995). There are no data available
for this velocity component.
A droplet trajectory analysis is developed assuming local momentum
equilibrium. The only forces acting on the droplet are the centripetal and
drag forces in the radial direction. For simplification it is assumed that the
droplet moves at the fluid velocity in the axial and tangential directions.
84
Based on the droplet trajectory the separation efficiency of the LHC is
determined using a similar procedure proposed by Wolbert et al. (1995).
The underflow purity can be computed for a given feed droplet size
distribution.
Through comparison with 17 cases, where the characteristic diameter of
the hydrocyclone, Dc, varies from 20 to 58 mm and the flowrate ranges
from 32 to 250 lpm, the model predicts the underflow purity with an
average relative absolute error of 4%. One of these cases is the study by
Wolbert et al. (1995), where their model predicted 90% of underflow
efficiency, while the experimental results reported 81%. However, the
proposed LHC model predicted 76% underflow efficiency for this same
case. This may suggest that in general the present model predicts a more
realistic velocity field within the LHC.
Based on the velocity field of the continuous phase and using an energy
balance equation, the pressure drop is predicted by the model. Comparison
with 20 data points reveals an average relative absolute error of 11.1% and
an average relative error equal to 7.9%. The pressure drop is compared
not only with the Colman and Thews Design but also with the Youngs
(1993) Design, and good results are also achieved for the latter. It is
important to mention that Youngs design has a conical section of a 3
semi-angle, what goes beyond the range for which the velocity correlation
was developed.
85
After a critical analysis of several experimental data available for the LHC, it is
possible to conclude that the LHC mechanistic model predicts with a good confidence
level the performance of liquid hydrocyclones with geometrical proportions similar to the
Colman and Thews (1988) Design.
5.2 Recommendations
The developed mechanistic model has proven to be a good tool to predict the
performance of various LHC sizes, for Colman and Thews Design. Unfortunately, most
of the experimental data published to date comes from this LHC Design. In order to use
the current model as a design tool, further comparisons with experimental data from
different designs are needed.
Some recommendations that may improve the performance of the model and help
to understand the limitations of its application are as follows.
Acquire local velocity measurements for the axial and tangential velocity
distribution at different tapered section angles, from 0 to 10 semi-angle
section. These data can be used to improve the set of correlation that
defines the LHC flow field.
The axial velocity profile needs to be further investigated, since under
high values of swirl intensity double reversal may occur, for which the
equation that the model uses will no longer be valid.
The model assumes a stable core. However, vortex instability may occur
under certain conditions, as confirmed by Weispfenning and Petty (1991).
They found that this phenomena is strongly dependent on the swirl
86
intensity and a characteristic Reynolds Number. Knowledge of the swirl
intensity values where these undesirable conditions occur will provide a
realistic range of applicability of the model.
This proposed model does not consider recirculation zones or short
circuits at the inlet. These two phenomena cause either the return to the
main flow of some of the fluid that goes with the oil core to the overflow
outlet, or cause the feed to go directly to the reject orifice. These
conditions may affect to some degree the separation efficiency, and they
have to be included in order to have a more robust model.
The model does not consider the overflow to underflow split ratio. This
parameter is crucial for a desirable operation of the LHC but does not
affect considerably the LHC flow field. At this point the model assumes
that the split ratio is sufficient to accommodate the volume of oil that is
separated and that the efficiency does not change with the split ratio, as
many researchers have reported. This assumption may be true in the
typical range of operation of the LHC, namely, 1 to 10 %, but outside this
range a change of the velocity field may occur, and that must be accounted
for.
There is a relationship between the swirl intensity and the reverse flow
radius. As shown by the experimental data and followed by the models
prediction, the reverse flow radius is reduced as the swirl decays. But there
is a point where there is no longer reverse flow and still some swirling
motion can occur. Under this condition the model will still consider a
87
reverse flow. A proper improvement will be to know for which small swirl
intensity values the flow will not exhibit reverse flow and incorporate this
aspect in the model.
