Proceedings of The South African Sugar Technologists' Association - June 1990

ENCRUSTATION AND SCALING IN CONTINUOUS SUGAR

VACUUM PANS
By P. W. REIN
Tongaat-Hulett Sugar Ltd, Glenashley

Abstract away causing choking ofthe crossover chutes between com-

partments,lodging oflumps in someof the tubes, and block-
Experience gained with encrustation and scaling occurring ing of some ofthe discharge valves, evident on drainingthe
in continuous vacuum pans is reviewed. This covers both pan. However, lumps were not noticedin the product sugar.
encrustation on internal pan surfaces and scaling of heating
surfaces. Practical results are compared with information Subsequently the cross-over port arrangement has been
published in the literature, enabling conclusions to be drawn changed from the Maidstone design. Instead of being in the
on the major mechanisms involved. Ways of minimising or downtake area where massecuite moves slowly, it is now
dealing with encrustation are given. directly above the calandria, where boiling massecuite keeps
the opening clear. Work done at Delft confirmed that en-
crustationonly occurs more than 100 mm above the boiling
Introduction massecuite level (Heffels et aI 3) .
In batch vacuum pans, the vessels are steamed out or It was found in Tongaat-Hulett pans that above-masse-
cleaned on a routine basis, as the pans are emptied every cuite encrustation can be eliminated entirely by the use of
few hours. In continuous pans however, the vessels operate fine sprays directing water onto exposed surfaces (Kruger"),
for long periods without emptying. Encrustation of sugarin Onlysmallamountsof waterare required. This is confirmed
the vessel becomes a concern under certain circumstances. by the fact that encrustation has beenobserved to have been
Indeed, fear of encrustation was one of the factors initially washed offbaffles directly below watersprays which are used
impeding acceptance and installation of continuous pans, periodically to clean light glasses or inspection port-holes.
particularly for high grade products. Experience withTongaat-Hulett pansat Felixton has shown
Although lowgradecontinuous vacuumpans may operate that any lumpsfound in A pans are usually friable, i.e.easily
on a continuous basis for long periods, it has been found broken up with the fingers, and any large lumps escaping
necessary to liquidate and clean high grade pans on a more through the massecuite outlet are normally broken up by a
frequent basis. This is due to two factors: positive displacement pump. Smaller lumps will however
• Unlesspreventive stepsare taken, sugarbuilds up on sur- get through the system, and if they go through the sugar
faces in the pan. This encrustation breaks off in pieces drier they will harden and end up as hard pieces. They can
which, on liquidating can block tubes or outlets. however be easily removed by scalping screens on the drier
outlets. Lumps have not been found to be a problem with
• Progressive scaling of the heat transfer surfaces occurs. Felixton A pansover a 2-3 week operating period,but lumps
This reduces the rate of heat transfer, requiring that cal- are found consistently from Felixton B pans (Montocchio").
andria pressures be increased. When the limit of this ad- These are large enough to be screened out at the end of the
justment is reached, cleaning of the heating surfaces is barometricsealboxand returnedto the remelter. The lumps
required if production rate is to be maintained. are usually flat pieces resembling encrusted massecuite bro-
Encrustation is clearly a function of purity of the mas- ken off pan internal surfaces. No lumps are observed from
secuite beingboiled. In C-massecuite continuous pansin raw the C pans.
sugar mills (± 52 purity) the pans can be operatedfor a full It is apparent that conditions favouring crystallisation in
season of about 38 weeks without having to stop and clean general also help to minimise encrustation, i.e. consistent
out the pan. In A-massecuite pans (± 85 purity), encrusta- mother liquorsupersaturation and optimum crystal content.
tion becomes severe after a few weeks and it is necessary to It has been found in Tongaat-Hulett Sugar mills that when
empty the pan and clean out the encrustation. continuous pans are operated under good control, ancillary
Encrustation can occur above the boiling massecuite or problems like encrustation are reduced.
on surfaces below the massecuite level. The mechanisms A different type of above-massecuite level fouling was re-
involved appear to be somewhat different. ported on a C pan in Australia (Pozetti and Sheedy"). Mas-
Continuouspans have nowbeenin operationin the South secuite which splashes onto hot incondensible outlet pipes
African sugar industry for 15 years. This paper relates ex- or steampipeswas reportedto formhard caramelised lumps.
perience gained in different types of continuous pans, on When these break off they cause havoc with centrifugal
different grades of massecuite. It outlines someofthe reasons screens. This was overcome by lagging the hot pipes inside
for the behaviour observed, and practices adopted to over- the pan.
come or cope with encrustation.
Mechanism of encrustation
Encrustation above the boiling level Some work on the factors affecting encrustation has been
done at Delft University of Technology. This work showed
(Heffels et al. 3) that encrustation wasa function of the crystal
Experience with continuous pans surface area/unit volume and the supersaturation in the
Experiments were undertaken at Maidstone Mill to in- mother liquor, both factors which affect the crystallisation
vestigate howlonga Tongaat-Hulett A-fuassecuite pan could rate. It was also affected by the rate of evaporation, which
be operated before boiling out was required. It was found influenced the rate at which splashes of massecuite attached
that four weeks was the limit of operation (Kruger"). If the to the wall, and the viscosity of the massecuite, i.e. the time
pan was operated for longer, lumps of encrustation broke spent by massecuite running down the encrusted surface.
204
Proceedings of The South African Sugar Technologists' Association - June 1990

