Sugar Research Australia Ltd.

eLibrary http://elibrary.sugarresearch.com.au/
Completed projects final reports Milling Efficiency and Technology

2017-05

Improved modelling of wet scrubbers

Mann, A

http://hdl.handle.net/11079/16424
Downloaded from Sugar Research Australia Ltd eLibrary
 Improved modelling of wet scrubbers

(2012/055)

Final Report submitted to Sugar Research Australia

Anthony Mann

Queensland University of Technology

May 2017

Page 1 Sugar Research Australia- Research Funding Unit- Research Project Final Report
© Sugar Research Australia Limited 2017

Copyright in this document is owned by Sugar Research Australia Limited (SRA) or by one
or more other parties which have provided it to SRA, as indicated in the document. With the
exception of any material protected by a trade mark, this document is licensed under a
Creative Commons Attribution-NonCommercial 4.0 International licence (as described
through this link). Any use of this publication, other than as authorised under this licence or
copyright law, is prohibited.

http://creativecommons.org/licenses/by-nc/4.0/legalcode - This link takes you to the relevant

licence conditions, including the full legal code.

In referencing this document, please use the citation identified in the document.

Disclaimer:

In this disclaimer a reference to “SRA” means Sugar Research Australia Ltd and its
directors, officers, employees, contractors and agents.

This document has been prepared in good faith by the organisation or individual
named in the document on the basis of information available to them at the date of
publication without any independent verification. Although SRA does its best to
present information that is correct and accurate, to the full extent permitted by law
SRA makes no warranties, guarantees or representations about the suitability,
reliability, currency or accuracy of the information in this document, for any
purposes.

The information contained in this document (including tests, inspections and

recommendations) is produced for general information only. It is not intended as professional
advice on any particular matter. No person should act or fail to act on the basis of any
information contained in this document without first conducting independent inquiries and
obtaining specific and independent professional advice as appropriate.

To the full extent permitted by law, SRA expressly disclaims all and any liability to any
persons in respect of anything done by any such person in reliance (whether in whole or in
part) on any information contained in this document, including any loss, damage, cost or
expense incurred by any such persons as a result of the use of, or reliance on, any
information in this document.

The views expressed in this publication are not necessarily those of SRA.

Any copies made of this document or any part of it must incorporate this disclaimer.

Page 2 Sugar Research Australia- Research Funding Unit- Research Project Final Report
SRA Project Code 2012/055 (QUT055)

Project Title Improved modelling of wet scrubbers

Key Focus Area in SRA

Milling efficiency and technology (KFA5)
Strategic Plan

Research Organisation(s) Queensland University of Technology

Dr Anthony Mann
Chief Investigator(s) Dr Floren Plaza
Dr Phil Hobson
The project seeks to gain an improved understanding of the flow processes in
wet scrubbers. These flow processes include but are not limited to:
 Flue gas and droplet flows;
 Agglomeration of droplets;
Project Objectives  Flow of bulk water; and
 Interaction between dust particles and water droplets.
With the improved modelling capability, existing scrubber designs can be
simulated with the aim of determining design modifications for improved
scrubber performance.
SRA measures of success
for Key Focus Area (from Adoption of improved or novel milling processes and technology
SRA Strategic Plan)

Page 3 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Contents

Executive Summary .......................................................................................................5

Background ........................................................................................................................6

Project Objectives ....................................................................................................... 11

Methodology ................................................................................................................... 12

Results ............................................................................................................................... 18

Discussion ....................................................................................................................... 39

Acknowledgements .................................................................................................... 40

References ....................................................................................................................... 40

Outputs and Outcomes ............................................................................................ 42

Intellectual Property (IP) and Confidentiality .............................................. 43

Industry Communication and Adoption of Outputs ............................... 43

Environmental Impact ............................................................................................... 44

Recommendations and Future Industry Needs .......................................... 44

Publications..................................................................................................................... 44

APPENDIX.......................................................................................................................... 45

Page 4 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Executive Summary

Recent record high levels of electricity export from sugar factories have corresponded to
record low levels of dust emissions from sugar factory boilers. One of the main reasons
for low emissions is the increased use of wet scrubbers for removing dust from boiler
flue gas.

While wet scrubbers have high collection efficiency, operational problems associated
with many wet scrubber installations reduce boiler steam output and, therefore, factory
crushing rates. These problems include:
 Blockages that increase the pressure drop across the scrubbers. This reduces the
gas flow through the boiler and, therefore, boiler steam output.
 Droplet carryover that causes build-up on and wear of boiler induced draft fans,
reduces collection efficiency and generates complaints from the local community.

The need to improve the design of centrifugal wet scrubbers provided the motivation for
this work which uses computational fluid dynamics (CFD) modelling and scale model
experiments. The modelling and experiments were used to develop an improved
understanding of the flow processes occurring in a centrifugal wet scrubber and to
evaluate options for improving their performance. This project falls within the SRA key
focus area of milling efficiency and technology.

A test rig with a scale model of a centrifugal wet scrubber module was constructed and
used for velocity and pressure measurements and flow visualisation. The flow
visualisation, with standard and high speed photography, and the measurements were
used for CFD model validation and to gain a better understanding of the air and water
flow patterns in the scrubber scale model. The flow visualisation identified the stages in
the process water sheet break-up into droplets and the development of a swirling water
bath that forms between the scrubbing and demisting vanes. The behaviour of this
water bath affects droplet carryover and scrubbing efficiency. At very high air flows the
water bath would extend beyond the demister vanes and lead to increased droplet
carryover. At low air flow the water bath would not form and this would lead to reduce
collection efficiency. Both these observations are consistent with the reported
performance of full scale wet centrifugal scrubbers with high and low gas flows.

A site visit to two factories with centrifugal wet scrubbers was completed and the
patterns of build-up in the scrubber modules were related to the predicted flow patterns
in the scrubber scale model.

Observations from the scrubber scale model flow visualisation experiments were used in
the development of a sub-model for dust particle capture by water droplets and water
films. With this sub-model incorporated into the CFD code, the collection efficiency of a
centrifugal wet scrubber can be predicted.

Observations from the scrubber scale model experiments and associated CFD modelling
formed the basis of proposed design changes. Further modelling and modifications to
the scrubber scale model were made to evaluate the proposed design changes. Feedback

Page 5 Sugar Research Australia- Research Funding Unit- Research Project Final Report
 was received from factory staff (end users) and the scrubber manufacturer on the
proposed design changes and further modelling was carried out based on this feedback.
This lead to updated scrubber design change recommendations.

Outputs from this project include the test rig and scrubber scale model, the improved
modelling capability with the updated code, the CFD modelling and flow visualisation
results and the recommended scrubber design changes.

Implementation of the recommended scrubber design changes should lead to reduced

factory downtime, increased factory throughput, reduced emissions from sugar factory
boilers and reduced complaints from the local community.

Some of the recommended design changes have already been implemented and it is
likely that more widespread implementation of the design changes will occur.

Background

Wet scrubbers are the Australian sugar industry's most effective means of removing dust
from boiler flue gas used in the Australian sugar industry. Currently wet scrubbers are
used on more than 60% of the boilers in the industry but this proportion is increasing as
new boilers with wet scrubbers installed replace older boilers that often use dry
scrubbers. The more widespread use of wet scrubbers is significantly reducing the
environmental footprint of the sugar industry.

In virtually all cases, the collection efficiency of wet scrubbers is high enough for boilers
fitted with them to comfortably comply with the dust emission limits set by
environmental authorities. However wet scrubbers often cause boiler operational
problems such as excessive vibration, deposit build up on and wear of boiler induced
draft fans due to water droplet carryover and reduced boiler steam output due to
blockages in and/or around the scrubber. The profitability of a sugar milling operation
is adversely affected by the reduced throughput and/or factory stoppages caused by
boilers being taken off line for scrubber cleaning.

