ROBUST DESIGN OF AN EVAPORATOR STATION AS APPLIED TO

THE XINAVANE REHABILITATION PROJECT

DJ LOVE'. DM MEADOWS' AND RG HOEKSTRA2

'Dnvid Love Process Engirzeeritig cc, 45 Kelvin Plcice, Durbun, 4051, South Africn
'Torgnat-Hulett Sugar Ltd., Private Bug 3, Glerzaslzlej, 4022, Soutlz Africa

Abstract shredder, a new 6 m wide diffuses, modification of the evap-

orator station. a new 42 m' batch A-pan, a new 43 m' con-
The rehabilitation ol' thc Xinavane sugar factory in
tinuous C-pan, two new 120 m' vertical crystallisers, two
Mozambique necessitated a redesign of the evaporator sta-
new 1 500 kg batch centrifugals and four new 1 300 mrti
tion. The particular arrangement at Xinavane proved to be
continuous centrii'ugals, as well as the refurbishment of most
extremely sensitive to changes i n operating parametcrs.
of the existing plant. The boiling house would be arranged to
Using this evaporator station as an example, the concept of
produce VHP sugar, using a threeboiling system with grain-
design robustness (suitability under a range of operating
ing of A-pans and the usc of C-magma for B-seed. In addi-
conditions) is introduccd. A rigorous evaporator simulation
tion, the process of recycle of clarifier mud to the diffuser
program was used to optimise heating surface and tempera-
was selected, obviating the need for refurbishment of the fil-
ture driving force distribution along the evaporator train, and
tcr station or ancillaries.
to model and minimise the potentially considerable effect on
evaporator performance of variations in bleed rate. The While various areas of the proccss design presented interest-
unique requirements of evaporator station design are dis- ing challenges, one of the more absorbing came from an
cussed, and sorne of the details of the equipment designed unexpected quarter - the evaporator station. Although evap-
for Xinavane are presented. orator design is ostensibly a routine exercise, the specifics of
the Xinavane dcsign forced consideration of the subtleties of
evaporator dcsign optimisation.
Introduction
A prelimina~-yevaluation of the evaporator station indicated
Recent design work for rcl'urbishmcnt of the Xinavane fac-
that, irrespcetive of the condition of the vessels, the existing
tory required thc redesign of the evaporator station. station design was not appropriate for the expected duty i n
Preliminary calculations highlighted the fact that some the refurbished factory. In particular, thc first effect area was
options for modifications would meet the new requirements
clearly too small and the distribution ol' heating surface
for the selected design paramcters, but the performance between the vessels in the tail was far from optimum. This
would deteriorate rapidly if the design parameters were discrepancy can probably be explained by the following
slightly different. This highlighted the need for a robust
changes, which are part of the refurbishment and signifi-
evaporator design which was not as sensitive to design cantly affect vapour blecd requirements:
parameters.
thc elimination of the usc of exhaust steam for boiling
pans (previously available for use on two calandria pans)
Refurbishment of the Xinavane factory the scrapping ol' old coil pans which used stcam at 400
Xinavane sugar mill (known previously as Inkomati mill) is kPa(g)
located i n southern Mozambique some 110 km from a changed boiling scheme whcl-c B-sugar is fully remelt-
Maputo. In 1998 Tongaat-Hulett Sugar purchased a share i n ed rather than being bagged
the mill, with the remaining share being held by the higher levels of imbibition (selected to achieve higher
Mozambican government. At its previous production peak, extraclion).
in the 197 1 season. the mill crushed 483 147 tons of cane. A further faclor is that the exisling evaporator station was
but the civil war had a devastating effect on production, and modified during the civil war ycars, with the modifications
throughput i n the 1998 season was a mere 97 745 tons. defined by vesscls that were available from other non-oper-
The Technical Managcment Dcpartment ot Tongaat-Hulett ational mills.
Sugar was appointed to carry out the design and detailed A detailcd investigation into the dcsign of the evaporator sta-
engineering for the rehabilitation of thc mill to a nominal tion was clearly requil-cd. It must be emphasised that evapo-
throughput of 120 tons canc per hour, with a view to a pos- ~.;ltorstation design cannot bc conductcd independently but
sible future doubling of capacity. The scope of the rehabili- is very closcly linked to overall factory steam demand and
tation included upgrading of boilers, a new heavy duty the requirements 1'0s fuel cconomy.

