See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/27469554

Improved CFD modelling of natural circulation vacuum pans

Conference Paper · January 2006

DOI: 10.13140/RG.2.1.1867.5046 · Source: OAI

CITATIONS READS

3 281

3 authors:

Darryn Rackemann R. Broadfoot

Queensland University of Technology Queensland University of Technology
45 PUBLICATIONS   517 CITATIONS    43 PUBLICATIONS   105 CITATIONS   

SEE PROFILE SEE PROFILE

Darrin Stephens
Applied CCM
44 PUBLICATIONS   94 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Specifications for renewable crude oils View project

Development of a general purpose, open-source overset library for Caelus View project

All content following this page was uploaded by Darrin Stephens on 16 July 2015.

The user has requested enhancement of the downloaded file.

Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

IMPROVED CFD MODELLING OF NATURAL

CIRCULATION VACUUM PANS

By

D.W. RACKEMANN1, R. BROADFOOT1 and D.W. STEPHENS2

1
Sugar Research and Innovation @ QUT, Mackay, Qld
2
CSIRO Minerals, Clayton, Vic.
[email protected]

KEYWORDS: Vacuum Pan, CFD Modelling,

Performance, Circulation, Heat Transfer.

Abstract
NUMERICAL and especially CFD modelling are becoming cost-effective and
reliable ways to develop improvements in vessel designs. Simulating boiling in
crystallisation vacuum pans is a very complex process that needs to consider
non-isothermal conditions, multi-component, multiphase boiling and
condensation. This level of complexity which has been captured in the SRI CFD
model was previously too complicated for most CFD software. With
improvements to the software, these complex physical processes can now be
modelled, albeit a little cumbersomely. This paper details the development of
CFD models to predict the circulation patterns and heat transfer occurring in
natural circulation crystallisation vacuum pans. Model validation involved
checking the circulation velocities predicted by the CFD model with circulation
velocity data measured on factory pans. The predictions were in reasonable
agreement with factory measurements. The validated CFD model was used to
investigate the effect of altering key dimensions on batch pans and on two
different continuous pan designs. The batch pan investigations included:
increasing the volumetric region above the calandria by flaring the pan body;
reducing the clearance underneath the calandria; and changing the dimensions of
the tubes (tube diameter and length) while maintaining the same heat transfer
area and keeping the evaporation rate constant. The effect of variations in pan
geometry, massecuite viscosity and operating level above the calandria were
considered for the batch pan simulations. These results provide interesting
insight into the complicated processes involved in the operation of natural
circulation vacuum pans.
Introduction
CFD modelling has emerged as a powerful tool for designing and improving the
performance of process equipment that involves heat transfer and fluid flows. CFD modelling
is able to predict flow patterns within entire process vessels.
Validation using experimental measurements is required in order to obtain
confidence in the CFD model predictions.
Simulating boiling in crystallisation vacuum pans is a complicated process that needs
to consider non-isothermal conditions, multi-component and multiphase boiling and
461
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

condensation. This level of complexity has previously been too complicated for most CFD
software to handle adequately. With improvements to the software, these complex physical
scenarios can now be modelled, albeit a little cumbersomely. Much of the complexity is due
to the heat transfer that occurs within the calandria tubes.
CFD model development
The initial development of the CFD models of vacuum pans was based on a
continuous pan element as it represents a simpler geometry than typical batch vacuum pans
and experimental data were available.
A continuous pan can be likened to a number of batch vacuum pans operated in
series and therefore the CFD model developed for continuous pans is readily applicable to
batch pans.
The vacuum pans were simulated using a commercial software package – CFX
(ANSYS Inc, 2004).
The geometry was modelled as a two-dimensional (2D) thin slice utilising the
symmetry of the pan design thus minimising the computer memory and CPU requirements. A
slice thickness of 50 mm was used. The sliced geometry contained 7992 nodes and 25 746
elements.
The thin slice model could not model the geometry of the tube openings in the top
and bottom tube plates (tube diameters are typically 100 NB). Instead the openings were
modelled as slots, with the width of the slots calculated to give the same ratio of the cross-
sectional area of the tube openings to the tube plate area.
A segmented modelling approach, as developed by Stephens (2001), was adopted so
that individual calandria tubes need not be modelled.
In the segmented approach, the pan is divided into two segments: the space inside the
calandria tubes (the calandria tube segment), and the remaining part consisting of the
downtake and the regions above and below the calandria (the external flow segment).
The external flow segment is modelled using the standard CFD approach, whereas
the calandria tube segment is modelled separately. Modelling of flow and heat transfer
behaviour within the calandria tubes is simplified to a one-dimensional (1D) model as the
heat transfer is too complex for current CFD codes.
The two segments are coupled together, through the calculated boundary conditions
at the top and bottom of the tubes, to obtain the overall model of the entire vacuum pan.
Stephens and Harris (2002) provide more detail on the 1D model.
A two-phase Eulerian-Eulerian approach is used to model the massecuite and vapour
in the CFD model. The massecuite which is essentially two-phase (made up of sugar crystals
and molasses) is lumped together as a single phase and is treated as a Newtonian fluid. In the
model, the only effect of the crystals is to increase the fluid viscosity. The vapour is modelled
as a Newtonian ideal gas.
Modelling of multiphase flows requires information regarding the interaction
between the phases. The interphase momentum transfer between the phases is modelled using
a particle-based model. In the particle model the dispersed phase is treated as a group of
particles with a uniform size when calculating the momentum transfer terms for the
momentum equations.
The bubble size and distribution are important as they define the interfacial area for
transfer of mass, momentum and energy. The specified bubble size is important since it
strongly influences the size of the flashing zone (froth layer) at the free surface.
462
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