Wet scrubbers are effective and in many cases the most economically viable means of
removing particulate and gaseous pollutants from gas streams (Calvert et al., 1974; Lee
et al., 2013). They remove particles from a gas stream via the mechanisms of impaction,
interception and diffusion (Schifftner and Hesketh, 1983). In most wet scrubbers,
impaction, where the kinetic energy of the particle is used to penetrate the surface
tension of the scrubbing liquid, is the dominant means of particle removal. Interception
occurs when particles meet the droplets at angles much less than 90° and at lower
relative velocity than that required for impaction. Diffusion, where particles migrate and
come into contact with droplets due to gas density and turbulence fluctuations, is only
important for very small particles (< 5 µm).

Several designs of wet scrubbers are used to control emissions from sugar factory
boilers around the world. The most common designs include venturi scrubbers (fixed or
variable throat) (Jones, 1949), sieve plate (flooded and impingement) scrubbers, fixed

Page 6 Sugar Research Australia- Research Funding Unit- Research Project Final Report
vane centrifugal scrubbers, spray tower scrubbers and static bath impingement
scrubbers (Lora and Jativa, 1999; Moor, 2007; Pennington, 1999). In the Australian
sugar industry only the fixed vane Ducon and Clyde Carruthers style and the multi-
venturi design wet scrubbers are currently used as the primary means of dust collection.
Sketches of these scrubber designs are shown in figures 1, 2, and 3 respectively. Most of
the wet scrubbers in the Australian sugar industry are the fixed vane Ducon style
(figure 1).

Figure 1 Ducon style fixed vane scrubber.

Page 7 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 2 Clyde Carruthers style fixed vane scrubber.

Figure 3 Multi-venturi fixed vane scrubber (www.mikropul.com – accessed on the 6-6-2008).

Page 8 Sugar Research Australia- Research Funding Unit- Research Project Final Report
The design of wet scrubbers has traditionally been based on past experience and rules of
thumb. In some cases CFD has been used in the design process but this modelling
usually only considers the gas flow patterns only. Modelling carried out by QUT in the
last few years is more sophisticated and includes the motion of the water droplets and
evaporation, but more development work is required to account for effects such as the
flow of bulk water, the agglomeration of water droplets into bulk water and the
interaction between dust particles and bulk water or water droplets.

The flows in a wet scrubber are very complex due to the interaction between dust
particles, water droplets, bulk water and flue gas. The flue gas flow is turbulent and
there are multiple phases (gas, liquid and solid).

CFD codes that predict gas flow patterns are based on conservation equations for mass,
momentum and energy (Bird et al., 1960; Roache, 1976). Most of the major CFD codes in
use are based on the TEACH set of codes developed at Imperial College (Patankar and
Spalding, 1972).

The modelling becomes significantly more complex when more than one phase is
present. Two phase modelling, which considers a gas phase and a dispersed phase
(usually liquid or solid) has been the focus of much modelling effort (Drew, 1983; Hinze,
1972; Melville and Bray, 1979). Most techniques used for solving two phase flow
problems can be categorised as either Eulerian or Lagrangian.

An Eulerian approach (Harlow and Amsden, 1975; Whitaker, 1973), often called a 'two
fluid' model, treats the cloud of particles as a continuum. Conservation equations can be
formulated for the particle phase in the same manner as for the gas phase. These
conservation equations will include the volume concentrations (void fractions) of each
phase. To do this an effective diffusion coefficient for the particle phase must be
assumed. This has been an area of much study. Coupling between the two phases is
modelled by source terms in the respective conservation equations. A significant
advantage of Eulerian methods is that numerical techniques developed for solving the
gas phase flow can be applied to the particle phase. However, storage requirements for
Eulerian schemes can become quite large when a number of representative particle size
classes are used because each size class usually requires a separate conservation
equation. As gas phase numerical techniques are used for the particle phase, the solution
for the particle phase will be affected by numerical instabilities and numerical diffusion.
In addition, modifications to the particle effective diffusion coefficient must be made at
the wall and the particle slip velocity at the wall must be established.

A Lagrangian approach involves following the motion of a representative number of

particles by solving an equation of motion for each particle. It was proposed (Migdal and
Agosta, 1967) that the particles could be regarded as a source of mass, energy and
momentum to the continuum phase (two way coupling). This was later formalised by
developing the PSI-Cell (Particle-Source-In Cell) method (Crowe et al., 1977) that has
since become widely used. This approach ignores particle to particle interactions and
thus cannot be used for very high concentrations of the particle phase. Simulating the
effect of turbulent dispersion can be a problem with this method. One option used a

Page 9 Sugar Research Australia- Research Funding Unit- Research Project Final Report
diffusion 'force' (Dukowicz, 1980) requiring the selection of a diffusion coefficient.
Others (Leonard et al., 1980) have also treated turbulent dispersion as a Fickian
diffusion process while modelling the flow in an electrostatic precipitator. Another
method for modelling turbulent dispersion is the Monte Carlo technique (Gosman and
Ioannides, 1981). The turbulent gas flow field is represented as steady flow plus a
random velocity fluctuation. Many particles must be used to simulate the cloud, thereby
increasing simulation run time. Consequently, this method can be computationally
expensive but has the advantage of low storage requirements.

A Lagrangian model of the particle phase allows easier treatment of particle wall
interactions and gives more information on particle trajectories and residence times.
Furthermore, reasonable predictions of particle trajectories are possible even when a
rather coarse flow grid is used. Lagrangian techniques have been used to model the flow
in a cyclone separator (Crowe and Pratt, 1974) and the flow of coal particles in an
axisymmetric furnace (Lockwood et al., 1980). This approach has also been used to
model spray units (Palaszewski et al., 1981).

The relative merits of Eulerian and Lagrangian approaches have been investigated by
many workers (Adeniji-Fashola and Chen, 1990; Crowe, 1982; Durst et al., 1984). In one
case (Durst et al., 1984) both approaches to model two phase vertical pipe flow were
compared. They found that Lagrangian particle tracking was superior when particle
accelerations were high. An Eulerian approach was better suited to cases with high
particle concentrations.

Earlier modelling of a Ducon style wet scrubber carried out at the Sugar Research
Institute (Downing et al., 2000) considered only the gas flow patterns because the
commercial CFD modelling software available at the time was not able to represent a
dispersed phase (dust particles, water droplets) on the mesh required for the model.
The FURNACE code (Boyd, 1986; Dixon et al., 2005; Luo and Stanmore, 1994; Mann,
1996; Woodfield, 2001) is more sophisticated and includes the motion of the water
droplets (with two way coupling between the gas and particle phases) and evaporation.
However this code, while ideally suited for modelling the flow patterns and combustion
in bagasse and coal fired boilers, can only solve on a cartesian structured grid which
makes it difficult to model the internal components of a scrubber. Furthermore, the k-
turbulence model used by the FURNACE code is known to be deficient in strongly
swirling flows (Weber et al., 1990).

CFD has been used by several researchers to predict single and two-phase flow in dry gas
cyclones (Boysan et al., 1982; Chuah et al., 2006; Cortes and Gil, 2007; Griffiths and
Boysan, 1996; Karagoz and Kaya, 2007) and multi-phase flow in a dense medium cyclone
(Narasimha et al., 2007). The earlier work on single phase cyclone flow identified that
the k- turbulence model could not accurately predict the flow patterns. More recent
CFD modelling of a sampling cyclone achieved reasonable agreement between predicted
and measured axial and tangential velocities when using an algebraic turbulence model
for the gas phase (Chuah et al., 2006). This modelling work only had one way coupling
between the gas and particle phases.

