Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

REFEREED PAPER

SUCROSE LOSSES ACROSS THE GLEDHOW EVAPORATORS

DETERMINED USING NIRS PREDICTIONS
DAIRAM N1, RAMARU R2, NGEMA S1, SUTAR N1 and MADHO S2
1
Gledhow Sugar Company, 1 Gledhow Mill Road, KwaDukuza, South Africa, 4450
2
Sugar Milling Research Institute NPC, c/o University of KwaZulu-Natal, Howard College Campus,
Durban, South Africa, 4041
ndairam@gledhow.co.za, spngema@gledhow.co.za, nsutar@gledhow.co.za, smadho@smri.org, and
rramaru@smri.org

Abstract

Shorter crushing seasons and reduced cane throughputs, as a result of the severe 2014 and 2015
drought, called for a concerted effort by the Gledhow sugar mill to optimise on sucrose
recoveries for mill profitability. This proved to be challenging as the factory was not designed
for lower throughputs, and led to general high Undetermined Losses (UDL); on occasion more
than three times that acceptable for back-end refinery mills.

A UDL audit verifying stock-take procedures, sugar movements and authenticating all known
losses did not reveal the source of the losses. However, some losses across the evaporator
station were identified by mill staff because of the inconsistent cane supply, erratic throughputs,
and steam supply. On many occasions only a single tandem was operated at substantially
reduced throughputs over extended periods. At full capacity, Gledhow mill usually operates
both 200 tonnes cane per hour (TCH) milling and 100 TCH diffuser tandems. Indications of
possible losses across the evaporators included inconsistent evaporation rates, increased juice
residence times, lower syrup pH, higher syrup colours and lower condensate pH. The Gledhow
evaporators do allow for some vessels to be taken off-line, however, the station was not
designed for substantially reduced throughputs.

A dedicated team of samplers and analysts was assigned to determine the extent of the losses
across the evaporator station using the SMRI-NIRS toolkit. Three modes of operation were
investigated:
a) Both tandems in operation – all evaporator vessels were on-line;
b) Milling tandem operational only – one second effect and one fourth effect evaporator
taken off-line; and
c) Diffuser tandem operational only – same second and fourth effect vessels taken off-line
as with milling tandem only tests.

The tests were performed in October and November 2015 just prior to the end of the season.
Over 34 individual tests were conducted across the evaporator station, with each test
comprising measurements across all evaporator effects. From the tests performed and other
plant data, it was estimated that the evaporator inversion losses accounted for a maximum of
20% of the Undetermined Losses experienced. The higher inversion losses were incurred when
low brix juices were concentrated at low throughputs. This, unfortunately, is a common
practice at the mill to meet its refinery steam demands. Sufficient information was supplied to
mill management from the tests to propose modifications to the evaporator station to minimise
inversion losses.

391
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

Keywords: losses, UDL, NIRS, sucrose inversion, evaporators, capacity

Introduction

The Gledhow sugar mill (GH) has a design throughput of 300 tonnes cane per hour (TCH) and
a back-end refinery rated to produce 35 tonnes of refined sugar per hour. The extraction plant
consists of a 200 TCH milling tandem and a 100 TCH diffuser. After the severe 2014/15
drought season, GH started the 2015/16 season by mainly operating the milling tandem only
because of cane supply shortages. The diffuser came into full operation after 12 weeks of
crushing as GH’s cane supply improved (see Figure 1), however, the cane supply was still
inconsistent, leading to the frequent operation of only a single tandem (either diffuser or milling
tandem) at low throughput rates. GH’s overall tonnes cane crushed in a season has been
declining for the past three seasons as shown in Figure 2.

Figure 1. Weekly tonnes cane crushed from milling tandem (A) and diffuser (B)

High Undetermined Losses (UDL), increased sucrose losses in molasses, and poor Boiling
House Recoveries (BHR) were experienced in 2015/16 compared to the 2013/14 and 2014/15
seasons (see Figure 3). With regards to the former, detailed investigative work and audits were
carried out throughout the factory to quantify all known losses but the source of the excessive
losses could not be identified.