The CFD model has the flexibility to incorporate numerous drag relationships, heat
transfer relationships, bubble size and distribution parameters as either a user-defined
variable such as a correlation derived from the literature or factory data, or as a constant
value. For the simulations the following were used:
! Density difference method to model the fluid buoyancy.
! Average and constant particle diameter (bubbles) of 100 mm; however, the
bubble diameter can be changed to vary with height or can be specified as
groups of bubbles of various sizes. However these changes were not
investigated.
! The interphase drag force according to the Schiller-Naumann model (Schiller
and Naumann, 1933), which is the simplest of the available models.
! The interfacial heat transfer according to the Plesset-Zwick correlation
(Plesset and Zwick, 1954).
The simulations were performed to reflect typical fluid properties observed during a
batch high grade strike. These fluid properties included massecuite dry substance and purity
values of 90, density of 1450 kg/m3 and steam properties for an operating pressure in the head
space of 12 kPa abs. The steam temperature in the calandria (required for the 1D tube model)
was set at 100°C (typically the steam temperature reaches a maximum of approximately
120°C at ‘pan full’).
Due to the nature of the flow it was decided to treat the interface between the
massecuite and vapour as a free surface. This assumption meant the computational domain
had to be increased to accommodate changes in the free surface height for various
evaporation rates. The free surface height is directly proportional to the hold-up of vapour
(vapour volume fraction) in the liquid and the non-isothermal bulk boiling effects (flashing
and condensation). The vapour hold-up is also a function of the drag force which is a function
of the bubble size and shape, liquid properties and various other factors.
The simulations using the free surface model were performed as steady-state and
obtained a satisfactory level of convergence even though the solution was somewhat transient
in nature.
Implementation of bulk boiling
Previous attempts at CFD modelling of vacuum pans by Bunton (1981), Brown et al.
(1991), Stephens (2001), Rein et al. (2004) and Echeverri et al. (2005) did not consider the
bulk boiling or condensation effects and assumed for simplicity that the pan was isothermal
and all the vapour was produced within the calandria tubes.
This situation is in contrast to several theories in literature (Honig, 1959; Wright,
1966) that hypothesise that the massecuite leaving the calandria tubes in the upward direction
is not uniformly heated, and as a consequence there are many tiny areas where massecuite is
flashing to produce vapour and then condensing as the heated vapour comes in contact with
areas of cooler massecuite.
The incorporation of bulk boiling and condensation into the CFD model provided for
more stable simulations. The condensation proved to be a very important factor in the
solution stability that had been ignored in other significant CFD model investigations
(Stephens, 2001; Rein et al., 2004; Echeverri et al., 2005).
With no condensation, the initial modelling attempts in this project were affected by
transient wave fronts that passed along the free surface. These wave fronts were caused by
the bubbles produced from the boiling in the calandria tubes.
463
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

By considering condensation, a large amount of the vapour is removed (condensed)

and the momentum build up from the vapour travelling from the exit of the tubes to the
boiling surface has a smaller impact on the free surface interactions. The reduction of
momentum also reduces the vorticity above the calandria as predicted by Stephens (2001),
Rein et al. (2004) and Echeverri et al. (2005). Observations in industrial vacuum pans do not
demonstrate these regions of vorticity.
Figures 1 and 2 depict the massecuite volume fraction and massecuite velocity
patterns respectively for the full CFD model (including bulk boiling and condensation
effects) for a viscosity of 2 Pa.s, a boiling level of 600 mm above the calandria and an
evaporation rate of 14.5 kg/h/m2 in a SRI continuous pan. The figures show one half of the
pan about the centreline, at the left. Similar, but larger magnitude massecuite volume
fractions and liquid flow profiles and patterns are predicted by the CFD model at higher
evaporation rates.