Page 10 Sugar Research Australia- Research Funding Unit- Research Project Final Report
 Outlet droplet distributions for wet venturi scrubbers have been predicted with some
success using a two-dimensional CFD model (Ahmadvand and Talaie, 2010). More
complex CFD models of a venturi scrubber that take into account two dispersed phases
(droplet and dust particles) have also been developed (Pak and Chang, 2006).
Lagrangian particle tracking is used to predict the motion of droplets and dust particles.
This model takes into account atomisation of the liquid jet (Wu et al., 1997) and capture
of the dust particles by the droplets (Mohebbi et al., 2003).

Another wet venturi scrubber investigation included the collection efficiency tests by
Costa et al. (2005) who operated a test scale venturi scrubber with mineral salt as the
dispersed phase. Collection efficiencies were measured for different scrubbing liquid
flow rates and air velocities. Experimental results and CFD predictions were compared
by Guerra et al. (2012); this work showed how scrubber pressure drop is affected by the
number of liquid injection orifices. Ali et al. (2015) have measured the collection
efficiency of a venturi scrubber using a filtration technique and compared the results
with a previously developed expression for collection efficiency via inertial impaction
(Calvert, 1970), which is usually the most important mechanism for dust collection in
wet scrubbers (Pak and Chang, 2006).

The studies noted above are just some of the experimental and CFD investigations of dry
cyclones and wet venturi scrubbers. By contrast there are few published studies of wet
centrifugal scrubbers like the Ducon and Clyde Carruthers used in the Australian sugar
industry. This could be because there are few empirical relations, experimental data and
measurements that can form the basis of such an investigation like there are for dry
cyclones and wet venturi scrubbers.

The work carried out in this project, which combines experimental and CFD
investigation into centrifugal wet scrubbers, therefore fills a void in the published
literature.

Project Objectives

The objectives of the project are to gain an improved understanding of the flow
processes in wet scrubbers. These flow processes include but are not limited to:

 Flue gas and droplet flows;

 Agglomeration of droplets;
 Flow of bulk water; and
 Interaction between dust particles and water droplets.

With the improved modelling capability, existing scrubber designs can be simulated with
the aim of determining design modifications for improved scrubber performance.

Page 11 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Methodology

Construction of the test rig

After the initial review of relevant literature and enrolment of the PhD student, a
laboratory scale model of a wet scrubber was constructed. This model, along with the
extraction fan and connecting duct work, form the test rig that was used for flow
visualisation CFD model validation. The test rig, in conjunction with CFD modelling, was
used to evaluate scrubber design modifications. The test rig is located at the QUT pilot
plant facility at Banyo and is shown in figure 4. A close up view of the scrubber scale
model is shown in figure 5.

Figure 4 Test rig for wet scrubber scale model.

Page 12 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 5 Wet scrubber scale model.

The scale model is made of transparent acrylic. Flexible and rigid duct work is used to
connect the scale model to the extraction fan. The small length of flexible duct work was
used to reduce the amount of fan vibration transmitted to the scale model. It was
originally proposed that the fan would be connected to the inlet duct, rather than the
outlet of the scale model. However the as constructed arrangement is superior because
the velocity and pressure measurements are not be affected by fan generated turbulence.
The as constructed arrangement is also closer to the actual arrangement of a wet
scrubber module in a boiler (upstream of the induced draft fan). The extraction fan can
run at full speed without affecting the structural integrity of the scale model or duct
work.

The scrubber scale model is approximately one tenth the size of a full scale scrubber
used to remove dust from boiler flue gas. This meant that to achieve full Reynolds
number similarity with the full scale scrubber the air velocities through the SSM needed
to be higher than 100 m/s, which was clearly impractical. Therefore, in the experiments,
the air flows through the scrubber scale model were adjusted using the variable speed
controller so that the air velocities in the scrubber scale model would approximately
match the gas velocities in the full scale scrubber. This gave Reynolds numbers based on
the air velocities and SSM diameter greater than 100 000 which corresponds to fully
turbulent flow. Velocity and pressure measurements were made using a multi-meter
equipped with a hot wire anemometer and a differential pressure meter.

Page 13 Sugar Research Australia- Research Funding Unit- Research Project Final Report
The test rig was operated with different air flows without and with water addition. To
avoid damaging the hot wire anemometer, velocity measurements were only taken when
the rig was operated with air flow only. On some of the occasions when the rig was
operated with air flow and water addition, high speed cameras (X-Stream XS-4 and
HiSpec 1) were used to analyse the water flow through the scrubber scale model.

Upgrade of the CFD model

ANSYS Fluent is arguably the most powerful CFD package for simulating multi-phase
flows and was chosen for the modelling work in this project. One of the strengths of the
code is its ability to model the dispersed phases using Lagrangian particle tracking or by
solving a separate conservation equation for each of the dispersed phases (the Eulerian
or volume of fluid approach). However the code, while powerful, cannot represent all
flow phenomena occurring in the wide range of applications that the code is used to
simulate. For this reason the code was written to allow the user the option of overriding
or adding to many of the sub models used. This is done with user defined functions.

A review of the relevant literature was carried out to determine the most appropriate
approaches for simulating the various flow processes occurring in a wet scrubber. In some
cases the preferred approach is already available in ANSYS Fluent while for others, user
defined functions are required to better simulate the multiphase flow in a wet scrubber.
One of the important processes occurring in tangential entry centrifugal wet scrubber is
the flow of bulk water, specifically over the water distribution cone and down the inside
wall of the scrubber module. This can already be simulated using the Eulerian approach
available in ANSYS Fluent.

Effective break-up of the water curtains that form over the edge of the water distribution
cone and underneath the scrubbing vanes is essential for good scrubbing efficiency and
for the reduction of water droplet carryover. The thickness of the water curtain can be
estimated by carrying out Eulerian modelling of the water flow over the water
distribution cone and the flow down the outside of the scrubbing vanes. The predicted
water curtain thickness was used as an estimate of the droplet mean particle size and an
assumed initial particle size distribution can be used to determine droplet sizes for
Lagrangian particle tracking in a manner similar to earlier modelling work on venturi
scrubbers (Pak and Chang, 2006).

The population balance model (Hounslow et al., 1988) has been used for bubble columns,
aerosols and atomization of liquid jets. It can predict the break-up of larger droplets into
smaller droplets and the coalescence of smaller droplets into larger droplets. The
population balance model is available as an option in the CFD code ANSYS Fluent.
The break-up of droplets is strongly dependent on the Weber number (We) which is
defined for a droplet in a gas flow as (Pilch and Erdman, 1987):
 V 2d
We 

where  is the density of the gas, V is the droplet velocity relative to the gas, d is the
droplet diameter and σ is the surface tension of the droplet material. The break-up
mechanism is strongly dependent on the Weber number as shown in figure 6.

Page 14 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 6 Droplet break-up mechanisms (Pilch and Erdman, 1987).

Droplet collisions are modelled in ANSYS Fluent with the collision volume approach
(O'Rourke, 1981) which predicts the probability that two particles in the same
computational cell will collide.

In this project the above approach for droplet coalescence has been updated using the
macro DEFINE_DPM_SPRAY_COLLIDE to include a kinetic energy test. This kinetic
energy test takes into account that a collision between a dust particle and a droplet will
only result in droplet capture if the kinetic energy associated with the collision exceeds
the work required for droplet capture defined by Pemberton (1960). The criterion for
dust particle capture by a droplet therefore can be expressed as:
1 2 8𝜋𝑟 2 𝜎
𝑚𝑑𝑢𝑠𝑡 |𝑉̃𝑟𝑒𝑙 | >
2 3
~
where mdust is the mass of the dust particle and Vrel is the velocity of the dust particle
relative to the droplet. Note the above assumes that the dust particle is smaller than the
water droplet. The assumption is justified because preliminary calculations have shown
that the minimum stable droplet diameter for the gas velocities encountered in a wet

Page 15 Sugar Research Australia- Research Funding Unit- Research Project Final Report
scrubber is of the order of 1 mm while nearly all dust particles from a bagasse fired
boiler have diameters less than 0.1 mm.