The longer massecuite takes to travel down the surface, the the same reason, particularly in high grade massecuites. In
greater the potential for crystallisation. C massecuites, conductivity electrodes need to be removed
Heffels and de Jong" report that encrustation is enhanced almost daily for cleaning, if goodcontrol is to be maintained.
by higher crystalcontent and small crystal size.Encrustation With A-massecuites, conductivity probes need to be re-
starts with crystals getting stuck to the surface. Both more moved every few hours, which is not practical. The Duotrac
and smaller crystals will promote this. This suggests that RF probe was developedto overcome this problem (Radford
encrustation could be more severe in early compartments. et aI.1l); because it measures electrical properties at high fre-
quency, the effect of scaling of the electrodes has much less
Some reports on pans operating with high grade masse- effect, but still the RF probesneed to be removedand cleaned
cuites indicate that encrustation is more severe in particular regularly on A massecuites.
parts of the pan. In practice, it has never been possible to
establish in Tongaat-Hulett pans that encrustation occurs to Experience at Maidstone has shown that where the inter-
a greater extent in early or late compartments of the pan. connecting ports between compartments are below the boil-
In the firstfewcompartments, crystallisationrates are higher ing massecuite level, encrustationaround the openingoccurs.
(due to the lower impurity/water ratio in the mother liquor), This can be a restriction to massecuite flow, and Fives-Cail
while in the later compartments, the crystallisation rate is Babcock(FCB) pans in South Africahave been suppliedwith
lower, but the viscosity is increased. steam jiggers at the cross-over ports to overcome this.
The Delft work has shown (Heffels et al. 3) that the initial Encrustation and build-up can occur in regions of a con-
onset of encrustation is delayed if the surface is coated with tinuous pan where circulation is poor. This has been evident
a synthetic material which is hydrophobic. Laboratory tests in C pans which have operated for long periods without
showed that encrustation started on a steel plate after five cleaning. Hard deposits in the bottom of an FCBC pan have
minutes in a laboratory pan, but after 25 minutes on a teflon been seen at Maidstone after a full season's operation. The
plate. However, once encrustation was initiated, it occurred formation of hard caramelisedlumps was also reported from
at the same rate on either steel or teflon. They also showed an Australian C pan (Pozetti and Sheedy"), thought to be
that the roughness of steel has no effect on encrustation, i.e. due to poor circulation in the last compartment following
there is no advantage in the use of polished plate. shut-down and start up of the pan.
Experiments by Tongaat-Hulett Sugar indicated that syn- The influence of the designof the pan on lump formation
thetic surfaces such as teflon or epoxy coatings can have a is evident at CO Smith's Sezela factory. Two different types
beneficial effect. However,the method of attachment of such of continuous C pan are installed, but only one is generally
a surface caused considerable problems, and this approach used at anyone time. Lumps are regularly observed in the
does not eliminate the problem. Further, concern was ex- massecuite when one of the pans is in use, and are absent
pressedat the potential for acceleratedcorrosion in the space when the other is operating. .
above the boiling massecuite provided by a pinhole or some
other discontinuity in an epoxy coating. Scaling of Calandria Heating Surfaces
Encrustation below the massecuite surface The factor determining raw sugar A-pan operating time
can sometimes be the scaling up of the calandria heating
Below the surface of the massecuite, some encrustation surface, rather than occurrence of encrustation. Tests done
does occur. It is far less severe,and has never caused serious at Maidstone on the effect of time on heat transfer rate in
problems. In certain cases it has caused difficulty in extract- an A-massecuite pan are shown in Figure 1. There is a fair
ing massecuite samples through small valves in the bottom amount of'.scatter, but the general trend is to lower heat
of the pan. The Delft researchers (Kuijvenhoven') suggest transfer coefficients after longer periods of operation. Typ-
that good insulation of the pan will prevent below-masse- icallyat Maidstone, the calandria pressureimmediately after
cuite level encrustation on external surfaces, but that en- pan clean-out is 80 kPa abs., and at the end of the period
crustation will be promoted at sharp comers or discon- has risen to about 100 kPa. Since V1 pressure is about 130
tinuities, e.g. probe tips or the edges of submerged baffles. kPa, they could continue to operate for slightly longer pe-
Alternatively, heating the walls to a temperature just above riods, but have generally adopted a routine of cleaning out
saturation temperature can be employed (Heffels and de the continuous A-pan every two weeks. This coincideswith
Jong").
The Seafordcontinuous pan installed at Aarbergin Switz- 400 r-------------------,
erland on seed production contains a rotating paddle turning • •
in the massecuite. Initially a sharp-edgedpaddle constructed •
from angle iron sections was used. Encrustation on the pad- ,.... 350 -
dle was observed to occur after a few hours in beet high
grade massecuites. Subsequently the paddles have been
.
"E
u •
•
changed, eliminating sharp edges, and the encrustation is <, • •
markedly reduced. ~ 300 f-