392
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

1600000 1 507 969

Crushing Weeks In Season:
1400000 2013 = 37 weeks
Tonnes cane crushed 1 257 948 2014 = 34 weeks
1200000 2015 = 30 weeks

1000000 938 522

800000
600000
400000
200000
0
2013 2014 2015
Year

Figure 2. Total tonnes cane crushed per year for the past three seasons

Undetermined losses
18 Molasses losses
Total losses in % sucrose in cane

16 Filter cake losses

3.91
14
3.61
12 2.10
10
8 12.08
6 9.76 10.03
4
2
0.39 0.33 0.53
0
2013 2014 2015
Year
Figure 3. GH annual sucrose loss distribution

Table 1 shows that the UDL in 2014/15 and 2015/16 seasons were substantially higher than the
previous drought years (1983/84 and 1992/93), as reported by Lionnet (1993).

In recent years, there has been a huge demand for refined sugar in the market. GH imports raw
sugar to maximise the refined sugar production since the cane supply has diminished over the
years. The cane supplied in the 2015/16 season could have been crushed over a much shorter
season i.e. within 25 weeks, however, due to erratic cane supply the cane was crushed over 30
weeks. This afforded the opportunity to import more raw sugar and maximise refined sugar

393
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

production. The erratic cane supply in the season though may have contributed to the UDL -
leading to the difference in UDL between the recent drought years (2014/15 and 2015/16) and
previous drought years (1983/84 and 1992/93).

Table 1. GH UDL (%) in different drought seasons

Drought Rainfall Cane crushed Season UDL
Season (mm) (tonnes) weeks %
1983/84 774 1 084 706 30 1.9
1992/93 521 774 414 20 2.2
2014/15 583 1 257 948 35 3.6
2015/16 845 938 522 30 3.9

In the past, sucrose inversion across southern African evaporators, and other unit operations,
was conventionally quantified by the Sugar Milling Research Institute NPC (SMRI) using gas
chromatography (GC) to analyse for fructose, glucose and sucrose (FGS). Since 2014, the
SMRI has made its SMRI-developed Near Infrared Spectoscopy (NIRS) prediction equations
available to its Full Members for the factories to determine pol, brix and FGS concentrations
on mixed juice and most factory intermediate processing streams. This means that sugar mills
can now quantify sucrose inversion losses themselves over a range of processing conditions.
The use of NIRS offers the possibility of rapidly conducting FGS analyses in mill laboratories
as opposed to SMRI-only tests previously possible by GC, which used to take longer to analyse
(Ramaru and Madho, 2015) and which were only snapshot tests performed at one processing
condition.

In the 2015/16 season, GH, using the SMRI-NIRS predictions equations, conducted 34
inversion loss surveys across its evaporator station under different operation configurations.
The surveys were performed in October and November 2015 prior to the end of the season by
GH factory staff.

This paper presents the results of the GH evaporator sucrose inversion loss survey conducted
during that period.

Methods

GH evaporator configuration
Figure 4 shows a schematic diagram of the GH evaporator station. The GH evaporator station
is designed for 300 TCH throughput. The total heating surface capacity of the station is
12 275 m2.

The evaporator station is currently not equipped with an option of reducing the heating surface
area in the first effect. As a result, water is added to the clear juice (CJ) tank to prevent drying
out of the tubes at low throughputs. Most importantly, water is added to CJ so that GH can
generate sufficient vapour bleed to operate the raw house and refinery. The second and fourth
effect heating surface areas can be reduced by taking vessels off-line at low throughputs. The
GH evaporator configurations used when tests were performed for different extraction plant
setups are shown below:

a) Both tandems in operation – all evaporators on line;

394
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

b) Milling tandem operation only – 2A and 4B vessels off line; and

c) Diffuser tandem operation only – 2 A and 4B vessels off line.