The above kinetic energy approach was also used to model the capture of dust particles
by the liquid film on the inside wall of a centrifugal wet scrubber.

In a centrifugal wet scrubber virtually all the droplet coalescence would be expected to
occur on or near the inside wall. Droplets that do not coalesce onto the wall are likely to
be carried over from the scrubber and cause build-up of ash on the induced draft fan and
black rain emissions from the stack. It has been found that particles above a certain size
are likely to coalesce onto a surface while smaller particles are unable to remove the air
between the contact regions of the wall film and the droplet. This critical size depends
on a modified version of the Weber number based on the droplet density and collision
velocity (Pan and Law, 2007). This critical collision Weber number is used as a criteria
for the droplet being collected at the scrubber walls and is introduced as a user defined
function in ANSYS Fluent using the Macro DEFINE_DPM_BC. However it has also been
found that if this collision Weber number is too high the droplet hitting the wall may
fragment into smaller droplets (Orme, 1997). These smaller droplets could have
collision Weber numbers too small for coalescence with the film on the wall which
means they will bounce off the wall and possibly exit the scrubber. This is also
incorporated into the user defined function for wall collisions.

The collection of dust particles by free stream droplets and the wall film were taken into
account using the approach described above. It is noted that the droplets that are not
collected by the wall film exit the scrubber with the gas flow as carryover. The dust
particles collected by these droplets are not in fact removed from the gas flow. The code
modifications take into account the reduction in dust collection efficiency due to droplet
carryover.

Wet scrubber inspections

Inspections of wet scrubber modules were carried out at Bingera Mill and Isis Mill. The
inspections included the inside of scrubber inlet duct, underneath the scrubbing vanes,
in between the scrubbing vanes and demisting (eliminator) vanes and above the
demisting vanes. Figure 7(a) shows the build-up of ash below the demisting vanes of a
scrubber module and figure 7(b) shows the build-up above the demisting vanes. Little
build-up was seen around the scrubbing vanes and pressure disc as seen in figure 8(a)
but figure 8(b) shows significant build-up on the roof of the scrubber inlet duct.

The scrubber inspections identified design variations such as the convergence of the
scrubber gas inlet duct, the use of a pressure disc below the scrubbing vanes and a dam
on the water distribution cone. Figure 9(a) shows a water distribution cone without a
dam and figure 9(b) shows a water distribution cone with a dam.

Page 16 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 7 Build-up (a) below and (b) above the demisting vanes.

Figure 8 (a) Scrubbing vanes and pressure disc and (b) deposit on top of scrubber inlet duct.

Figure 9 Water distribution cone (a) without and (b) with a dam.

Page 17 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Results

Test rig scale model measurements and observations

Velocity and pressure measurements were carried on the test rig scale model for
different air volumetric flow rates. The air volumetric flow rates through the scale
model were chosen so that average air velocities in the scale model would be similar to
the average gas velocities in a full scale scrubber at different boiler steam loads. The
measured velocity magnitudes have been used for CFD model validation.

Figures 10 and 11 show measured tangential and axial air velocities in a plane below the
scrubbing vanes for traverses parallel to and perpendicular to the direction of the flow in
the inlet duct with an air volumetric flow rate of 0.2 m3/s. The pressure drop across the
scale model corresponding to these measurements was approximately 380 Pa. Figure 10
shows that the velocity distribution is not symmetrical with higher velocities on the side
of the scale model opposite to the direction of flow in the inlet duct (bottom of the sketch
in figure 10). The minimum velocities are also closer to the side of the scale model
opposite the direction of flow in the inlet duct. Velocity measurements along the
traverse perpendicular to the direction of flow in the inlet duct also show non
symmetrical flow with highest velocities near the inside wall of the scale model (left of
the sketch in figure 11) and near the protrusion of the inlet duct into the scale model.

Page 18 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 10 Measured velocity components and velocity magnitudes (m/s) along a traverse parallel to
the longitudinal axis of the scale model inlet duct. Locations of the 1 and 38 cm distances are
shown on the sketch.

Figure 11 Measured velocity components and velocity magnitudes (m/s) along a traverse
perpendicular to the longitudinal axis of the scale model inlet duct. Locations of the 1, 33
and 38 cm distances are shown on the sketch.

Figure 12 shows the measured velocity magnitudes between the scale model scrubbing
vanes. The velocity distribution is non uniform with velocity magnitudes varying by a
factor of 2.9.

Page 19 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 12 Measured velocity magnitudes (m/s) between the scale model scrubbing vanes.

Further tests were carried out with water added to the trough on the water distribution
cone as air flowed through the scale model. The rate of water addition was chosen so
that the rate of water addition per unit inlet air volume flow would be approximately
equal to the design rate of water addition per unit inlet gas volume flow for a full scale
scrubber module. It was not possible to measure the air flow distribution in the main
body of the scale model because the water would have damaged the hot wire
anemometer but it was still possible to measure velocities in the scale model inlet duct
(and therefore the air volumetric flow rate) and the pressure drop across the scale
model. Tests for different air flow rates found that water addition reduced the
volumetric flow through the scale model by approximately 7% and increased the
pressure drop by approximately 23%. This corresponds to an increase in fan power
consumption of approximately 14%. This increase in pressure drop with water addition
is much less than that reported in the literature for venturi scrubbers (Schifftner and
Hesketh, 1983). Note that in a full scale operating scrubber there will be a reduction in
gas volume flow as the gas is cooled by the water; this should partially compensate for
the increase in pressure drop due to water addition. This effect is not picked up in the
scale model experiments because air enters at ambient temperature and there will be
little cooling of this air when water is added.

Figure 13 shows that a highly agitated water bath forms between the scrubbing and
demisting vanes with water addition. It appears that a lot of the gas cleaning can occur
as it passes through this water bath. It was also noted that at high rates of water
addition this water bath overflowed past the demisting vanes. This could be a
mechanism for the water droplet carryover that often occurs when too much water is
added to an operating tangential entry fixed vane wet scrubber. Furthermore, it was

Page 20 Sugar Research Australia- Research Funding Unit- Research Project Final Report
found that when the air flow rate was very high, the water bath lifted up into the
demisting vanes. This is a likely mechanism for the increased droplet carryover
experienced by tangential entry fixed vane scrubbers when a boiler is operated with very
high gas flows resulting from tramp air leakage and/or the boiler operating at high
steam loads.

Figure 13 Part view of the scrubber scale model showing the agitated water bath between the
scrubbing and demisting vanes.

Flow visualisation in the scrubber scale model

Images from the flow visualisation experiments are shown in figure 14. Figure 14 (a)
shows a standard camera image of the flows through the scrubber scale model. Note
that with this image the wall water film obscures the internal flow patterns. The image
from the high speed camera taken at 250 frames per second (fps) in figure 14 (b) clearly
shows the agitated water occupying most of the volume between the scrubbing and
demisting vanes. A close up image of some droplets above a demister vane taken with a
very small exposure time (5000 fps camera speed) is shown in figure 14 (c) and
figure 14 (d) shows some droplets in the above demister vane region taken with the high
speed camera operating at 1000 fps. A close up view showing the breakup of a water
sheet from the water distribution cone is shown in figure 14 (e). Figure 14 (f) shows a
close up view between two scrubbing vanes taken by the high speed camera at 2000 fps.
The liquid film flowing down the top of the vane shown in the bottom of the image is
broken up into ligaments as it flows over the bottom of the vane by the upward air flow.

Page 21 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 14 Scrubber scale model images from the flow visualisation experiments: (a) regular camera,
(b) high speed camera at 250 fps, (c) close up view showing part of a demister vane at 5000
fps, (d) close up view showing region above demister vanes at 1000 fps, (e) close up view
showing the breakup of a sheet from the water distribution cone at 1000 fps, (f) close up
view showing the liquid film separating from the scrubbing vanes breaking up into ligaments
and droplets at 2000 fps.