Another example of encrustation at sharp discontinuities u

•
is given by Witte." In a beet white (BMA) continuous pan,
I-
:I: • •
serious encrustation occurred around the ends of tubes pro- 250 ... •
truding from the calandria tube sheet. This problem dis- • •
appeared when the tubes were recessed into the tube sheet
and weldedin to givea smooth inlet. The fact that this effect
200 '--"'--'---'----'---l.........- ' -..........-'--L.....----'---'--'-..l-..........-'--.......--l
has not occurred in tubular calandria pans on raw A mas- o 5 10 15 20
secuitesindicatesthat encrustation at discontinuitiesis more
serious with high purity or refined sugar massecuites, Oa~s after clean
FIGURE 1 Overall heat transfer coefficients measured on Maidstone
Scaling of conductivity electrodes or temperature ele- continuous A pan in relation to time since cleaning out
ments used on boiling point elevation control is severe for the pan

205
 Proceedings of The South African Sugar Technologists' Association - June 1990

the shut-down of their diffuser extraction line, so taking the the South African industry, this could also have explained
pan off-line does not present a problem in terms of why the problem was experienced at Felixton but never at
production. Maidstone. However, further observations have shown that
this severe scaling problem is not always relatedto the sand
At Felixton it has been established that the continuous A- content of the cane. The factthat the occurrence of the prob-
pans can be operated for at least three weeks without en- lem is associated with downtime following rain suggests that
crustation becoming problematical. Generally however it is the phenomenon is related to stale cane. In the past this
scaling of heat transfer surfaces which causes Felixton to phenomenon has been observed in A-batch pans at one or
boil out the pan on a two-week (or less) cycle. two factories only, but has beensufficient to slow down boil-
The scaling of heating surfaces has never been a concern ing even in batch A boilings. Its occurrence there has also
at Maidstone. At Felixton however, some strange and un- been ascribed to excessive cane delays.
predictable scaling effects have been experienced. In some In the season following the problems at Felixton, the SMRI
cases virtually no reduction in heat transfer rate has been regularly analysed encrusted material from the A pan tubes.
experienced over a two-week period. But over a period of Extensive tube fouling did not occurduringthis period, and
a few months in the 1986/87 season, heat transfergenerally the analytical results were indicative of normal conditions.
dropped sharply, sometimes stabilised, and in some cases However, the results tend to indicateselective accumulation
even increased again. in the scale of calcium and silica(Bachan and du BoW). This
During this time, very severe scaling of heat transfersur- severe scaling problem has not recurred.
faces wasexperienced, with the pan sometimes havingto be
emptied and boiledout afterjust a few days. This was found
to occurafter periodsof heavy rain. Measurements of evap- Implications for Continuous White Sugar Boiling
oration rate during this period are shown in Figure 2. It is unlikely that the scaling of heating surfaces willoccur
in white cane sugar refinery boilings. The sugar has been
30 through an intensive decolourisation and filtration process,
~

t\ ~ v>:
25 FX A1 Pan and the scale-forming compounds are absent.This is proven
by the operation of batch white pans at Hulett Refinery
20 which are not steamed out at all over a four-week period,
