Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

REFEREED PAPER

UNDETERMINED LOSS REDUCTION AT MAFAMBISSE

SUGAR FACTORY IN MOZAMBIQUE
DUNN NC1 AND MUZZELL DJ2
1
Tongaat Hulett Mafambisse, P.O. Box 1121, Beira, Mozambique.
2
Tongaat Hulett Sugar, Technology and Engineering Group, Private Bag 3,
Glenashley, 4022, South Africa
nelson.dunn@mafambisse.co.mz dave.muzzell@tongaat.com

Abstract

Undetermined loss, while by name suggesting a measure of ‘uncontrollability’, is considered

by many production managers as a key measure of performance in the raw house. The control
and reduction of undetermined loss can present many challenges to operational staff in a raw
sugar factory.

Mafambisse (MB) factory is situated in central Mozambique in the Sofala Province. From
2005 through to 2009, undetermined loss averaged 3% at MB. Investigations revealed that
the main cause of these sucrose losses was entrainment via the evaporator and Oliver filter
condensers. This paper discusses interventions made during 2007 and 2008 to overcome
these problems and eventually reduce undetermined loss to an acceptable average of 1.25%.

Keywords: undetermined loss, boiling house recovery, BHR, entrainment, condensers

Introduction

Mafambisse (MB) mill was established in 1972, and the majority of the plant that was
installed at that time is still in use. MB uses a pol based mass balance. No sucrose data is
available for mass balance calculations. The reduction in undetermined pol loss was achieved
over the period of a year in two separate areas, namely the filter and evaporator stations. The
cumulative benefit derived from reductions in undetermined loss in both these areas has
contributed to the current overall undetermined pol loss of 1.25% on mixed juice, down from
3% in previous seasons.

Filtration

Mafambisse has a six mill extraction tandem. The suspended solids in mixed juice is 0.8-
1.0% under normal operating conditions, and can rise during rainy periods to 1.4% or more.
The average suspended solids % mixed juice at 0.9% is marginally higher than most milling
factories. Harvesting of cane in low lying wet fields often results in soil attachment on the
cane and hence the slightly elevated suspended solids in mixed juice.

Filter cake and filtrate tonnages vary in proportion to the suspended solids level in mixed
juice. When abnormal pol losses are experienced at the filter station, these losses will be
inflated further when suspended solids in mixed juice is high.

427
Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

Traditionally, undetermined pol losses at the filter station occur when filtrate entrains into the
condenser. As the MB cooling water is ‘once through’ river water and the condenser tailpipe
is in an obscure location, entrained pol losses were not easily detected. Process staff at MB
suspected that pol losses were taking place through entrainment into the filter condenser. No
continuous sampling was undertaken; catch samples of tailpipe water were taken every hour
and tested for sucrose. Sucrose contamination was detected in the tailpipe water, and values
as high as 500 ppm sucrose were recorded from time to time.

There was insufficient data available to accurately determine filtrate tonnage, therefore an
estimate had to be made. Based on reasonable throughput estimates, Process staff calculated
the pol entrainment losses that could possibly be taking place.

MB averages 200 tons of mixed juice per hour. At suspended solids values of 0.9% on mixed
juice, it is reasonable to assume filtrate % mixed juice of about 15%. Historical data from the
Darnall sugar factory in South Africa, which is also a milling tandem, supported this
assumption. If 5% of filtrate at 12.5% pol is used as an arbitrary loss through entrainment,
then the average hourly pol loss in filtrate will be:

200 x (15/100) x (5/100) x (12.5/100) = 0.1875 tons/hour

Over the period of a normal week, this converts to about 0.9% pol loss on mixed juice.

Although the calculation was theoretical and based on assumptions, the magnitude of the
possible loss was enough to spur Process staff at MB to eliminate sucrose entrainment to the
filter condenser tailpipe water. It was decided to re-engineer the relevant equipment at the
filter station to eliminate this unquantified loss of pol to the cooling water. The modifications
undertaken in the 2007/08 offcrop included the installation of new vacuum equipment and
refurbishment of filtrate extraction equipment to prevent high liquid levels in the filtrate
tanks. Both these changes were designed to reduce the incidence of entrainment.