Process conditions such as CJ flow, CJ pH, exhaust steam temperature, and Vapour 1 and
Vapour 2 bleed pressures were recorded every hour during the trial.

Figure 4. A schematic diagram of the GH evaporator station

Sampling
A team of samplers and an analyst were assigned to take catch samples across the evaporator
station from CJ to syrup, in sequential order, at the sampling points depicted in Figure 4. The
sampling intervals between the vessels were based on the estimated retention times at the test
throughputs. The SMRI trained the GH team on how to perform the tests. The sampling
procedure detailed in a Standard Operating Procedure (SOP) by Ramaru and Madho (2015)
was followed throughout the survey, unless otherwise stated.

The SMRI recommended collection of samples at each sampling point every five minutes for
a period of an hour but due to staff shortages samples were collected only every 15 minutes.
Thus, a total of five samples was collected at each sampling point for each survey. All samples
were cooled quickly to 20°C in an ice bath to prevent sucrose inversion after collecting the
samples.

All evaporator intermediate samples were diluted to about 13% brix as per the SOP. The
samples were analysed by NIRS. The SMRI-NIRS CJ equation was used to predict FGS and
brix concentrations of the samples collected.

Sucrose inversion calculation

In the absence of accurate masses of processing streams, the sucrose inversion losses were
estimated using glucose/brix and sucrose/brix ratios as shown by Schäffler et al. (1985) (see
Equation 1). Schäffler (1986) reported that small losses of sucrose will be detected more easily

395
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

by a relatively large increase in glucose concentration. This approach assumes that there is no
glucose destruction in the evaporators.

glucose glucose
( )out -( )in
Brix Brix Sucrose_molecular_weight
%sucrose_lost= × ×100
sucrose Glucose_molecular_weight
( )in
Brix
… …Equation 1

An assumption made during the inversion loss survey was that the steady state conditions
persisted while the tests were performed. The NIRS data were not included in the evaluation if
unsteady state conditions persisted during sampling or if the CJ brix was below 5% (SMRI-
NIRS predictions below 5% brix were not validated at the time of the GH tests). Of the 34 GH
inversion loss surveys conducted in the 2015/16 season, only results from 18 surveys were
included in the evaluation of the results.

The results were categorised as follows:

Placed under low CJ brix category if CJ brix < 8.5%, otherwise high brix category; and
Placed under low CJ flow category if CJ flow < 240 m3/hr, otherwise high flow category.

Statistica® software was used to evaluate if the differences in sucrose losses between different
scenarios were statistically significant (Least Significance Difference (LSD) method used).

Results and Discussion

Overall sucrose lost i.e. from CJ to final effect syrup

Figure 5 shows the overall sucrose lost (calculated using Equation 1) over the GH evaporator
station at different CJ brixes. It is apparent that higher sucrose losses were obtained at low CJ
brixes. It is generally accepted that higher sucrose losses would be expected in a low brix
stream compared to a high brix stream if both streams were under similar physical conditions
– this would be due to higher brix retention times (verified using SMRI Vukov model -
Schäffler, 1994). The effect of brix on sucrose inversion loss is larger at high temperature i.e.
early effects.

Equation 1 is not suitable for quantifying small sucrose losses accurately and precisely because
small changes in glucose and brix levels lead to much variation (Davis, 1995 and Crebo, 1994).
This may explain some of the negative sucrose losses observed in Figure 5. The SMRI will
determine the level of uncertainty in the sucrose loss equation analytically in the 2016/17
season.

396
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

Figure 5. The overall sucrose lost (%) across GH evaporator station at different CJ brixes

Figure 6 shows the overall sucrose lost (%) over the GH evaporator station at different CJ brix
and flow levels. The average sucrose lost was about 1.8% when GH was operating at low CJ
flow and low CJ brix. The average sucrose lost was about 0.5% at high CJ brixes both at low
and high CJ flows. This was deemed to be an acceptable level of losses across an evaporator
station (personal communications with various South African sugar technologists, February
2016).