with no evidence of heating surface scaling.
.-.. 15
N

E
Likewise, it is considered highly unlikely that heating sur-
c 10 l- face scaling will be a significant factor in beet high grade

I I I
.I:
white boilings, because of the more intensive purification
<,

.=.
0\ 5

0
I-

I
t
I I
and filtration processes employed in the beet factory.
In both cases therefore, the factor determining the length
....10
OJ

FX A.2 Pan of operation of continuous pans between cleaning is likely

~
c- 25
to be encrustation. In the absence of any measures taken to
V
e

=\ \ '\
....
0
.... 20 avoid encrustation, it is likely that periods between cleaning
10
c-
0 IS
will be similar to those experienced on raw sugar A-grade
a.
10
boilings, i.e. of the order of two to four weeks. Optimum
>
ILl 10 l- operation of continuous white pans should result in high
5
o
'-
II t tI I
f I
crystal contents, and with high crystallisation rates, rela-
tively low supersaturation in the mother liquor and lower
mother liquor viscosity. This may be countered from an
o 5 10 15 19 encrustation point of viewby the higher crystallisation rates,
Time(days) so that the net effect is likely to be a similar encrustation
FIGURE 2 EVlIPoration rates measured on Felixton continuous A rate to that experienced by raw A-pans. This pre-supposes
pans during time of severe tube scaling. Arrows indicate however, that conditions in each and everycompartmentof
times when pans were emptied and cleaned. the pan are well controlled.
Analysis of the scale inside the tubes when this first oc- In practice, operating periods between cleaning on white
curred showed the scale to contain 41 % silica. This led to pans have beenvariously reportedas being between one and
the conclusion that the excessive sand accompanying cane four weeks.
after rain was implicated in the scaling process. However,
subsequent analyses of scale after other severe scaling con- Methods of reducing or coping with encrustation
ditions gave a very high loss of weight on ignition, i.e. the
scale was mainly carbonaceous, originating from carbohy- Use of water orfeed syrup sprays
drate material. Subsequent scale analyses confirmed this
finding (Bachan and du Boil'), The scale was always found Encrustation above the massecuite level can be overcome
to be easily removed by a water wash. almost entirely by the periodic use of water sprays directing
Attempts to establish the cause of this serious scaling have water onto the exposed surfaces (Kruger"), If the sprays are
been in vain. Since there are two identical continuous A- activated by a timer the amount of water added can be a
pans at Felixton, it was possible to run different operating negligibly small proportion of the total evaporation load.
conditions on the two pans during the scaling periods. This An alternative which has been adopted in some contin-
established conclusively that the scaling is not caused by uous pans is to direct the syrup feed onto the exposed sur-
changes in operating conditions, but that it is solely de- faces above the massecuite level. This can also be effective
pendent on certainconstituents of the canejuice,which have but suffers from the drawback that the lower density feed
not yet been identified. does not get rapidly mixed in with the massecuite and can
Initial theories linked scaling to excessive sand in juice. be the cause offalse grain formation (Kuijvenhoven'), It is
Since Felixton contains the highest average ash % cane in generally better to add the feed in an area where intense
206
Proceedings of The South African Sugar Technologists' Association - June 1990