Finally, to use this model as a design tool, a good prediction of the swirl intensity
with the axial position for different taper sections is crucial. It is believed that in the small
angle tapered section the swirl intensity decreases at a slower rate as compared to a
cylindrical section. Nevertheless, a point can be reached at the conical section where
lower values of swirl intensity are generated, as illustrated in the next hypothetical
diagram.
Figure 5.1 Hypothetical Swirl Intensity Decay
At this point, not sufficient information is available to confirm this notion, but it is
important to note that the swirl intensity is crucial for the LHC performance and also for
design purposes. If a model is able to predict accurately the swirl intensity, this can be
used as a design parameter, where an optimum design will be the one with the highest
possible swirl intensity.
88
NOMENCLATURE
A = cross sectional area
B = factor that determines the peak tangential velocity radius
C
D
= drag coefficient
d = droplet diameter
D = diameter
Dc = LHC characteristic diameter
f = friction factor
h = losses
I = inlet factor
k = concentration
L = length
m& = mass flow rate
M
t
= momentum flux at the inlet slot
M
T
= axial momentum flux at the characteristic diameter position
n = centrifugal force correction factor
P = pressure
Q = volumetric flow rate
r = radial position
R = LHC radius
89
Re = Reynolds Number
t = time
T
m
= maximum momentum of the tangential velocity at the section
u = continuous phase local axial velocity
U = bulk axial velocity
v = continuous phase local radial velocity
V = volumetric fraction / velocity
V
r
= droplet radial velocity
V
sr
= droplet slip velocity in the radial direction
V
z
= droplet axial velocity
w = continuous phase local tangential velocity
W = mean tangential velocity
Greek Letters:
= swirl intensity
= taper section semi-angle
= efficiency / purity / pipe roughness
= axis horizontal angle
= viscosity
= density
90
Subscripts:
av = average
c = characteristic diameter location / continuous phase
cf = centrifugal
crit = critical
d = dispersed phase / droplet
f = frictional
g = gravity acceleration
i = inlet
is = inlet section
o = overflow
r = resultant
rev = reverse flow
u = underflow
z = axial position
Abbreviations:
CFD = Computational Fluid Dynamics
ESP = Electric Submergible Pump
LDA = Laser Doppler Anemometry
LHC = Liquid Hydrocyclones
91
REFERENCES
Algifri, A., Bhardwaj, R. and Rao, Y., 1988, Turbulence Measurements in Decaying
Swirl Flow in a Pipe. Applied Scientific Research, Vol. 45, pp. 233-250.
Ali, S., Wesson, G., Petty, C. and Parks, M., 1994, The Use of Small Hydrocyclones for
Produced Water Clarification. Hydrocyclone Development Consortium, Michigan State
University.
Bloor, M., 1987, "On Axially Symmetric Flow Models for Hydrocyclones". In 3rd
International Conference on Hydrocyclones, Wood, P. (ed), Elsevier, Oxford, England,
pp. 83-89.
Bloor, M. and Ingham, D., 1973, "Theoretical Investigation of the Flow in a Conical
Hydrocyclone". Trans. Instn. Chem. Engrs., Vol. 51, pp. 36-41.
Bloor, M. and Ingham, D., 1984, "The Influence of Vorticity on the Efficiency of the
Hydrocyclone". In 2nd International Conference on Hydrocyclones, BHRA, Bath,
England, paper B2, pp. 19-21.
Bloor, M. and Ingham, D., 1987, "The Flow in Industrial Cyclones". Journal of Fluid
Mechanics, Vol. 178, pp. 507-519.
Bloor, M., Ingham, D. and Laverack, S., 1980, "An Analysis of Boundary Layer Effects
in a Hydrocyclone". In International Conference on Hydrocyclone, Cambridge, England,
paper 5, pp. 49-62.
92
Bowers, B., Lloyd, D., Schrenkel, P. and Matthews, C., 1996, "Downhole Application of
Liquid-Liquid Hydrocyclones". In Hydrocyclones 96, Claxton, D., Svarovsky, L. and
Thew, M. (eds), M.E.P., London, England, pp 423-434.
Bradley, D., 1965, "The Hydrocyclone". Pergamon Press.