Test rig scale model air flow modelling

Figure 15 shows a side elevation view of the predicted air velocity vectors in the
scrubber scale model. The green colour and longer vector lengths in the figure
correspond to higher air velocities. The highest air velocities occur near the wall of the
scrubber scale model below the scrubbing vanes and through the scrubbing vanes.
Recirculation zones are predicted to occur around the centreline of the scrubber scale
model below the scrubbing vanes and demisting vanes and above the demisting vanes.
The predicted recirculation zones above and below the scrubbing vanes correspond to
the observed ash build-up in existing scrubbers seen in figure 7. The high predicted
velocities through the scrubbing vanes seen in figure 15 correspond to the minimal ash
build-up on the scrubbing vanes seen in figure 8(a).

Page 22 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 15 Predicted air velocity (m.s-1) distribution through the scrubber scale model.

Initial full scale scrubber gas flow modelling

CFD simulations of two different full scale scrubber modules (scrubber module A and
scrubber module B) were performed. These modules are different size and have similar
but not identical geometries. These simulations used calculated gas flows based on
operating data for the two boilers.

The first simulations carried out only considered the gas flow distributions through the
scrubber modules without any water addition and it was assumed that the properties of
the gas were the same as those of air at 25 °C (test rig scale model conditions) and that
these properties did not change. This is clearly not the case in an actual scrubber but the
purpose of the first set of simulations was to predict full scale scrubber module gas
velocity distributions that could be directly related to the test rig measurements and the
CFD simulations of the test rig scale model.

Figures 16, 17 and 18 show plan views of the predicted gas velocity distribution at two
different elevations in scrubber module A for a modelled gas flow rate of 34 kg/s. The
elevation of the plane of view in figure 16 is through the scrubbing vanes, the elevation
of the plane of view in figure 17 is above the water distribution cone and the elevation of
the plane of view of figure 18 is through the demisting vanes. The scrubber module gas
inlet duct is to the upper right and below the planes of view of all figures. Figure 17
shows the predicted gas velocities are higher on the gas inlet duct side of the scrubbing
vanes and the lower gas velocities are on the opposite side of the scrubber module.

Page 23 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Above the water distribution cone the flow is a lot more concentrated towards the inside
wall and the location of the highest gas velocities has moved clockwise around the
scrubber module. The flow through the demister vanes (figure 18) becomes a lot less
concentrated towards the inside walls of the scrubber module and the predicted gas
velocities are lower. As the flow spirals up through the demister vanes the location of
the highest gas velocities has moved clockwise to the opposite side of the module to the
gas inlet duct. All three plots show highly non-symmetrical flow.

Figure 16 Plan view of the predicted gas velocity (m/s) distribution through the scrubbing vanes of
scrubber module A at 34 kg/s gas flow.

Page 24 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 17 Plan view of the predicted gas velocity (m/s) distribution above the water distribution cone
of scrubber module A at 34 kg/s gas flow.

Page 25 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 18 Plan view of the predicted gas velocity (m/s) distribution through the demisting vanes of
scrubber module A at 34 kg/s gas flow.

Figure 19 shows plan views of the predicted gas velocity distribution through scrubber
module B at four different gas flow rates. The modelled gas flow rate determines
velocity magnitudes as expected but has little effect on the relative velocities; the
locations of the high and low velocity zones are almost identical at the different gas flow
rates. Furthermore, the predicted gas velocity distributions for scrubber module B are
quite similar to those for scrubber module A (figure 16).

Page 26 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 19 Plan view of the predicted gas velocity (m/s) distribution through the scrubbing vanes of
scrubber module B at (a) 34 kg/s, (b) 30 kg/s, (c) 24.25 kg/s and (d) 15 kg/s gas flow.

Prediction of collection efficiency

The upgraded CFD code was used to model the air and water flows through the scrubber
scale model. Once a converged solution for the air and water flows was obtained, dust
particles with a size distribution measured at the inlet of a full scale dust collector, were
introduced into the simulation. This process was repeated for different air flows and
post processing was carried out to predict the collection efficiency for each of these air
flows.

The predicted variations of overall collection efficiency with air flow through the
scrubber scale model without and with droplet carryover being taken into account are
compared in figure 20. The predicted droplet carryover, expressed as a percentage of
the inlet water flow, is also included in the figure. The predicted collection efficiency
without taking droplet carryover into account increases with the air flow over the range
of air flows simulated but the rate of increase starts to taper off at air flows above
0.22 kg/s. When the reduction in collection efficiency due to droplet carryover is taken
into account the collection efficiency of the SSM is predicted to decrease when the air
flow exceeds 0.287 kg/s. This reduction in the predicted collection efficiency
corresponds to the predicted droplet carryover exceeding 3% of the inlet water flow.

Page 27 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 20 Predicted variation of collection efficiency (without and with droplet carryover being taken
into account) and droplet carryover with air mass flow rate in the scrubber scale model.

Proposed design changes for improving scrubber performance

Based on observations during the scrubber scale model experiments and modelling of the
scrubber scale model and full scale scrubber, several scrubber design changes were proposed
which included:
 Raising the demisting vanes
 Installing a breakwater annulus between the scrubbing and demisting vanes and
raising the demister vanes
 Installation of a vertical breakwater between the scrubbing and demisting vanes
 Addition of a drainage slot in in the annulus supporting the scrubbing vanes
 Addition of a vertical breakwater in the bottom cone of the scrubber
Modelling and/or modifications to the scrubber scale model were used to evaluate these
design changes.

Raising the demister vanes

Droplet carryover can become excessive in a full scale scrubber when the agitated
swirling water bath (figure 13) above the scrubbing vanes extends close to or beyond the
demisting vanes.

Page 28 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Raising the demisting vanes makes it less likely for the water bath to extend past the
demisting vanes. Eulerian modelling of the air and water flows through the scrubber
scale model was carried out without and with the raised demister vanes. Figure 21
compares predicted volume fraction distribution of 550 µm diameter droplets in the
scrubber scale model with the existing distance between the demisting and scrubbing
vanes and with the distance between the two sets of vanes increased by raising the
demisting vanes. The predicted volume fraction of droplets above the raised demisting
vanes is significantly lower than the droplet concentration above the demisting vanes in
the existing position. This suggests that droplet carryover should be reduced with the
raised demisting vanes.

Figure 21 Predicted contours of the volume fraction of 550 µm droplets in the scrubber scale model
with (a) the existing spacing between the scrubbing and demisting vanes and (b) the
proposed raised demisting vanes.

Installing a breakwater annulus between the scrubbing and demisting vanes and raising
the demister vanes
The breakwater annulus is intended to work by restricting the flow of water droplets up
the inside wall of the scrubber towards the demister vanes. CFD simulations predicted
and subsequent modifications to the scrubber scale model confirmed that provided the
inside diameter of the breakwater was greater than the inside diameter of the annulus
supporting the scrubbing vanes, installing the breakwater would not increase the
pressure drop across the scrubber module. Figure 22 shows the predicted volume
fraction distribution of 550 µm diameter droplets in the scrubber scale model with the
proposed breakwater installed and the raised demister vane assembly. Very little
carryover of 550 µm diameter droplets is predicted to occur with these changes.

This breakwater annulus and raising the demisting vanes modification was carried out
on the scrubber scale model and the effects of these changes are shown in figure 23.
With these modifications much less water enters the demisting vanes.

Page 29 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 22 Predicted contours of the volume fraction of 550 µm droplets in the scrubber scale model
with the breakwater installed and the raised demister vane assembly.

Page 30 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 23 Flow patterns in the scrubber scale model (a) before and (b) after the addition of the
breakwater and the raising of the demister vanes.