MB has four rotary drum filters, from which filtrate is drawn into separate high and low
vacuum filtrate receiving tanks. The incondensable gases and filtrate vapour flow into the
condenser via an entrainment separation tank. Incondensable gases are extracted from the
condenser by a single stage vacuum pump. Entrained filtrate is removed in the entrainment
separation tank and is returned to the high vacuum filtrate tank through a drain.

Both filtrate tanks have floats inside the tanks themselves. A stem on the float extends
through a vent valve located on top of each filtrate tank. When the filtrate level rises above
the tank high level mark due to pump cavitation, the float also rises up. The vent opens and
air is drawn into the filtrate tank that has the high level. The vacuum in this tank will decrease
and the suction head on the filtrate pump will increase. When the suction head increases, the
pump will start pumping again and this will reduce the filtrate level in the relevant tank. As
the filtrate level drops, the float moves down and the air vent is sealed again. The vacuum is
restored to the relevant filtrate tank. A mechanical pressure reducing regulator controls the
reduced vacuum in the low vacuum filtrate tank.

There were three filtrate pumps, a pump for each duty and a common standby pump. A
balance line connected the pump suction to the filtrate tank above it. The balance lines have
not been shown in Figure 1.

428
Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

Figure 1. Original filtrate station layout and control before 2008 season.

Modified filtrate station

The original installed system was prone to failure for a number of reasons, including
punctured floats, floats sticking and not opening the air vent, and jamming of the
counterweight pressure controller. It was decided to convert the old system to a more modern
level and vacuum control system for the 2008/09 season.

1. Low vacuum pressure is set on the pressure indicator controller (PIC). The pressure is
regulated by a butterfly control valve.
2. In the event of a high level in one of the filtrate tanks, the Mobrey switch float will rise,
activating and opening a solenoid valve to bleed air into the relevant filtrate tank. The
increased pressure in the tank will increase the suction head on the filtrate pump, resulting
in rapid removal of filtrate.
3. When the filtrate level drops, the Mobrey switch float will drop down, deactivating and
closing the solenoid valve. Vacuum will be re-established in the filtrate tank.
4. To visually monitor possible entrainment, a 120 mm diameter sightglass was installed on
the vapour/incondensable gas pipe leading from the filtrate tanks to the entrainment
separator and condenser. Entrainment is normally observed when large droplets travel up
the pipe to the entrainment separator.

429
Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

5. The entrainment return pipe from the entrainment separator to the filtrate tank was also
equipped with a sightglass for the same purpose. Any flow of filtrate in this pipe indicates
that filtrate has managed to reach the separator and corrective action is required.
6. A conductivity probe is installed in the vapour/incondensable gas pipe to notify operators
when there is any entrainment up this pipe. Notification takes place through an audible
alarm. The conductivity output is also trended for management inspection and evaluation
on a daily basis.
7. Filtrate pump inspection is undertaken regularly during the season to ensure that casing
and impeller erosion has not taken place. Mechanical erosion will adversely affect pump
efficiency and contribute to high filtrate tank levels and possible entrainment.
8. The low and high vacuum filtrate pump duties were totally separated, with both a
dedicated pump and standby pump for each duty.

The layout of the modified station is illustrated in Figure 2.

Incondensable Gas

Conductivity Electrode &

Audible Alarm
Condenser

Entrainment Sight
Separation Glass
Tank
Water PI PX
C
In
Solenoid Solenoid
Valve Valve

Mobrey Mobrey
Sight Switch
Switch
Glass

High Vacuum Low Vacuum

Filtrate Tank Filtrate Tank

Seal Well

Filtrate to MJ
Tank

Figure 2. Modified filtrate station.