Bradley, D. and Pulling, D., 1959, "Flow Patterns in the Hydraulic Cyclone and Their
Interpretation in Terms of Performance". Trans. Instn. Chem. Engrs., Vol. 37, pp. 35-44.
Chakraborti, N. and Miller, J., 1992, "Fluid Flow in Hydrocyclones: A Critical Review".
Mineral Processing and Extractive Metallurgy Review, Vol. 11, pp. 211-244.
Chang, F and Dhir, V., 1994, Turbulent flow field in tangentially injected swirl flows in
tubes. International Journal of Heat and Fluid Flow, Vol. 15, pp. 346-356.
Colman, D., 1981, The Hydrocyclone for Separating Light Dispersions. Ph.D. Thesis.
University of Southampton, England.
Colman, D. and Thew, M., 1980, "Hydrocyclone to Give a Highly Concentrated Sample
of a Lighter Dispersed Phase". In International Conference on Hydrocyclones, BHRA,
Cambridge, United Kingdom, paper 15, pp. 209-223.
Colman, D. and Thew, M., 1983, "Correlation of Separation Results From Light
Dispersion Hydrocyclones". Chem. Eng. Res. Des., Vol. 61, pp. 233-240.
Colman, D. and Thew, M., 1988, "Cyclone Separator". U.S. Patent 4764287.
Colman, D., Thew, M. and Corney, D., 1980, "Hydrocyclones for Oil/Water Separation".
In International Conference on Hydrocyclones, BHRA, Cambridge, United Kingdom,
paper 11, pp. 143-165.
93
Colman D., Thew, M. and Lloyd, D., 1984, "The Concept of Hydrocyclones for
Separating Light Dispersions and a Comparison of Field Data with Laboratory Work". In
2nd International Conference on Hydrocyclones, BHRA, Bath, England, paper F2, pp.
217-232.
Dabir, B., 1983, "Mean Velocity Measurements in a 3''-Hydrocyclone Using Laser
Doppler Anemometry". Ph.D. Thesis. Michigan State University, Michigan.
Fanglu, G. and Wenzhen, L., 1987, "Measurements and Study of Velocity Field in
Various Cyclones by Use of Laser Doppler Anemometry". In 3rd International
Conference on Hydrocyclones, Wood, P. (ed), Elsevier, Oxford, England, pp. 65-74.
Fraser, S. and Abdullah, M., 1995, "LDA Measurement on a Modified Cyclone". ASME
Laser Anemometry, FED-Vol. 229, pp. 395-403.
Flanigan, D., Stolhand, J., Shimoda, E. and Skilbeck, F., 1992, "Use of Low-Shear
Pumps and Hydrocyclones for Improved Performance in the Cleanup of Low-Pressure
Water". SPE Production Engineering, August, pp. 295-300.
Hall, N., 1957, Thermodynamics of Fluid Flow. Longmans, Green, New York.
Hargreaves, J., 1990, Computing and Measuring the Flowfield in a Deoiling
Hydrocyclone. Ph.D. Thesis. University of Southampton, England.
Hargreaves, J. and Silvester, R., 1990, Computational Fluid Dynamic Applied to the
Analysis of Deoiling Hydrocyclone Performance. Trans. IchemE., Vol. 68, Part A, pp.
365-383.
Hashmi, K., Friesen, W., Bohun, D. and Thew, M., 1996, "Application of Hydrocyclones
for Treating Produced Fluids in Heavy Oil Recovery". In Hydrocyclones 96, Claxton, D.,
Svarovsky, L. and Thew, M. (eds), M.E.P., London, England, pp 369-382.
94
He, P., Salcudean, M., Branion, R. and Gartshore, I., 1997, "Mathematical Modeling of
Hydrocyclones". In ASME Fluids Engineering Division Summer Meeting, FEDSM97-
3315.
Hsieh, K. and Rajamani, R., 1991, "Mathematical Model of the Hydrocyclone Based on
Physics of Fluid Flow". AIChE Journal, Vol. 37, No. 5, pp 735-746.
Hubred, G., Mason, A., Parks, S. and Petty, C., 2000, "Dispersed Phase Separations: Can
CFD Help?". Proceeding of ETCE/OMAE Conference, New Orleans, Louisiana.