Installation of a vertical breakwater between the scrubbing and demisting vanes

Figure 24 shows the geometry of the vertical breakwater attached to the inside wall of
the scrubber module. The purpose of this vertical breakwater is to disrupt the swirl of
the water bath above the scrubbing vanes and force the water to run down along this
breakwater back towards the central scrubbing vanes. The circumferential location of
this breakwater should correspond to the area of the wall that faces the incoming flow.
This will make more of the water running down through the scrubbing vanes come in
direct contact with the incoming flue gas so that as much dust as possible can be
removed from the gas prior to it entering the scrubbing vanes. This proposed
modification was evaluated by modelling the gas flows through a full scale scrubber
module with this modification. Figure 25 shows a plan view of the predicted gas velocity
distribution in a plane between the scrubbing and demister vanes of a full scale scrubber
module with the vertical breakwater.

Page 31 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 24 The proposed vertical breakwater on the scrubber wall between the scrubbing and
demisting vanes.

Page 32 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 25 Plan view of the predicted gas velocity distribution in a full scale scrubber module in a plane
between the scrubbing and demisting vanes with the added vertical breakwater baffle.

Addition of a drainage slot in in the annulus supporting the scrubbing vanes

To help reduce the level of the water bath between the scrubbing and demister vanes a
drainage slot could be installed in the annulus supporting the scrubbing vanes.
Figure 26 shows the predicted volume fraction distribution of 550 µm droplets in the
scrubber scale model with such a slot installed. It is recommended that the
circumferential location of any such slots should correspond to the area of the wall that
faces the incoming gas flow. This will help ensure that there will be a substantial film of
water on this part of the wall below the scrubbing vanes that can capture the incoming
dust particles in a full scale scrubber module.

Figure 26 Side elevation view of the predicted volume fraction of 550 µm droplets in the scrubber scale
model with the slot installed in the annulus supporting the scrubbing vanes.

Page 33 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Addition of a vertical breakwater in the bottom cone of the scrubber
During the scale model tests it was noted that water accumulates and swirls in the
bottom cone which can cause water to splash into the inlet duct. In a full scale scrubber
module this water would be dirty and could cause build-up in the scrubber inlet duct.
The scrubber scale model was modified by installing a breakwater in the bottom cone
that stops most of the swirl occurring and reduces the water level in the cone. Figure 27
shows water flowing in the bottom cone of the scrubber scale model with the
breakwater plate installed.

Figure 27 Water flowing in the bottom cone of the scrubber scale model with the breakwater plate
installed.

Page 34 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Feedback from factory staff and a scrubber manufacturer
There was general agreement that the breakwater annulus and the raised demister
vanes should reduce droplet carryover and that the vertical breakwater in the bottom
cone should reduce swirl in that region. However it was pointed out that raising the
demister vanes can be difficult to implement and could the option with only the
breakwater annulus be considered.

The vertical breakwater proposal provoked a mixed response; one respondent thought it
would reduce swirl which was undesirable while another thought a spiral vertical
breakwater that would direct the water flowing down the wall into drainage slots would
be more useful. Some respondents agreed with putting a drainage slot in the annulus
supporting the scrubbing vanes and suggested that multiple drainage slots would be
even better while other respondents did not think the drainage slots would be useful.

Updated modelling based on feedback

Figure 28 shows the predicted volume fraction distributions of 550 µm diameter
droplets in the unmodified scrubber scale model and in the scrubber scale model with
the breakwater installed and the height of the demister vanes unchanged. The predicted
volume fraction of droplets above the demisting vanes and therefore droplet carryover is
reduced with the breakwater installed. However the predicted reduction in the droplet
volume fraction above the demister vanes is less than that predicted with the raised
demister vanes and with the raised demister vanes and breakwater annulus combination.

Figure 28 Predicted contours of the volume fraction of 550 µm droplets in (a) the unmodified scrubber
scale model and (b) the scrubber scale model with the breakwater annulus installed.

A side elevation view of predicted air velocity vectors through the scrubber scale model
with the breakwater annulus installed is shown in figure 29. Note the high velocity air
flow from the scrubbing vanes to the right of the figure that is initially directed towards
the wall before passing inside the breakwater annulus and then into the demister vanes.
The upper limit of the swirling water bath is restricted to the height of the breakwater
annulus. If the breakwater annulus is too low the high velocity air flow from the
scrubbing vanes will be directed towards the scrubber scale model wall above the

Page 35 Sugar Research Australia- Research Funding Unit- Research Project Final Report
breakwater annulus and would not come in contact with the water bath. In a full scale
scrubber this corresponds to dust laden gas flow not being collected by the scrubbing
water and reduced collection efficiency.

Figure 29 Side elevation view of the predicted air velocity vectors through the scrubber scale model
with the breakwater annulus installed.

Note also from figure 29 that the direction of the air flow from the scrubbing vanes
corresponds roughly to the angle of the scrubbing vanes. A projection drawn from a
scrubbing vane to the wall at the angle of the scrubbing vane would give an estimate of
the lowest vertical position of the breakwater annulus.

CFD modelling of the test rig scrubber scale model was carried out to predict the effect of
the proposed vertical breakwater on the water and air flows. Figure 30 shows a side
elevation view of the iso-surface in the scrubber scale model where the water volume
fraction was predicted to be equal to 0.001 with the proposed vertical breakwater
installed. More water is predicted to flow down through the scrubbing vanes on the side
of the scrubber scale model with the vertical breakwater installed. The vertical
breakwater is on the opposite site of the scale model to the inlet duct so that the
incoming air flow is directed towards the water draining through the scrubbing vanes.
This corresponds to the incoming dust laden gas flow in a full scale scrubber being
directed towards where the water drains through the scrubber vanes for improved
mixing between the dust particles and the water and therefore improved collection
efficiency.

Page 36 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 30 Side elevation view of the iso-surface in the scrubber scale model where the water volume
fraction was predicted to be equal to 0.001 with the proposed vertical breakwater installed.

One of the concerns raised about the vertical breakwater proposal was the possibility
that it would reduce the swirl in the flow between the scrubbing and demister vanes.
Figure 31 shows a part plan view of the predicted air and droplet velocity vectors
through the scrubber scale model with the vertical breakwater installed. There is only a
small disruption predicted to the main swirling air flow (black velocity vectors) that
occurs around the vertical breakwater. It is therefore unlikely that a vertical breakwater
of the modelled relative size will have much effect on the overall swirling flow pattern.
Figure 31 also shows that the droplet velocity vectors have a large radial velocity
component while the air velocity vectors have virtually no radial velocity component.

Page 37 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Figure 31 Part plan view of the predicted air (in black) and droplet (other colours) velocity vectors
between the scrubbing and demisting vanes of the scrubber scale model with the proposed
vertical breakwater installed.

Recommended scrubber design changes

From the modelling and experimental work carried out in this project the combination of
raising the demister vanes and installing a breakwater annulus should significantly
reduce droplet carryover. The modelling indicates that raising the demister vanes would
be the next most effective option for reducing droplet carryover followed by the annular
breakwater between the scrubbing and demister vanes. It is noted that if the breakwater
is too close to the scrubbing vanes collection efficiency could reduce. There was general
agreement between the factory staff and the scrubber manufacturer that these
modifications should reduce droplet carryover.

When a vertical breakwater was installed in the bottom cone of the test rig scale model,
the water drained more quickly as the breakwater reduced the swirl. This suggests that
a vertical breakwater installed in the bottom cone of a full scale scrubber module should
reduce the swirl and help the dirty water in the cone drain more quickly. There was
general agreement that this modification would be useful. If the bottom cone of a full
scale scrubber can drain more quickly this should reduce the amount of dirty water
splashing up into the gas inlet duct and therefore reduce inlet duct blockages.