430
Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

Results and Discussion

Apart from trending injection water conductivity measurement, sampling of filter condenser
tailpipe water was initiated at short 15 minute intervals for a 24 hour period over a number of
days. No sugar traces were detected. Monitoring of this area has become a normal
management function to ensure that no sucrose entrainment occurs.

Evaporator entrainment separators

During the 2005 season a decision was made to replace the corroded Sugar Milling Research
Institute (SMRI) vane type save-alls in all the evaporator vessels with vertical chevron plate
(VCP) entrainment separators. The decision was based on the knowledge that the VCP
entrainment separator is a new generation design that had yielded good results. The
undetermined loss (UDL) for the 2005 season was 2.6%. Drawings were made for the VCP
installations and an engineering firm was contracted to manufacture and install the VCPs on a
phased basis over a period of three years.

In the 2005/06 offcrop three evaporator vessels including the two last effects were fitted with
VCP entrainment separators. The UDL for the following 2006/07 season increased to 3.63%.
During the next offcrop, a further four vessels were completed. UDL remained high at 3.12%.
In the 2007/08 offcrop the remaining two 3rd effect vessels were completed. All the
evaporator vessels now had VCP entrainment separators. It was after the fabrication of the
last two sets of VCPs but prior to their installation that the two sets were examined by
Process staff and an alarming discovery made.

Around the periphery of the VCP packs large gaps of up to 50 mm were identified, together
with some areas between the louvres where the gaps were large enough to allow
uninterrupted sucrose entrainment to take place. After this discovery the other evaporator
vessels were also inspected as these had not been done at the time of installation during the
previous two offcrops. The same problems were identified in all the other vessels. MB
removed all the VCP packs and refabricated them according to design, sealing off all the
gaps.

Evaporator syrup extraction pumps

Mafambisse has two sets of quadruple effect evaporators. The layout before 2008 is
illustrated in Figure 3.

431
Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

Figure 3. Evaporator layout before 2008 season with one set of syrup extraction pumps.

The following were included in the evaporator arrangement:

1. The A-line evaporators had two 1st effects in parallel followed by single 2nd, 3rd and 4th
effects.
2. The B-line consisted of four individual vessels arranged in a quad layout.
3. Each evaporator line had individual water ejector condensers to generate vacuum. This is
not shown in Figure 3.
4. Syrup from each of the A and B-line final effects combined into a common syrup
manifold on the suction side of the syrup extraction pumps.
5. Syrup from the evaporators was pumped to the syrup storage tank on the pan floor.

Observations and Modifications

Level control
Syrup level in the A and B-line last effects was extremely difficult to control. Although the
absolute pressures in both lines varied slightly on occasion, the pressure differences were
small and had no measurable effect on the syrup level. However syrup levels in the two last
effects were seldom the same. When the syrup level in one of the last effects increased, it
created a higher suction head on the single syrup extraction pump than the line with a low
syrup level. As a result, syrup was extracted from the line with the higher syrup level in the
last effect vessel. The syrup level now started to fall in the vessel where syrup was being
extracted, while the syrup level in the other last effect continued to rise. When the syrup level
rose to a critical suction head, the syrup extraction changed from the last effect where the
level was dropping to the last effect where the syrup level was rising. In this way, the syrup
was extracted from the last effects in batches and the batching process was dependent upon
the rate at which the levels in the last effect changed. There was preferential extraction of
syrup from one line where the syrup level was high, while there was no syrup flow from the
other line.

432
Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

Periodically, the absolute pressure increased suddenly. When that happened in the last effect
where the syrup level was rising, it triggered an increase in suction head on the syrup
extraction pump, which forced a change in syrup removal to this vessel. The reverse applied
when the absolute pressure dropped suddenly and the suction head decreased on the syrup
extraction pump.

When the syrup level in one of the last effects increased, the disengagement height decreased
and sucrose entrainment increased. The rise and fall of syrup level in the last effects repeated
in a continuous cycle. Sucrose entrainment took place each time the syrup level increased and
reached a critical height in the last effect vessel.