Fig. 1—CFD model predictions of massecuite volume fraction for the continuous
pan element.

Fig. 2—CFD model predictions of massecuite superficial velocity for the

continuous pan element.
464
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

Figures 1 and 2 show that more fluid travels up the inner tubes which are also the
longest than travels up the outer tubes closest to the downcomer. This condition results in
more vapour being produced (due to longer residence time in the tubes) in this region. The
massecuite volume fraction profile predicts an increase in level close to the centreline
reducing to a lower level at the side of the pan. This prediction is very similar to observations
made in factory crystallisation pans where there is a rolling action from the middle to the
outside of the pan caused by the change in level. The flow of vapour from the tube openings
(calandria tube outlets) pluming to the free surface is predicted to produce a swirling
recirculation liquid flow pattern as shown in Figures 1 and 2.
The large amount of vapour produced near the centreline of the pan influences the
flow patterns above the top tube plate and causes a recirculation region to form above the
middle of the tube plate. The recirculation is in some part caused by the low boiling level
above the top tube plate causing the intersection of the flow rising out of the tubes near the
centreline of the pan and the resulting downwards flow after the liquid flashes at the surface
and flows towards the downcomer. The ability of the models to predict similar trends to
observations made in factory vessels gives confidence in the applicability of CFD modelling
techniques.
Validation of model
Factory trials were conducted on a SRI continuous pan to obtain circulation
measurements (using anemometer probes) and heat transfer measurements (Broadfoot et al.,
2004). The objective of the factory trials was to collect localised flow circulation data that
could be used to validate numerical models of circulation within crystallisation vacuum pans.
The technique and procedure for using anemometers to measure circulation velocities in
crystallisation vacuum pans have been reported by Rackemann and Stephens (2002).
Circulation velocity measurements were undertaken on the SRI continuous pan
across the downtake (near the top) and in a region below the middle of the calandria
(Broadfoot et al., 2004). The measured velocity data are in reasonable agreement with the
predicted data. The model parameters mentioned previously (such as the drag coefficient,
bubble size and size distribution) were not used to improve the CFD predictions to better
match the factory data. The reason for this was that accurate validation data is not available to
individually validate each of the model parameters mentioned (drag coefficient, bubble size
and size distribution etc.). At this stage, however, it was considered that the CFD model
provides reasonable results and trends to allow qualitative investigations of changes to design
to be undertaken.
Simulations
The continuous pan work was used to develop the CFD model to a stage where it
could be readily implemented to batch pan designs. Different batch pan geometries were
investigated to identify design features that would achieve improved circulation and provide
better productivity. These simulations show that the CFD models can be used as a tool to
determine and optimise vacuum pan designs.
Batch pan simulations
The validated CFD model was first used to investigate the effect of altering various
dimensions from a typical batch pan design (Figure 3). The pan is shown as a half cross-
section from the centreline, at the left, to the outer wall. The simulations were conducted for
two conditions that were chosen to represent the early and late parts of a batch high grade
strike. For the early part of the strike the boiling level was simulated at approximately
600 mm above the top tube plate and the viscosity of the solution was set at 2 Pa.s.
465
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

Fig. 3—Typical batch vacuum pan (Case 1): (a) dimensions; and (b) three
dimensional representation of a 5 degree wedge.

In simulating the later part of a batch high grade strike the boiling level above the top
tube plate was increased to approximately 1600 mm and the viscosity increased to 20 Pa.s.
Other fluid properties were as detailed previously. The evaporation rate used in both sets of
simulations was 17 kg/h/m2. The cases studied in the investigation are designated in Table 1.

Table 1—Batch pan geometries investigated in the CFD simulations.