Proc S Afr Siig Teclulol Ass ( 1999) 73 21 1

 Requirements of evaporator station design boilers)
IOW HP stcam prcssurc (2 l 0 0 kPa(a))
On first consideration, it may seem that the requirement of
an cvaporator station is simply to remove water fromjuicc to lowcl- efficiency turbines
producc syrup. However, Inore dctailccl consideration will an unrcliablc clcctricity supply I'rom the national grid
identify a significant number ol' rcquircmcnts and con- a market Sor generation of clcctrical power (I'or irrigation)
straints. Good evapol-atol- station design involvcs balancing thc usc of local trees as the supplementary I'uel source.
conflicting rcquil-cnicnts and constraints while attempting to
approach each of them as closcly as possible. Unfortunately, bccausc of thc decline in performance of the
mill during thc civil war in Mozamhicluc, Ihcrc is limited
The major rcquircmcnts arc: qu~llitydata on plant pcrl'ormancc ancl many 'best estimates'
Evaporate the required quantity of watcr from clear juice havc bccn necessary as the basis for calculations of expect-
to producc syrup at the appropriate brix for pan boiling. ed pcrformancc.
Condense all exhaust steam horn the turbines (both primc
Taking thcse I'aclors into account, steam balances for a range
movers and turbo altcrnato~-S),avoiding blow-01'1' uncl o f cxpcctcd operating conditions highlighted the following
allowing a sul'l'icient let-down from HP to exhaus1 (to points spccil'ic to Xinavanc.
allow exhaust steam prcssurc contl-01).
Have an efficiency dcsigncd in conjunction with the rest When opcrat~ngw ~ t ha cluudruplc el'l'ect evaporator (V 1
of the l'actory to eliminate (or at least minimise) cithcr bleed only):
unwanted bagassc sul-pluses or the need [or supplemen- supplcmcntary I'ucl woultl bc rcquircd cvcn during the
tary fuel (i.e. achicvc a 'I'ucl balance'). high I'ibrc portion ol. rhc scason
Supply vapour bleeds at thc rcquircd quantities ancl prcs- the export ol'powcr would bc possihlc without the need to
surcs to meet the proccss demands of other sections of the blow off cxhaust s t c ~ ~ m
I'actory. a reasonable quantily ol' H P to exhaust let-down would
Supply thc rccluircd quality and cluantity of boiler feed hcilitatc cxhaust steam prcssurc contl-ol.
water, primarily l'rom condensed exhaust steam, but sup- When operating with a quintuple cvaporator (with V2 blccd
plemented with other acceptable condcnsatcs. for ~ ~ ~ ' I ' L I s c I - and primary juice heating):
Act as a reactor Ihr thc destruction of ?larch whcn su~tably thcrc would be a fuel shortage only during the low I'ibrc
dosed with the appropriate cnLymc (not necessary in dif- 1mrtion ol'thc season, with a bagassc surplus during high
fuser factories). 1.i hrc periods
These must bc achieved subject to the following constraints: thc maximum gcncration of clcctricity Ior export would
The evaporator dcsign must I'acilitatc stable operation and rcsulr i n cxhaust steam blow-01.1'
control. during pcl-iods of low fibre throughput, thcrc would bc
The thermal degradation ol' sucrose and reducing sugars vcry lirtlc HP slcam to Ict down to cxhaust which could
must bc minimised. I-csult In unstable cxhaust steam p~cssurccontrol.
The capital cost must bc minimised - a major factor i n Thc dil'l'iculty in dcciding between thcsc two options is com-
achieving this being thc optimum distribution ol' heating pounticd by the possibility tlia1 the I'ucl shortage could be
surfacc between cffects. eliminated by an cxpansion in the ncar Suture, doubling lhe
The dcsign should be robust, i.c. it should continue to pcr- capacity ol' thc mill. Preliminary calculations indicated that
form acceptably ovcr a reasonable range of operating con- thc lowcr spccil'ic stcam demand 01. u largcr l'actory would
ditions. cl iminatc the ncccl for supplerncntary I'ucl even when opcr-
The design should be conipatiblc with I'uture cxpansion ating ;I cluadruplc cvaporator. The added cxpcnsc o f cl-cating
plans. a cjui~it~~plc el'l'ect evaporator as part of the rc1u1-bishnient
would thcn be particularly dil'l'icult to justify, unless the
design was I'ully compatible with thc cxpansion.
Requirements of the Xinavane evaporator station
Given thcsc unccrtuintics, i t was ncccssary to generate alter-
The first stcp in clctcrniining the requiremcnts of the native designs Sol- refurbishment 01- thc cxisling evaporator
Xinavanc evaporator station was to pcllhl-m detailed stcam station at Xinavanc us cithcr a cluadruplc 01- quintuple effect
balances ovcr the entire factory. This was done using a con)- evaporation station. To cvaluatc the s~~itability ol' both ol'
putcr program called SLOB (Steam Load Overall Balance) thcsc dcsigns for future cxpansion. i t was also necessary to
which was developed in-housc by Tongaat Hulctt Sugar dcsign cvapol-ator stations lbr the cxpa~ision.This includes
(Rein and Hockstl-a, 1994). These calculations were able to I'urthcr possibilities sincc the bcncl'its ol' generating export
take into account I'actors spccific to Xinuvane that make it power I'or irrigation would rcclui~-ca cluintuple evaporator if
significantly different from the average South African sugar a condensing turbine proved economic o r a quadruplc cvap-
factory, viz: orator il' the condensing turbine was not an economic
a low cane crush rate (avcragc ol' 120 tons canelh) option.
lowcl- efficiency boilers (2 x 30 tonlh Dutch oven type