The statistical analysis showed that the difference in sucrose lost between the low CJ brix level
and high brix level (at high and low flows) was statistically significant at 5% level of
significance (i.e. p-value<0.05) (see Appendix , Table 1).

Low CJ brixes result in a longer dissolved solids retention time in the evaporators. Longer
retention times could lead to higher sucrose losses. The low CJ brixes at GH were due to the
addition of water to the CJ tank for reasons previously mentioned. This only occurred when
the cane throughput was low hence there was no case whereby GH recorded low CJ brixes at
high CJ flows. The sucrose lost at low CJ brix and low flow scenario, i.e. 1.8% sucrose loss
(see Figure 6), was equivalent to almost half of the GH UDL if only this processing condition
persisted, which was not the case. An exercise was conducted to determine the frequency and
duration of the low brix, low flow condition. However, this was not possible to quantify with
the recorded plant data alone. It was then assumed that this condition prevailed at times of high
Lost Time % Available (LTA) and shorter no-cane stops. If the higher losses persisted for this
percentage of the available time, it was estimated that the evaporator inversion losses were a
major contributor to the UDL, accounting for a maximum of about 20% of the UDL. It,
however, appears to be not the only contributor.

397
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

If CJ brix < 8.5%  Low brix

If CJ flow < 240 m3/hr  Low flow

High brix range Low brix range

Figure 6. Overall sucrose lost (%) at different brix range and flow range

Low pH promotes inversion of sucrose, however, measured CJ pH at GH was always between

6.8 and 7.3 throughout the study, indicating that poor pH control was not the cause of sucrose
inversion. Final syrup pH was between 5.8 and 6.1. Higher temperatures also promote sucrose
inversion, yet Figure 7 shows that this trend is not evident between the exhaust steam
temperatures recorded during the trial and the overall sucrose lost. The highest exhaust steam
temperature recorded was 128°C.

3.0

2.5
Overall sucrose lost (%)

2.0

1.5

1.0

0.5

0.0
116 118 120 122 124 126 128 130
-0.5
Exhaust steam temperature (°C)
-1.0
Figure 7. Overall sucrose lost vs. exhaust steam temperature

The fructose/glucose ratio under normal conditions should not change substantially across an
evaporator station, i.e. CJ (F/G) ≈ Syrup (F/G)), however, if there is destruction of glucose due
to possible Maillard reactions the fructose/glucose ratio will increase (Rein, 2007). Figure 8

398
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

shows that syrup (F/G) ratio was mostly approximately equivalent to the CJ (F/G) ratio during
the trial, indicating that there was limited destruction of glucose by Maillard reactions.
2.50
CJ (F/G)

2.00 Syrup (F/G)

1.50
F/G

1.00

0.50

0.00
0 20 40 60
Corresponding CJ and syrup samples

Figure 8. Corresponding CJ and syrup F/G ratios at GH during the survey period

Sucrose lost per effect

Previous studies (Davis, 1995 and Crebo, 1994) reported that it was difficult to determine small
sucrose losses across each effect because small changes in glucose and brix levels led to much
variation. The variation in sucrose losses is apparent in this study with large standard deviations
(SD) from the mean (see Figure 9). This makes it difficult to distinguish the difference in
sucrose lost between different scenarios. An LSD statistical analysis was used to evaluate the
results.

Figure 9 shows that the average sucrose losses (%) across the first effect vessels were below
0.3% at different CJ brix and flow levels. Statistical analysis of sucrose losses in the first effect
(Appendix – Table A 2) shows that the differences between sucrose lost under different CJ
brixes and flow levels were statistically insignificant at the 5% level of significance, i.e. p-
value >0.05. The statistical analysis outcomes of the second effect results (Appendix – Table
A 3) were similar to those of the first effect but the average sucrose losses were negligible in
all scenarios compared to the UDL observed.