Case Description
Case 1 Typical design of a 100 t pan. The calandria consisted of 1250 tubes
that were 1180 mm in length with an outside diameter of 114.3 mm. The
heating surface area of the calandria (including the area of tube plate
2
metal and the downtake) was ~556 m .
Case 2 Same as the typical design except the diameter of the vessel above the
calandria is increased by 600 mm. This change increases the volume of
the pan at the end of the strike, for the same boiling level, by 13–15 t.
Case 3 The apex of the double cone (bottom of the ‘W’ shape) in the typical
design is raised by 110 mm in the vertical direction. This height
reduction changes the ‘flow ratio1’ from 0.93 to 0.75 and reduces the
footing volume by ~ 2 t.
Case 4 The height of the calandria in the typical design is reduced by 105 mm
and the radius of the downtake increased by 85 mm. Additionally the
tube outside diameter was changed to 101.6 mm. For the same heating
surface area, the footing volume is reduced by ~ 2 t. In this simulation
1541 tubes were modelled.
Case 5 The height of the calandria in the typical design is increased by 105 mm
and the radius of the downtake decreased by 30 mm. Additionally the
tube diameter was changed to 127 mm. For the same heating surface
area, the footing volume is increased by ~ 2.7 t. In this simulation 1031
tubes were modelled.

1
The flow ratio is the ratio of the vertical area for massecuite flow below the downtake wall to cross-
sectional area of the downtake.
466
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

The CFD modelling was conducted on a wedge of five degrees which represented the
entire pan in much the same way that the thin slice represented the continuous pan element.
Figure 3 shows the dimensions of the original batch vacuum pan that was
investigated as well as a five degree wedge of the three dimensional representation.
In each of the batch pan geometries modelled, the calandria is split into ten
concentric segments. The tubes in each concentric segment (labelled tube groups 1 to 10 in
Figure 3) are represented by a stub at the top and bottom of the calandria region, on which the
boundary conditions (flow, temperature and vapour fraction) are imposed.
Batch pan results
Simulation of the base case (Case 1)
Some typical model results for Case 1 are displayed in Figures 4 and 5 for the
massecuite volume fraction profile and the massecuite velocity vector plot respectively. From
these figures, the vapour produced in the tubes surges to the surface taking the massecuite
along with it and in doing so producing the circulation patterns observed in Figure 5.
Quite high velocities (~1 m/s) are predicted within the downtake close to the centre.
The high velocity massecuite travels to the bottom of the pan and is then directed outwards
along the pan floor.
Two recirculation regions are produced: one that is small in magnitude close to the
downtake wall, and another above the calandria towards the outer wall which is larger in
magnitude.
The drag on the liquid and the recirculation draws the vapour away from the outer wall
as it rises to the surface. As the vapour is flashed at the surface, the liquid then flows back
towards the outer wall.
It is quite common to observe the flow of massecuite towards the outer wall in factory
pans due to this recirculation zone.

Fig. 4—Massecuite volume fraction profile for the batch vacuum pan – Case 1 at
2 Pa.s viscosity, operating with a boiling level of ~600 mm.
467
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

Fig. 5—Massecuite velocity vector plot for the batch vacuum pan – Case 1 at
2 Pa.s viscosity, operating with a boiling level of ~600 mm.

Effect of flared region above the calandria

Similar flow patterns to Case 1 are produced for Case 2 which incorporates the flared
region above the calandria. The velocity in the recirculation zone above the top tube plate
near the outer wall as shown in Figure 6 is slighter lower than compared to Figure 5.

Fig. 6—Massecuite velocity vector plot for the batch vacuum pan – Case 2 at
2 Pa.s viscosity, operating with a boiling level of ~600 mm.
468
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