mixing takes place, e.g. immediately above a stirrer or di- Table 1

rectly below the bottom tube plate. Time taken to empty and clean a 120 m' A pan at Felixton
An alternative approach is not to lag the top of the con-
tinuous pan. Then some of the vapour condenses on the side Before modification After modification
walls and runs down the walls with the same effect as the Time Time
water sprays. However this does not ensure that the dividing Closing steam and Closing steam and
plates between compartments get washed. breaking vacuum 5 min breaking vacuum 5 min
Liquidating 45 min Liquidating 30 min
Raisingof massecuite boiling level Steamingout 10 min Steamingout
Feeding water 60 min Feeding water
Experience with the Seaford pan at Aarberg has shown Boiling water 180 min Steamingpan 120 min
that encrustation above the boiling massecuite can be re- Draining to Draining to
A-molasses 15 min A-molasses
moved by raising the boiling level. This is an application of Cutting over 60 min Cutting over 60 min
the findings at Delft that the action of the massecuite at the
boiling level washes the surfaces clean. Total ± 61/4 hours Total ± 3'12 hours

Vibrating plates or membranes

A suggestion has been made by Austmeyer' that thin plates
or membranes could be fitted to areas susceptible to en-
crustation. These membranes vibrate from the pressure fluc-
Needfor good control
tuations caused by the flow of the massecuite, thus cracking Since encrustation is largely affected by the condition of
off encrustations. It is suggested that membranes can be fit- the massecuite splashing up onto the surrounding surfaces,
ted to walls and baffles as well as to stirrer blades. A similar it can be significantly reduced by maintaining good crystal-
result could perhaps be achieved by using flexible polyeth- lisation conditions in each compartment. If a high crystal
ylene baffles. content and a relatively low supersaturation can be main-
tained in the compartment, the encrustation which occurs
above that compartment must be lower than in cases where
Positioning of crossover ports either of these important variables strays away from opti-
If crossover ports are situated below the massecuite boil- mum values. In this respect it must be an advantage to be
ing level they are subject to encrustation. Because the area controlling the massecuite condition in every compartment
immediately above the boiling massecuite is not susceptible as is done in Tongaat-Hulett pans (Rein "), rather than reg-
to encrustation, the positioning of crossover ports imme- ulating only the massecuite density at the exit of the pan
diately above the calandria ensures that they are always kept and making manual adjustments to the feeds to individual
clean. compartments. The latter system does not automatically
compensate for changes in heat transfer or crystallisation
Bypassing of compartments rates in individual compartments.
The BMA continuous pan which consists of four com- If control in any compartment is inadequate and false
partments, one above the other, has in some cases been fitted grain forms (nucleation), the small crystals resulting can also
with additional piping such that any single compartment can promote above-massecuite encrustation (Heffels and de
be bypassed on the run. During the time that one of the Jong").
compartments is boiled out and cleaned, production capac-
ity would generally be reduced to 3/4 of normal capacity. A
further disadvantage however is that crystal size distribution Pan circulation patterns
suffers as the reduction in number of compartments from In order to reduce below-massecuite encrustation to a low
four to three is severe. level, it is important to ensure that the design of the pan
An alternative is to use two smaller pans on high grade ensures good circulation and particularly that there are no
duty, or to divide a pan into two separate sections, so that dead zones where crystal can settle. Care needs to be taken
one half can be liquidated at a time, to reduce the effect on also to ensure that on pan start-up circulation starts quickly,
production rate. or else caramelization of massecuite in contact with hot cal-
andria surfaces can occur.
Arrangements for easy draining and cleaning
Disruption to production can be minimised by giving spe- Elimination ofdiscontinuities
cial thought to all the individual activities involved in emp- In order to reduce below-massecuite level encrustation, it
tying, cleaning and re-filling. . is important to minimise any sharp edges in contact with
Since the occurrence of the severe scaling problems at the massecuite, particularly where high grade massecuites
Felixton, attention has been given by the Mill staff to re- are being boiled. It is a well established fact that encrustation
ducing the time required to empty and clean out a pan. Thus, will occur at any sharp discontinuities.
even if severe scaling occurs during a high throughput period, Since most of the measurement probes employed will en-
disruption to production will be minimised. crust for the same reason is it desirable to have them easily
A breakdown of the activities involved in emptying and removable for periodic cleaning.
cleaning out a 120' m' A-pan before and after this investi-
gation (Montocchio') are given in Table 1. Heating ofsurfaces below massecuite level
An improvement was achieved by changes to liquidating Some pans have also been built with heating jackets around
procedures, but the major improvement was due to a change the areas containing the massecuite .In certain cases the heat-
from boiling water in the pan to merely blowing vapour into ing jackets have been dispensed with and, instead, the sur-
the pan. This is done with the drain valves open, so that all faces in contact with massecuite have been well insulated to
liquid drains as it forms into the molasses tank. prevent cooling.