The average sucrose losses in the third effect was also negligible at high CJ brixes, however, a
high average sucrose loss of approximately 0.81% was apparent at low CJ brixes (Figure 9). A
statistical analysis of the third effect sucrose losses (Appendix – Table A 4) showed that the
difference in sucrose lost between the low CJ brix and low flow scenario, and the high CJ brix
and low flow scenario, was statistically significant.

The average sucrose losses across the combined fourth/fifth effects at high CJ brix and low CJ
brix level were approximately 0.3% and 0.7%, respectively. However, the difference between

399
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

the high CJ brix and low CJ brix was found to be statically insignificant (Appendix – Table A
5).

In summary, the results show that the sucrose losses across the third effect were a major
contributor to the overall sucrose losses (at low CJ brix). The difference in sucrose lost between
low CJ brix and high CJ brix was only statistically significant in the third effect. However, the
sucrose losses, at low CJ brix and flow, in the combined fourth and fifth effects also appear to
be higher than the practical acceptable sucrose loss of 0.5%. Table 2 shows that the retention
times in the later effects were considerably longer at the low CJ brix scenario. This may have
promoted high sucrose losses in the later effects.

A B

C D

Figure 9. Average sucrose lost across individual effects at different flows and brixes.
…………A = 1st effect; B = 2nd effect; C = 3rd effect; D = 4th & 5th effect

Theoretical sucrose losses

The theoretical losses, calculated by using a Vukov model (Schäffler, 1994), underestimated
the sucrose losses (see Table 2). Purchase et al. (1987) stated that the theoretical losses can be
lower than measured amounts due to erroneous assumptions about the conditions in the
evaporator vessels used in the calculations e.g. juice levels used to estimate the retention times.
Tracer testing would be required to determine the juice retention times in the GH evaporator
station more accurately. For various scenarios modelled using the Vukov model, viz. low CJ
brix and low flow, high CJ brix and low flow, and high CJ brix and high flow, the total sucrose

400
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

losses amounted to 0.12%, 0.1% and 0.059%, respectively. The theoretical sucrose losses trend,
however, is similar to that of measured sucrose losses with higher sucrose losses at low CJ brix
and low flow scenario compared to high brix scenarios. The Vukov model did show a
substantial difference in sucrose lost between scenarios of low and high flows with a high brix
CJ. Again, this may be due to errors in assumptions made.

Table 2. Theoretical losses calculated using the SMRI Vukov model (Schäffler, 1994) with the
given measured brix profiles, evaporator surface area available and the estimated brix retention
time
Measured brix profile
Low brix-low High brix-low High brix- high
flow flow flow
CJ brix 6.93 11.29 11.29
1st effect brix 13.95 21.78 21.53
2nd effect brix 18.21 27.96 34.97
3rd effect brix 21.69 34.35 40.02
Syrup brix 65.57 65.65 65.65
Total evaporator heating surface area
st 2
1 effect area (m ) 5580 5580 5580
nd 2 *
2 effect area (m ) 930 930 2325
3rd effect (m2) 1395 1395 1395
Combined 4th-5th effect
(m2)* 2325 2325 2975
Estimated brix retention time
st
1 effect retention (mins) 3.33 3.32 2.21
nd
2 effect retention (mins) 5.55 5.45 3.64
3rd effect retention (mins) 10.50 10.53 7.02
Combined 4th-5th effect
(mins) 41.18 29.66 19.78
Theoretical losses using Vukov model
st
1 effect (%) 0.044 0.040 0.026
nd
2 effect (%) 0.024 0.020 0.013
rd
3 effect (%) 0.037 0.029 0.017
th th
4 -5 effect (%) 0.011 0.006 0.004
Overall sucrose losses (%) 0.120 0.100 0.059
-*at low CJ flow GH bypassed one vessel in the second effect and one vessel in the fourth effect.
Assumptions: Low=200m3/hr and high flow= 300 m3/hr. Juice level 1st effect=10% and 60% from 2nd effect.