The reduction in velocity is likely to have resulted from the increased volume caused
by the flared region. The volume of the recirculation zone is greater for the pan design with
the flared body.
Overall, the volumetric flow rate produced through the calandria section is lower in
Case 2 compared to Case 1. This reduction is caused, to some degree, by the energy wasted to
produce recirculation in the flared region which does little to contribute to the overall flow
patterns within the vessel. However, there is 13 to 15 t more material being circulated within
the pan compared to Case 1.
Effect of clearance under the calandria
Similar flow patterns and recirculation regions as in Case 1 are produced with the
reduced clearance under the calandria in the simulations of batch vacuum pan Case 3. In fact,
there is a slight reduction in the average circulation velocity. The reduction in velocity is
attributed to the increased resistance in the region below the calandria resulting from the
reduced clearance.
Effect of calandria tube geometry
In comparing the simulations of Case 4 (smaller tube diameter and tube length) and
Case 5 (larger tube diameter and greater tube length) to the base case (Case 1), similar flow
patterns and recirculation regions are produced.
The magnitude of the circulation passing through the calandria for Case 4 slightly
increased while the volume of massecuite in the pan has slightly reduced (by 2 t). For Case 5,
the magnitude of the recirculation region in the outer region of the pan above the top tube
plate was reduced relative to Case 1.
The magnitude of the circulation passing through the calandria also slightly reduced
compared to the base case while the volume of massecuite in the pan has slightly increased
(by 2.7 t).
Effect of viscosity and boiling level
For the second set of operating conditions simulated, the viscosity of the liquid was
increased from 2 Pa.s to 20 Pa.s and the static liquid level above the calandria was increased
from 600 mm to 1600 mm.
The increase in static level has the effect of raising the temperature required to
initiate boiling (caused by increasing hydrostatic pressure) which then suppresses the boiling
within the massecuite above the tube plate.
As well, the vapour is condensed after exiting the calandria tubes, as shown in
Figure 7 for the base case (Case 1). Figure 7 clearly shows the vapour produced in the
calandria tubes is condensed directly within the massecuite above the top tube plate due to
the higher boiling point.
However, as the massecuite flows towards the surface, flashing occurs and thus
provides net evaporation in the pan. Similar trends were observed for the simulations of
Cases 2 to 5.
Overall the circulation profiles predicted for the second set of operating conditions
for all cases showed the magnitude of the average velocity reduced to about 10% of the
average velocity when compared to the lower boiling level and lower viscosity conditions.
The results for the predicted massecuite velocity vector plot are shown for the base case
(Case 1) in Figure 8. A number of recirculation regions above the calandria were also still
present for all the cases considered.
469
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

Fig. 7—Massecuite volume fraction profile for the batch vacuum pan – Case 1 at
20 Pa.s viscosity, operating with a boiling level of ~1600 mm.

Fig. 8—Massecuite velocity vector plot for the batch vacuum pan – Case 1 at
20 Pa.s viscosity, operating with a boiling level of ~1600 mm.
470
Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

Summary of results
The volumetric flow rate through the calandria provides a measure of the overall
circulation within the vacuum pan. The measurements indicate that there are approximately
equal flows through each tube with a slight reduction in flow observed in the tubes closest to
the outer wall of the batch pans. The vapour flows produced in each tube also showed the
same trend. Past experience based on velocity measurements in factory pans suggests that
there should be a bias in flow through the tubes closest to the downtake as massecuite is
thought to preferentially flow up these tubes in some circumstances (Rackemann, 2004,
unpublished data). The reason for this bias is the reduced resistance to flow in the shorter
circulation circuit which exists for flow up the inner group of tubes.
The results of the total volumetric flow through the calandria are summarised in
Table 2 for all the five batch pan cases operating under the two sets of operating conditions.
Table 2—Summary of volumetric flows through the calandria of all the batch
CFD simulations.
Total volumetric flow, kg/s (in 5° model)
Batch vacuum pan Viscosity of 2 Pa.s, operating Viscosity of 20 Pa.s, operating
design level 600 mm above top tube level 1600 mm above top tube
plate plate
Case 1 44.14 4.42
Case 2 40.11 4.01
Case 3 40.87 4.12
Case 4 47.85 3.83
Case 5 43.57 4.95

Table 2 shows that at the lower viscosity and operating level, more circulation is produced
in the smaller diameter and shorter calandria tube (provided the same overall heat transfer area is
maintained) than in the larger diameter and longer calandria tube configurations. This trend is
reversed for the higher viscosity and higher operating level simulations. It was not possible to
determine the direct relationship between volumetric flows produced and tube diameter or tube
length separately since, in the five cases studied, both parameters were changed to maintain the
same heating surface area.
In comparing the two sets of operating conditions, higher vapour and liquid flows are
predicted above the calandria for the lower viscosity and lower operating level simulations, as
expected. Additionally, in the lower viscosity simulations, less condensation of the vapour produced
in the calandria tubes takes place immediately above the calandria and this vapour rises and flashes
at the liquid surface. The rise of vapour contributes to the flow patterns produced in the vessel. The
relative ratio of vapour flow to total flow (massecuite plus vapour) in the calandria tubes is reduced
for the lower viscosity simulations, caused by the faster flow rates through the calandria. The
massecuite has reduced time in the tubes to absorb heat and produce vapour. The simulations of
different viscosity used the same evaporation rate.
The higher viscosity simulations show lower massecuite flows above the calandria. The
vapour produced in the calandria is almost totally condensed after exiting the calandria tubes before
flashing as the flow nears the boiling surface. This condition results in much slower flow patterns
which are caused by non-isothermal currents rather than from the driving force caused by density
variations of multiphase flow.
Discussion and conclusions
The CFD model of batch pans predicts an even distribution of flow for the whole cross-
section of the calandria (rather than a significant flow bias). This results from a relatively uniform
471
 Rackemann, D.W. et al. Proc. Aust. Soc. Sugar Cane Technol., Vol. 28: 2006
__________________________________________________________________________