Proc S Afi. SLIS Tecll~lolAss (1 999) 73

Robust design oj'arl e ~ ~ z l ~ o r nstotioll
t o r as applierl to Xinavarle

Design of an evaporator station cient is directly equivalent to a 10% loss of heating surface
area.
Tongaat-Hulctt Sugar has for many years used a computer
simulation program (known as Program for Evaporator The evaporation rate W in kg/s can be calculated from the
Simulation and Testing, or PEST for short) for the design heat transferl-ed as:
and evaluation ol' evaporator stations. This program was
developed in-house (Hoekstra, 198 1) and provides detailed
calculations which avoid many of the simplifying assump-
tions of evaporator calculations used i n standard sugal- texts
describing evaporator calculations (Hugot, 1986). While where h is the enthalpy of evaporation in J/kg.
PEST or other such programs will handle the details of the This assumes that heat losses arc negligible and that there is
calculations, i t is important for the evaporator designer to no sub-cooling of condensate below its saturated tempera-
understand the principles of evaporator operation and the ture.
underlying calculations to use the program effectively.
Elucidating these principles does, however, require using
many simplifying assumptions. The three principles of multiple-effect evaporation original-
A simple quadruple evaporator station (with vapour one ly espoused by Rillicux (the pioneer of this technique) are
bleed only) is shown in Figure 1 , as the basis for the follow- (Spencer and Mcade, 1945):
ing discussion. First principle In a multiple-efl'ect evaporator, for each
kilogram of steam used. as many kilograms
of evaporation will result as there are units
i n the sct.
Second principle If vapours are withdrawn from any unit of
;I multiple-effect evaporator to replace
steam in a concurrent process, the saving
of steani will be equal to the amount of
vapour so used divided by the number of
units in the set and multiplied by (he
scquencc position of the unit from which
I 1 I At2 4I At, 4( At4 4 the vapour has been withdrawn.

I AT,
I-
_I (
AT,
AT.,
'
-I
Effective
Temperature
Differences
Third principle In any apparatus in which steam or vapour
is condensed, i t is necessary continuously
to withdraw the accumulation of non-con-
densable gas which is unavoidably left in
Figure 1. Representation of quadruple evaporator station with V1 the heating sul-face compartment.
bleed.
The first and second principles arc useful approximations
which relate to the el'ficiency oi' evaporation but ignore the
F~rridari~erltc~ls
oflletzt trtlrlsfer- effects of juice flashing and variations in cnthalpy of vapor-
isation (latent heat) with temperature. The nomenclature
The fundan~entalheat transfer ecluation which describes how
used in Figure 1 is based on the assumption that these prin-
heat is transferred across a heating surface is:
ciplec arc Isue.
Q=U.A.AT
Distribiltiorl of'teri~perot~rre
drivirlg,force over evaporators
where Q is the heat transferred i n W
The temperature driving force available for achieving the
U is the heat transfer cocf'ficient in W/m2/K required evaporation is the difference betwecn the saturated
A is the hcating surface area in m? temperature of the exhaust steam and that of the vapour in
the final effect. Thc exhaust steam pressure is normally
sclected as 200 kPa(a) (being a compromise between the
and AT is the temperature driving force in K.
requirements Ibr power gcneration and evaporator heating
In the case ol'an evaporator, AT is the difference between the surfacc recluirements). Smith and Tay lor (l 98 l ) have shown
saturated temperatuse of the steani in the calandria and boil- that the optimum pressure Ihr i'inal cl'fect vapour is between
ing temperature of the liquid. The convention used in this 16 and 20 kPa(a). This total available temperature difference
work is to calculate A and U based on the outside diameter will be reduced by thc elevation oi'hoiling point of the juice
of the evaporatol- tubes and thc distance between the tube in each effect as a consequence of concentration and hydro-
plates. This formula clearly shows the equivalent inlluence static head. The net or cffective temperature difference will
of the heat transfer cocfl'icient, U , and the heating surface then be distributed between the etfects as per the heat trans-
arca, A, on hcnt tsansfcr. A 10% drop in heat transfer coeffi- fcr ey uation.