207
Proceedings of The South African Sugar Technologists' Association - June 1990

Conclusions 3. Heffels, SK, Kuijvenhoven U and de Jong, EJ (1984). Incrustation

abovethe level in sucrose crystallisers. Process Technol Proc2:217-222.
It has been shown that encrustation problems are largely 4. Heffels, SK and de Jong, EJ (1988). Incrustation in sucrose-water so-
dependent on massecuite purity, and really onlybecome sig- lutions. Zuckerind 113 (10): 873-877.
5. Kruger, GPN (1983). Noteson observations on the continuous A-pan.
nificant in high grade boilings. Tongaat- Hulett Sugar Internal Report, 12/8/83.
Some understanding of the factors influencing encrusta- 6. Kruger, GPN (I 983).Continuous A-pan boiling at Maidstone sugarfac-
tion has been obtained. This has influenced the design and tory. Proc S Air Sug Techno! Ass 57: 46.
7. Kuijvenhoven, U (1983). Aspects of continuous sucrose crystallisa-
operation of continuous pansin orderto minimise theeffects tion. WTHO No. 156, Delft University of Technology.
of encrustation. However it is still not possible to predict a 8. Montocchio, G (1988). Continuous boiling at Felixton. Comm 3eCongr
priori the extent of encrustation for a given situation. Assoc Reun Dev Tech Agric Suer 198-207.
9. Pozetti, C and Sheedy, BB (1989). Improving the performance of 120
m' continuous lowgradepan at Farleigh. Proc AustSoc SugCane Tech-
no!ll: 168-174.
10. Radford OJ, Tayfield, OJ and Cox, MGS(1988). Furtherdevelopments
REFERENCES in automated white pan boiling using radio frequency control. Proc Sug
Ind Technol47: 90-107. .
I. Austmeyer, KE (1988). Somethoughts on further development of su- II. Rein, PW(1987).A review-of experience withcontinuous vacuum pans
crose crystallisation. Zuckerind 113 (5): 389-397. in Tongaat-Hulett Sugar. Int Sug J 89 (1058): 28-34.
2. Bachan., L and Morel du Boil, PG (1989). Felixton pan tube fouling. 12. Witte, G (l988).Further experience with continuous crystallisation at
SMRI Tech. Note No. 6/89. Wabem sugarfactory. Zuckerind I i3 (5): 414-420.

208