calculated pressure distribution below the calandria. In the continuous pan simulations the flow is
biased to directly below where recirculation regions are produced.
In the batch pan simulations, more circulation is produced in the smaller diameter and
shorter calandria tubes compared to the larger diameter and longer calandria tube configurations at
the lower viscosity operating conditions. This trend is reversed for the higher viscosity and higher
operating level simulations of batch pans. The direct relationship between volumetric flow produced
and tube diameter or tube length could not be determined in the limited simulations conducted.
The higher viscosity simulations show lower flows are generated. The vapour produced in
the calandria is almost totally condensed after exiting the calandria tubes before flashing as the flow
nears the boiling surface. This condition results in significantly lower buoyancy and flows. For the
lower viscosity and lower boiling level simulations only some of the vapour produced in the tubes
condenses before it reaches the boiling level.
The effect of the flared region above the calandria in the batch pan simulation was shown to
reduce the circulation velocities through the calandria section in comparison to a non-flared region
above the calandria. However, the additional massecuite quantity that the flared region provides per
pan cycle will likely be beneficial to the productivity of the pan and compensate for the slight
reduction in pan circulation. Reducing the clearance underneath the calandria in the batch pan
simulations also showed reduced volumetric flow rates through the calandria.
Acknowledgements
The funding assistance provided by the Department of State Development and Innovation
of the Queensland Government is acknowledged.
REFERENCES
ANSYS Inc. (2004). ANSYS CFX Version 5.7 Documentation. Pittsburgh, Pennsylvania, USA.
Broadfoot, R., Rackemann, D.W. and Cripps, L. (2004). Performance of the Tully continuous A
pan. Proc. Aust. Soc. Sugar Cane Technol., 26: (CD-ROM).
Brown, D.J., Alexander, K. and Boysan, F. (1991). Crystal growth measurement and modelling
of fluid flow in a crystalliser. Proc. Commission Int. Technique de Sucrerie, 19: 77–94.
Bunton, J.J. (1981). Natural convection, two-phase flow, and crystallisation in a vacuum pan sugar
crystalliser. Ph.D thesis, Louisiana State University.
Echeverri, L.F., Rein, P.W. and Acharya, S. (2005). Numerical and experimental study of the
flow in vacuum pans. Proc. Int. Soc. Sugar Cane Technol., 25: 212–221.
Honig, P. (1959). Principles of Sugar Technology, Volume 2, Elsevier, Amsterdam.
Plesset, M.S. and Zwick, S.A. (1954). The growth of bubbles in superheated liquids. Jnl. Applied
Physics, 25: 493–500.
Rackemann, D.W. and Stephens, D.W. (2002). Determining circulation velocities in vacuum
pans. Proc. Aust. Soc. Sugar Cane Technol., 24: 452–458.
Rein, P.W., Echeverri, L.F. and Acharya, S. (2004). Circulation in vacuum pans. Jnl. Amer. Soc.
Sugar Cane Technol., 24: 1–17.
Schiller, L. and Naumann, A. (1933). Uber die grundlegenden Berechnungen bei der
Schwerkraftaufbereitung. Zeitung des Vereins deutscher Ingenieure, 77: 318.
Stephens, D.W. (2001). Studies on modelling circulation in sugar vacuum pans. Ph.D. Thesis,
James Cook University, Townsville.
Stephens, D.W. and Harris, J.A. (2002). Prediction of evaporation, pressure driving force, and
heat transfer in calandria tubes. Proc. Aust. Soc. Sugar Cane Technol., 24: 431–436.
Wright, P.G. (1966). Circulation movements in sugar vacuum pans. Proc. Qd Soc. Sugar Cane
Technol., 33: 179–184.
472

View publication stats