Conclusions and Recommendations

The GH mill inversion loss survey, using SMRI-NIRS prediction equations, showed that higher
overall sucrose losses resulted when low brix CJ was processed at low flow rates. The sucrose
losses amounted to approximately 1.8% under this condition. From assumptions about the

401
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

frequency and duration of this condition, it was estimated that the evaporator inversion losses
accounted for a maximum of about 20% of the undetermined sucrose losses. It was therefore
deemed to be a major contributor to the undetermined losses at the mill but it appears to be not
the only major contributor. Sucrose losses appeared to be mainly occurring in the third to fifth
effects when the CJ brix was low.

It is acknowledged that the GH mill has to generate sufficient Vapour 1 bleed to operate the
raw house and refinery, however, it is recommended that GH should avoid substantially
diluting the CJ to supply the vapour bleed demands. One possible solution to this dilemma of
sucrose lost due to diluted CJ versus vapour bleed demands is to run water in one kestner only.
This would require the installation of isolation valves at the GH first effect so that the two
kestners can operate separately when required. Given the extent of the losses, it has been
recommended that senior GH mill management assess if the evaporator modifications can be
justified. It was approximated that a revenue of about R2 million was lost due to evaporator
inversion losses when the factory ran at a capacity of 60% or less. Losses will also occur at
higher throughputs, but at a lesser extent.

Further tests are recommended across individual effects to determine the actual retention time
by tracer testing and to confirm the results of the tests performed.

Acknowledgements

The authors would like to express their gratitude to the Gledhow staff that made these tests
possible viz. factory supervisors, mill laboratory staff and Marissa Munian. The SMRI
project team is also thanked.

REFERENCES

Crebo C (1994). Inversion Survey of the Gledhow evaporator. Sugar Milling Research Institute
Technical Note No. 33/94, 23 September, 7 pp.
Davis SB (1995). Inversion survey of the Noodsberg evaporator station. Sugar Milling Research
Institute Technical Note No. 6/95, 8 March, 9 pp.
Lionnet GR (1993). Some effects of the drought during the 1992/93 season. Proc S Afr Sug Technol
Ass: 209-216.
Purchase BS, Day-Lewis CMJ and Schäffler KJ (1987). A comparative study of sucrose degradation in
different evaporators. Proc S Afr Sug Technol Ass: 8-13.
Ramaru R and Madho S (2015). Standard Operating Procedure (SOP) - Determination of evaporator
inversion losses using NIRS analyses and SMRI inversion loss spreadsheet-toolkit. Sugar Milling
Research Institute Technical Note No. 12/15, 01 October, 11 pp.
Rein P (2007). Cane sugar engineering. Verlag Dr. Albert Bartens KG, Berlin: 420.
Schäffler KJ (1986). The magnitude of inversion losses in the Sezela evaporator. Sugar Milling
Research Institute Technical Report No. 1431, 29 January, 8 pp.
Schäffler KJ (1994). Example of a Vukov model for estimating inversion losses in Sezela evaporators.
Sugar Milling Research Institute Technical Note No. 2/94, 13 January, 3 pp.
Schäffler KJ, Muzzell, DJ and Schorn, PM (1985). An evaluation of sucrose inversion and
monosaccharide degradation across evaporation at Darnall mill. Proc S Afr Sug Technol Ass 60: 73
– 78.

402
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

Appendix: Statistical analyses

Table A1 - Table A 5 show the statistical analyses of the sucrose lost under different CJ brix
levels and CJ flow levels. The p-values of the conditions that are significantly different to each
other are underlined.