Jirun, X., Qian, L. and Qui, J., 1990, "Studying the Flow Field in a Hydrocyclone With
no Forced Vortex I, II". Filtration and Separation, July/August, pp. 276-278,
September/October, pp. 356-359.
Kang, J., 1984, "On the Analysis of the Flow and Particle Motion in a Hydrocyclone".
Society of Petroleum Engineers, paper 13407.
Kang, J. and Hayatdavoudi, A., 1985, "Analytical Model Determining the Flow Pattern
and the Distribution of d50 in a Hydrocyclone". Society of Petroleum Engineers, paper
15202.
Kelsall, D., 1952, "A Study of the Motion of Solid Particles in a Hydraulic Cyclone".
Trans. Instn. Chem. Engrs., Vol. 30, pp. 87-108.
Kraipech, W., Chen, W. and Parma, F., 2000, "Prediction of Hydrocyclone Performances
- How Much Can the Models Do?". American Filtration & Separation Society Annual
Conference, Myrtle Beach, SC, March 14-17.
Mantilla, I., 1998, Bubble Trajectory Analysis in Gas-Liquid Cylindrical Cyclone
Separators. M.Sc. Thesis. The University of Tulsa.
95
Meldrum, N., 1988, "Hydrocyclones: A Solution to Produced-Water Tratment". SPE
Production Engineering, November, pp. 669-676.
Moraes, C., Hackenberg, C., Russo, C. and Medronho, R., 1996, "Theoretical Analysis of
Oily Water Hydrocyclones". In Hydrocyclones 96, Claxton, D., Svarovsky, L. and Thew,
M. (eds), M.E.P., London, England, pp 383-398.
Morsi, S. and Alexander, A., 1972, An Investigation of Particle Trajectories in Two-
Phase Flow Systems. Journal of Fluid Mechanics, Vol. 55, part 2, pp. 193-208.
Munson, B., Young, D. and Okiishi, T., 1994, Fundamentals of Fluid Mechanics. John
Wiley & Sons, Inc. Second Edition.
Nezhati, K. and Thew, M., 1987, Aspects of the Performance and Scaling of
Hydrocyclones for Use with Light Dispersions. In 3rd International Conference on
Hydrocyclones, Wood, P. (ed), Elsevier, Oxford, England, pp. 167-180.
Parks, S. and Petty, C., 1995, Prediction of Angular Momentum Distribution in Deoiling
Hydrocyclones Using a Constant Eddy Viscosity Model. Technical Report No. HDC-
R2. Hydrocyclone Development Consortium, Michigan State University.
Rajamani, K. and Devulapalli, B., 1994, "Hydrodynamic Modeling of Swirling Flow and
Particles Classification in Large-Scale Hydrocyclones". KONA Powder and Particle, No.
12, pp. 95-104.
Rajamani, K. and Hsieh, K., 1988, "Hydrocyclone Model: A Fluid Mechanic Approach".
In Society of Mineral Engineers Annual Meeting, Phoenix, Arizona, preprint # 88-163.
Rhodes, N., Pericleous, K. and Drake, S., 1987, "The Prediction of Hydrocyclone
Performance with a Mathematical Model". In 3rd International Conference on
Hydrocyclones, Wood, P. (ed), Elsevier, Oxford, England, pp. 51-58.
96
Rietema, K., 1961, "Performance and Design of Hydrocyclones - I,II,III,IV". Chemical
Engineering Science, Vol. 15, pp. 298-320.
Seyda, B. and Petty, C., 1991, "Separation of a Light Dispersion in a Cylindrical Vortex
Chamber". Technical Report No. HDC-R6. Hydrocyclone Development Consortium,
Michigan State University.
Sinker, A. and Thew, M., 1996, Drop Size Distribution in De-watering Type
Hydrocyclones. In Hydrocyclones 96, Claxton, D., Svarovsky, L. and Thew, M. (eds),
M.E.P., London, England, pp 345-356.
Smyth, I., 1988, Cyclonic Dewatering of Oil. Ph.D. Thesis. University of Southampton,
England.