Even though it received mixed feedback, a vertical breakwater installed between the
scrubbing vanes and the demister vanes may also be worth considering because it is
predicted to help drain some of the water from between the scrubbing and demisting
vanes which will help when the level of the swirling water bath gets too high. It should
also direct more water into the path of the incoming dust laden gas flow which should
improve collection efficiency.

Page 38 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Discussion

This project has achieved all its objectives. An improved understanding of the flow
processes occurring in centrifugal wet scrubbers has been obtained. Based on the
knowledge gained and the CFD modelling carried out in this project, design changes for
improved scrubber performance have been recommended.

The air flow only tests on the scrubber scale model and the associated modelling
identified the strong non symmetrical flow patterns and the low velocity zones that
correspond to build up in full scale scrubber modules. The predicted air flow
distributions through the scrubber scale model were similar to the predicted gas flow
distribution in the modelled full scale scrubber modules.

The tests on the scrubber scale model with both air and water flow showed how a
swirling water bath forms between the scrubbing and demisting vanes. At low air flows
the water bath did not form because the air velocities in the scrubber scale model were
not sufficient to prevent water draining back down through the scrubbing vanes. At very
high air flows the water bath extended into the demister vanes which resulted in
excessive droplet carryover. This is consistent with the increased droplet carryover that
occurs in a full scale scrubber module when the gas flows are very high. It is likely that
the water bath captures the dust particles that pass through the scrubbing vanes.
Without the water bath, this capture of dust particles that pass through the demister
vanes is unlikely to occur which is consistent with the lower collection efficiency of wet
centrifugal scrubbers at low loads. The CFD modelling of the air and water flows
through the scrubber scale model with different air flows produced results that were
similar to the observations. The high speed photography identified the different stages
of water curtain break-up. Agglomeration of water droplets was observed to occur
primarily at the scrubber wall which is mostly covered by a film formed by these
droplets below the scrubbing vanes.

The CFD code was updated to predict the capture of dust particles by water droplets and
the wall film. The model updates included a kinetic energy test for dust particle capture
that takes into account the velocity of the dust particle relative to the water and the
water surface tension. With the upgraded CFD code the collection efficiency of the
scrubber scale model could be predicted for a range of air flows. The collection
calculations were adjusted to take into account the reduction in collection efficiency due
to droplet carryover. With the updated collection efficiency calculation the air flow rate
at which collection efficiency is maximised could be determined. Many other collection
efficiency models used in the literature do not take droplet carryover into account and
will therefore predict that collection efficiency always increases with gas flow, which is
not the case in practice.

Observations from the scrubber scale model experiments and associated CFD modelling
formed the basis of proposed design changes. Further modelling and modifications to
the scrubber scale model were made to evaluate the proposed design changes. The
feedback from factory staff and the scrubber manufacturer on the proposed design

Page 39 Sugar Research Australia- Research Funding Unit- Research Project Final Report
 changes and further modelling was carried out based on this feedback. This lead to
updated scrubber design change recommendations.

Acknowledgements

The input and assistance provided by Bingera Mill and Isis Mill staff during the scrubber
inspections is greatly appreciated. The input and assistance provided by Neil McKenzie
and Barry Hume with the experimental measurements and flow visualisation is
gratefully acknowledged. The authors appreciate the feedback from the factory and
Fowlerex staff on the proposed scrubber design modifications and the computing
support provided by QUT high performance computing. QUT is acknowledged for
funding the attendance by Hassan Ali at the Australian Heat and Mass Transfer
Conference at QUT and the 11th European Fluid Mechanics Conference in Seville.

References

Adeniji-Fashola A, Chen CP (1990) Modeling of confined turbulent fluid-particle flows

using Eulerian and Lagrangian schemes. International Journal of Heat and Mass
Transfer 33, 691-701.
Ahmadvand F, Talaie MR (2010) CFD modeling of droplet dispersion in a Venturi
scrubber. Chemical Engineering Journal 160, 423-431.
Ali M, Yan C, Sun Z, Gu H, Mehboob K (2015) Dust particle removal efficiency of a venturi
scrubber. Annals of Nuclear Energy 54, 178-183.
Bird RB, Stewart WE, Lightfoot EN (1960) 'Transport phenomena'. Wiley New York.
Boyd RK (1986) Computer modelling of a coal fired furnace. PhD thesis, University of
Sydney, Sydney.
Boysan F, Ayers WH, Swithenbank J (1982) A fundamental mathematical modelling
approach to cyclone design. Transactions of the Institution of Chemical Engineers
60, 222-230.
Calvert S (1970) Venturi and other atomizing scrubbers efficiency and pressure drop.
AIChE Journal 16, 392-396.
Calvert S, Goldshmid J, Leith D (1974) Scrubber performance for particle collection.
AIChE Symp. Ser 72, 357-364.
Chuah TG, Gimbun J, Choong TSY (2006) A CFD study of the effect of cone dimensions on
sampling aerocyclones performance and hydrodynamics. Powder Technology 162,
126-132.
Cortes C, Gil A (2007) Modeling the gas and particle flow inside cyclone separators.
Progress in Energy and Combustion Science 33, 409-452.
Costa M, Riberio A, Tognetti E, Aguiar M, Gonclaves J, Coury J (2005) Performance of a
venturi scrubber in the removal of fine powder from a confined gas stream.
Materials Research 18.
Crowe CT (1982) Review - Numerical models for dilute gas-particle flows. Journal of
Fluids Engineering 104, 297-303.
Crowe CT, Pratt DT (1974) Analysis of the flow field in cyclone separators. Computers &
Fluids 2, 249-260.

Page 40 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Crowe CT, Sharma MP, Stock DE (1977) The particle-source-in-cell (PSI-Cell) model for
gas-droplet flows. Journal of Fluids Engineering 99, 325-332.
Dixon TF, Mann AP, Plaza F, Gilfillan WN (2005) Development of advanced technology
for biomass combustion - CFD as an essential tool. Fuel 84, 1303-1311.
Downing CM, Dixon TF, Pennisi SN (2000) Computational fluid dynamics modelling of
HMA Ducon wet scrubber - poster paper. Proceedings of the Australian Society of
Sugar Cane Technologists 22, 505-531.
Drew DA (1983) Mathematical modeling of two-phase flow. 15, 261-291.
Dukowicz JK (1980) A particle-fluid numerical model for liquid sprays. Journal of
Computational Physics 35, 229-253.
Durst F, Miloievic D, Schonung B (1984) Eulerian and Lagrangian predictions of
particulate two-phase flows: a numerical study. Applied Mathematical Modelling 8,
101-115.
Gosman AD, Ioannides E (1981) Aspects of computer simulation of liquid-fueled
combustors. AIAA 81-0234.
Griffiths WD, Boysan F (1996) Computational fluid dynamics (CFD) and empirical
modelling of the performance of a number of cyclone samplers. Journal of Aerosol
Science 27, 281-304.
Guerra VG, Béttega R, Gonçalves JAS, R. CJ (2012) Pressure drop and liquid distribution
in a venturi scrubber: experimental data and CFD simulation. Industrial and
Chemical Engineering Research 51, 8049-8060.
Harlow FH, Amsden AA (1975) Numerical calculation of multiphase fluid flow. 17, 19-52.
Hinze J (1972) 'Turbulence'. McGraw, New York.
Hounslow MJ, Ryall RL, Marshall VR (1988) A discretized population balance for
nucleation, growth, and aggregation. AIChE Journal 34, 1821-1832.
Jones WP (1949) Development of the Venturi scrubber. Industrial and Engineering
Chemistry 41, 2424-2427.
Karagoz I, Kaya F (2007) CFD investigation of the flow and heat transfer characteristics
in a tangential inlet cyclone. International Communications in Heat and Mass
Transfer 34, 1119-1126.
Lee B-K, Mohan BR, Byeon S-H, Lim K-S, Hong E-P (2013) Evaluating the performance of
a turbulent wet scrubber for scrubbing particulate matter. Journal of the Air &
Waste Management Association 63, 499-506.
Leonard G, Mitchner M, Self SA (1980) Particle transport in electrostatic precipitators.
Atmospheric Environment 14, 1289-1299.
Lockwood FC, Salooja AP, Syed SA (1980) A prediction method for coal-fired furnaces.
Combustion and Flame 38, 1-15.
Lora ES, Jativa IH (1999) Air pollution control in the sugar cane industry. Proceedings of
the XXIII Congress, International Society of Sugar Cane Technologists321-330.
Luo M, Stanmore BR (1994) Modelling combustion in a bagasse-fired furnace. 1:
Formulation and testing of the model. Journal of the Institute of Energy 67, 128-
135.
Mann AP (1996) Applications of computational furnace modelling. PhD thesis, University
of Sydney, Sydney.
Melville WK, Bray KNC (1979) A model of the two-phase turbulent jet. 22, 647-656.
Migdal D, Agosta VD (1967) A source flow model for continuum gas-particle flow. Journal
of Applied Mechanics 34, 860-865.