Table A 1. Average overall sucrose lost at different CJ brix range and CJ flow range,
with the statistical analysis p-values obtained by using Fisher’s Least Significant
Difference (LSD) test
Summary: Overall
Average
CJ Brix sucrose lost
CJ flow rates (%) No. of samples Stdev
High High 0.48 6 0.32
High Low 0.34 45 0.41
Low High 0
Low Low 1.77 7 0.67
All Groups 0.53 58 0.64
Statistical analysis (LSD test): Overall
p-value at
High brix; High brix; Low brix; Low brix;
specified
High flow rate Low flow rate High flow rate Low flow rate
conditions
High brix ;
0.44 0.00
High flow rate
High brix ;
0.44 0.00
Low flow rate
Low brix ;
High flow rate
Low brix ;
0.00 0.00
Low flow rate

403
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

Table A 2. Sucrose lost across GH 1st effect evaporators at different CJ brix level and
CJ flow level, with the statistical analysis p-values obtained by using LSD test
Summary: 1st effect
Average sucrose
CJ Brix Flow rate lost (%) No. of samples Stdev
High High 0.25 6 0.22
High Low 0.03 45 0.73
Low High 0
Low Low 0.32 7 0.43
All Groups 0.09 58 0.67
Statistical analysis (LSD test): 1st effect
p-value at
High brix; High brix; Low Low brix; Low brix;
specified
High flow {1} flow {2} High flow {3} Low flow {4}
conditions
High brix;
0.45 0.86
High flow {1}
High brix;
0.45 0.30
Low flow {2}
Low brix;
High flow {3}
Low brix;
0.86 0.30
Low flow {4}
Table A 3. Sucrose lost across GH 2nd effect evaporators at different CJ brix level and
CJ flow level, with the statistical analysis p-values obtained by using LSD test
Summary: 2nd effect
Average
CJ Brix Flow rate sucrose lost (%) No. of samples Stdev
High High -0.13 6 0.38
High Low 0.08 45 0.53
Low High 0
Low Low -0.05 7 0.75
All Groups 0.04 58 0.54
nd
Statistical analysis (LSD test): 2 effect
p-value at
High brix High High brix; Low Low brix; High Low brix; Low
specified
flow {1} flow {2} flow {3} flow {4}
conditions
High brix;
0.39 0.80
High flow {1}
High brix;
0.39 0.56
Low flow {2}
Low brix;
High flow {3}
Low brix;
0.80 0.56
Low flow {4}

404
Dairam N et al Proc S Afr Sug Technol Ass (2016) 89: 391-405

Table A 4. Sucrose lost across GH 3rd effect evaporators at different CJ brix level and
CJ flow level, with the statistical analysis p-values obtained by using LSD test
Summary: 3rd effect
Average
sucrose lost
CJ Brix Flow rate (%) N Stdev
High High 0.00 6 0.45
High Low -0.09 45 1.04
Low High 0
Low Low 0.81 7 1.28
All Groups 0.03 58 1.05
Statistical analysis (LSD test): 3rd effect
p-value at
High brix High High brix Low Low brix High Low brix Low
specified
flow {1} flow {2} flow {3} flow {4}
conditions
High brix;
0.84 0.16
High flow {1}
High brix;
0.84 0.04
Low flow {2}
Low brix;
High flow {3}
Low brix;
0.16 0.04
Low flow {4}

Table A 5. Sucrose lost across GH 4th and 5th effect evaporators at different CJ brix
level and CJ flow level, with the statistical analysis p-values obtained by using LSD test
Summary: 4th- 5th effect
Average sucrose
CJ Brix Flow rate lost (%) N Stdev
High High 0.36 6 0.57
High Low 0.32 45 1.18
Low High 0
Low Low 0.66 7 0.95
All Groups 0.37 58 1.10
Statistical analysis (LSD test): 4th and 5th effect
p-value at High brix Low brix
High brix Low Low brix High
specified High flow Low flow
flow {2} flow {3}
conditions {1} {4}
High brix;
0.94 0.63
High flow {1}
High brix;
0.94 0.46
Low flow {2}
Low brix;
High flow {3}
Low brix;
0.63 0.46
Low flow {4}

405