Smyth, I. and Thew, M., 1987, A Comparison of the Separation of Heavy Particles and
Droplets in a Hydrocyclone. In 3rd International Conference on Hydrocyclones, Wood,
P. (ed), Elsevier, Oxford, England, pp. 193-204.
Smyth, I. and Thew, M., 1996, A Study of the Effect of Dissolved Gas on the Operation
of Liquid-Liquid Hydrocyclones. In Hydrocyclones 96, Claxton, D., Svarovsky, L. and
Thew, M. (eds), M.E.P., London, England, pp 357-368.
Smyth, I., Thew, M. and Colman, D., 1984, The Effect of Split Ratio on Heavy
Dispersion Liquid-Liquid Separation in Hydrocyclones. In 2nd International Conference
on Hydrocyclones, BHRA, Bath, England, paper E2, pp. 177-190.
Smyth, I., Thew, M., Debenham, P. and Colman, D., 1980, Small-Scale Experiments on
Hydrocyclones for De-watering Light Oils. In International Conference on
Hydrocyclone, Cambridge, England, paper 14, pp. 189-208.
97
Stroder, S. and Wolfenberger, E., 1994, "Hydrocyclone Separation: A Preferred Means of
Water Separation and Handling in Oilfield Production". Society of Petroleum Engineers,
paper 27671.
Svarovsky, L., 1984, "Hydrocyclones". Holt, Rinehart & Winston.
Svarovsky, L., 1996, "A Critical Review of Hydrocyclones Models". In Hydrocyclones
96, Claxton, D., Svarovsky, L. and Thew, M. (eds), M.E.P., London, England, pp 17-30.
Taitel, Y., 1995, Advances in Two Phase Flow Mechanistic Modeling. Society of
Petroleum Engineers, paper 27959.
Thew, M., 1986, "Hydrocyclone Redesign for Liquid-Liquid Separation". The Chemical
Engineer, July/August, pp. 17-21.
Thew, M., Silk, S. and Colman, D., 1980, "Determination and Use of Residence Time
Distribution for Two Hydrocyclones". In International Conference on Hydrocyclones,
BHRA, Cambridge, United Kingdom, paper 16, pp. 225-248.
Thew, M. and Smyth, I., 1997, "Development and Performance of Oil-Water
Hydrocyclone Separators - A Review". In Innovation in Physical Separation
Technologies Conference, Institution of Mining and Metallurgy, Falmouth, United
Kingdom, pp. 77-89.
Thew, M., Wright, C. and Colman, D., 1984, "R.T.D. Characteristics of Hydrocyclones
for the Separation of Light Dispersions". In 2nd International Conference on
Hydrocyclones, BHRA, Bath, England, paper E1, pp. 163-176.
Weispfennig, K., Parks, S. and Petty, C., 1995, An Experimental Study of Internal Flow
Structures of a Deoiling Hydrocyclone. Technical Report No. HDC-R3. Hydrocyclone
Development Consortium, Michigan State University.
98
Weispfennig, K. and Petty, C., 1991, "Flow Visualization in a Confined Vortex Flow".
Technical Report No. HDC-R5. Hydrocyclone Development Consortium, Michigan State
University.
Wolbert, D., Ma, B. and Aurelle, Y., 1995, "Efficiency Estimation of Liquid-Liquid
Hydrocyclones Using Trajectories Analysis". AIChE Journal, Vol. 41, No. 6, pp 1395-
1402.
Young, G., 1993, Desalting Crude Oil Using Hydrocyclone-Centrifuge Arrangement.
Produced in Amoco Production Company Research Department Tulsa Production
Research Division.
Young, G., Taggart, D. and Hild, D., 1993, Improved Understanding of Deoiling
Hydrocyclones Leads to Significant Performance Improvement. Produced in Amoco
Production Company Research Department Tulsa Production Research Division.
Young, G., Walkley, W., Taggart, D., Andrews, S. and Worrel, J., 1990, Oil-Water
Separation Using Hydrocyclones: An Experimental Search for Optimum Dimensions.
American Filtration Society, Advanced in Filtration and Separation Technology, Vol.3,
Conference held in Baton Rouge, Louisiana.