Solution
The interactive effects of conditions in the last effects had to be removed.

During the 2007/08 offcrop changes were made to the syrup extraction system for both
evaporator lines. Each evaporator line was equipped with two syrup extraction pumps as in
Figure 4. The suction head condition on the pump was now affected only by the last effect
vessel to which it was assigned.

At the same time, the water ejector condensers (water jet condensers) from both lines were
replaced with a single standard Tongaat Hulett design condenser to service both the A and B-
evaporator lines. The last two effects were linked with a common vapour pipe that connected
to the common condenser. The condenser was equipped with liquid ring vacuum pumps. The
problem with variations in the absolute pressure in both last effect vessels was removed.

Figure 4. Modified evaporator layout from 2008 season

to include two sets of syrup extraction pumps.

433
Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

Combined results of four changes at MB

When MB started the 2008/09 season, four important changes had been made to the plant to
minimise sucrose entrainment into the cooling water system.

 The filtrate station was automated.

 VCP entrainment separation packs replaced the SMRI vane type save-alls in all
evaporator vessels.
 Additional syrup extraction pumps were installed to provide separate syrup extraction
facilities for each evaporator line.
 Absolute pressure variations between the A and B-lines were nullified by the installation
of a single condenser catering for both lines of evaporators.

All four of these changes were made to reduce sucrose entrainment from both the filter
station and the evaporator station. MB did not attempt to quantify the reduction in
entrainment from each of the individual changes. The end results reported here have been
attributed to the combined input from all four changes.

At the start of the 2008/09 season an immediate reduction in UDL was reported. For the first
eight weeks of the season, the UDL was below 1% on average. This was a massive
improvement when compared with the first four weeks of the 2007/08 season, where the
UDL was in the order of 4%. The results are shown in Figure 5. In the middle of the 2008/09
season the UDL drifted out due to very low time efficiencies. A number of A-centrifugal
breakdowns were experienced. In weeks 12 and 13 overall time efficiencies (OTEs) were
22% and 47.7% respectively, and the UDL for these weeks was 5.26% and 6.49%. The high
UDL in the last two weeks of the season can also be attributed to very low OTE. The adverse
effects of low OTE are clearly illustrated in Figure 5 trends.

14

13

12

11

10

UDL % 6 Before changes - 2007

5 After changes -2008

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-1

-2
Week

Figure 5. Undetermined loss (UDL) before and after

the four changes made at the Mafambisse mill.

434
Dunn NC and Muzzel DJ Proc S Afr Sug Technol Ass (2013) 86: 427 - 435

Except for 2009, the average UDL continued to drop after the changes to the filtrate system,
the installation of the VCP entrainment separator packs, the changes to the evaporator syrup
extraction pump configuration and the common vacuum system for both evaporator lines.
The UDL trends from 2005 to 2012 are illustrated in Figure 6. In 2009 a few weeks of very
low OTE, coupled with delivery of stale cane, resulted in a large quantity of mixed juice
being dumped. The UDL for the season was understandably high, but MB recovered from
this poor performance season to register excellent UDL figures for the following three
seasons. No further changes were made to the filtrate and evaporator stations during this
period.

UDL Trend 2005 - 2012

3.75 After four process changes

3.25

2.75

% 2.25

1.75

1.25

0.75
2005 2006 2007 2008 2009 2010 2011 2012
Season

Figure 6. Undetermined loss (UDL) trends at

Mafambisse mill from 2005 to 2012.

Conclusion

Process staff at Mafambisse mill recognised that there was scope to reduce the average UDL
from a value of more than 3% to less than 1%. The route to success was attention to detail.
This involved questioning all aspects of design and operation that could contribute to high
UDL. Automation at the filter station, the correction of the VCP pack installation in the
evaporators, the separation of the syrup extraction system for both evaporator lines, and a
common evaporator vacuum system for both evaporator lines all contributed in some way to
the reduction in UDL.

435