Page 41 Sugar Research Australia- Research Funding Unit- Research Project Final Report
 Mohebbi A, Taheri M, Fathikaljahi J, Talaie MR (2003) Simulation of an orifice scrubber
performance based on Eulerian/Lagrangian method. Journal of Hazardous
Materials 100, 13-25.
Moor BSC (2007) Flue gas scrubbing equipment for bagasse fired boilers. Proceedings of
the XXVI Congress, International Society of Sugar Cane Technologists1273-1283.
Narasimha M, Brennan MS, Holtham PN, Napier-Munn TJ (2007) A comprehensive CFD
model of dense medium cyclone performance. Minerals Engineering 20, 414-426.
O'Rourke PJ (1981) Collective drop effects on vaporizing liquid sprays. PhD thesis,
Princeton University.
Orme M (1997) Experiments on droplet collisions, bounce and coalescence. Progress in
Energy and Combustion Science 23, 65-79.
Pak SI, Chang KS (2006) Performance estimation of a Venturi scrubber using a
computational model for capturing dust particles with liquid spray. Journal of
Hazardous Materials 138, 560-573.
Palaszewski SJ, Jiji LM, Weinbaum S (1981) A three-dimensional air-vapor-droplet local
interaction model for spray units. Journal of Heat Transfer 103, 514-521.
Pan KL, Law CK (2007) Dynamics of droplet-film collision. Journal of Fluid Mechanics 587,
1-22.
Patankar SV, Spalding DB (1972) A calculation procedure for heat, mass and momentum
transfer in three-dimensional parabolic flows. International Journal of Heat and
Mass Transfer 15, 1787-1806.
Pemberton C (1960) Scavenging action of rain on nonwettable particulate matter
suspended in the atmosphere. Intern. J. Air Polution 3.
Pennington I (1999) Controlling particulate emission levels of bagasse fired boilers to
meet the new DENR regulations. Philsutech.
Pilch M, Erdman CA (1987) Use of breakup time data and velocity history data to predict
the maximum size of stable fragments for acceleration-induced breakup of a
liquid drop. International Journal of Multiphase Flow 13, 741-757.
Roache PJ (1976) 'Computational fluid dynamics'. Hermosa, Albuquerque.
Schifftner KC, Hesketh HE (1983) 'Wet scrubbers'. Ann Arbor Science Publishers Inc.,
Ann Arbor, Michigan.
Weber R, Visser BM, Boysan F (1990) Assessment of turbulence modeling for
engineering prediction of swirling vortices in the near burner zone. International
Journal of Heat and Fluid Flow 11, 225-235.
Whitaker S (1973) The transport equations for multi-phase systems. 28, 139-147.
Woodfield PL (2001) Combustion instability in bagasse-fired furnaces. PhD thesis,
University of Sydney, Sydney.
Wu P-K, Kirkendall KA, Fuller RP, Nejad AS (1997) Breakup processes of liquid jets in
subsonic crossflows. Journal of Propulsion and Power 13, 64-73.

Outputs and Outcomes

Outputs
 Test rig and scrubber scale model
 Improved modelling capability
 Updated CFD model

Page 42 Sugar Research Australia- Research Funding Unit- Research Project Final Report
  CFD modelling predictions
 Flow visualisation images of the water flows in the scrubber scale model showing
the water curtain break-up mechanisms
 Proposed scrubber design changes and the associated modelling and scrubber
scale model modifications that have been communicated to factory staff and the
scrubber manufacturer Fowlerex
 Recommended scrubber design changes
Outcomes
 Reduced factory down time due to fewer stoppages for cleaning
 Increased factory throughput by reducing steam output limitations imposed by
scrubber blockages
 Reduced emissions from sugar factory boilers
 Reduced complaints from the local community
Benefits
 Increased profitability of the sugar manufacturing process
 Reduced environment impact of the sugar industry

Intellectual Property (IP) and Confidentiality

No protectable IP was generated in this project.

Industry Communication and Adoption of Outputs

What key messages have come from the project to date, when and how they have been
communicated and to whom? Has there been any communication with the SRA
communication team or adoption team?

Project progress has been reported at the 2014, 2015, 2016 and 2017 regional research
seminars. Initial measurements and comparisons with model predictions were
presented at Hassan Ali’s PhD confirmation seminar on the 9 February 2015 to QUT staff
and students. Webinars on the project organised through Andrea Evers were completed on
the 10 September 2014 and 22 February 2017. The paper ‘Inside a wet scrubber’ was
presented at the 2016 ASSCT conference and was well received. Modelling and experimental
results from the project were presented at the Australian Heat and Mass Transfer
Conference at QUT (July 2016) and at the 11th European Fluid Mechanics Conference in
Seville (September 2016).

What new information, if any, is available on the adoption of project outputs?

There was widespread interest in implementing some of the modelled design changes.
Two of the recommended scrubber design changes: the vertical breakwater in the
bottom cone and a drainage slot in the annulus supporting the scrubbing vanes, were
implemented in two of the four scrubber modules on the Macknade No. 7 boiler. Initial

Page 43 Sugar Research Australia- Research Funding Unit- Research Project Final Report
 reports from factory staff suggest that there is reduced carryover from the two modules
where the design changes were implemented but this needs further follow up to confirm.

List any newsletters, fact sheets or any other media coverage.

There was an article on the project in issue 2 of Milling Matters in 2014.

Identify any further opportunities to disseminate and promote project outputs at seminars,
field days etc.

It is intended that a wrap up of the project will be presented at the 2018 regional
research seminars.

Environmental Impact

There was no environmental impact associated with carrying out the project. It is likely
that the scrubber design change recommendations, when implemented, will reduce the
environmental footprint of the sugar industry.

Recommendations and Future Industry Needs

Further follow up on the effects of the implemented design changes on operating

scrubbers is recommended.

Publications

The 2016 ASSCT publication from this project ‘Inside a wet scrubber’ is included in the
appendix. Two manuscripts based on the work in this project: ‘Flow visualization and
modelling of scrubbing liquid flow patterns inside a centrifugal wet scrubber’ and
‘Numerical prediction of dust capture efficiency of a centrifugal wet scrubber’ have been
submitted to international journals and are currently under review.

Page 44 Sugar Research Australia- Research Funding Unit- Research Project Final Report
 APPENDIX

2016 ASSCT paper ‘Inside a wet scrubber’

Page 45 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 46 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 47 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 48 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 49 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 50 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 51 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 52 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 53 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 54 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 55 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 56 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 57 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 58 Sugar Research Australia- Research Funding Unit- Research Project Final Report
Page 59 Sugar Research Australia- Research Funding Unit- Research Project Final Report