Proc S Afr Silg Teclzrlol Ass (1999) 73

Kob~lstdesigrl of nil e\.trporrrtor- ~tntiorros rrl>plied to Xiilcrvai~e DJ Love, D M Meadows & RC; Hoekstrn

Effect of q~lantityo j vrrl7orrr hleed or1 evrrl7orrrtor .strrtiorl Since heat transfel- coel'ficients not-mally decrease towards
capacity the last cl'l'ect, causing temperature diff'erences to increase,
Detailed evaporator calculations on many practical evapora- the hcating surface area will need to increase down thc tail
tor stations in sugar factories have shown that incrcasing the for effective use of installed arca. On this basis existing
quantity of vapour bleed will increase the evaporation capac- cvaporator stations with equally sized vessels in the tail indi-
ity or the evaporator station. This increase in capacity will be catc a design that is not optimal.
associated with a decrease i n blccd pressure. This phenome-
non is not a natural consequence of multiple effect evapora-
Design for the refurbishn~entof Xinavane
tor stations but only occurs when the effects before the bleed
have a higher combination of area and heat transfer cocffi- evaporator station
cient than the effects further down the evaporation train.
Existirlg rllcrporrrtor stcitiorl
Appendix I givcs a derivation of this principle for a quadru-
ple cvaporator with vapour one bleed only. Simply stated, if The existing quadruple cl'l'cct evaporator consists of the fol-
I
we consider K, to be the resistance ol'an cvaporator effect lowing vessels:
to heat transfer. an increase in bleed will increase the cvapo-
rator capacity if the resistance of the first effect is less than Eflect Ty pc Area ( m 2
the average resistance of the effects in the tail. First Semi-Kcstncl- 1 523
Second Robcrt 703
Effect of corrde~~snteflash
on evaporator statioil ccyxrcitj,
Third Robert 402
In contrast to the effect of vapour bleed on capacity, the
Fourt h Robcr~ 877
return of condensate flash into vapour streams will normally
increase vapour pressures, and therefore reduce capacity but
improve steatn utilisation efficiency. This effect is, however, Preliminary calculations showed this station to have signifi-
usually small. cantly less than the required capacity. In particular the first
effect area was too small to provide an adequate bleed pres-
The use of vapo~rrtl~i.ottlirlg sure and the distribution of heating surlhcc between the
Vapour throttling (nor~nallyused after the last vapour blccd) cffccts i n the tail was t'ar from optimal. The heat transfer
can be used as a control variable to reduce evaporator capac- coefi'icients (listed below i n Table I ) used i n these calcula-
ity without affecting evaporator efficiency. This will reduce tions, and those for 311 modified configurations are based on
capacity while increasing bleed pressures as opposed to the cxtcnsivc mcasurcmcnts by Tongaat-Hulctt Sugar over many
alternative of rcducing capacity by reducing exhaust steam years and represent practically attainable design figures.
pressure (which will reduce bleed pressures). Throttling can
bc thought of as consuming driving forcc hut not stcam. Tahle l . Heat transfer coefficients used in evaporator
simulations.
Heat transfer c o e f f ~ c ~ e n(kWImYKj
t
Effect
Vessel arca prior to the vapour blccds should be installed to Quadruple effect Qu~ntupleeffect

provide the required blccd prcssurcs. Excess arca installed F~rst

here will result in bleed pressures above the required mini- Second
mum values and, although this will increase the cvaporator Thlrd
capacity (by driving the tail harder). the extra arca would Fourth
have been more effective had i t been installed in the tail.
For effective distribution of heating surlace area betwccn the
effects in the cvaporator tail (i.e. alicr the vapour blccd or
bleeds) Buczolich and Zadori (1963) provtde guidelines 1'0s The design for the rcfurbish~nentof' the quadruple cvapora-
the optimum distribution of hcating surfacc, as summarised tor was invcstigatcd i n detail on the basis of either replacing
by Hoekstra (l 98 l). the i'irst cl'l'cct vcsscl with a new Kestner type evaporator or
The criterion f'or optimum distribution of heating surface is converting the existing vcsscl into an (cxpandctl) Kestner.
that the ratio ol' healing surl'acc to cl'fcctivc tempcl-aturedriv- The cxisting Robcrt vessels wcrc to bc rctaincd.
ing lhrcc should I-emainconstant for each effect. The estimation of vapour bleed quantities is dependent on a
Expressed mathcrnaticall y: number of operating practices and without established oper-
ating norms, a range ol' assumptions are ~i~cessary. Given
this uncertainty in the quantity ol' vapour bleed requir-ed, a
!LzC
At;
number of simulations were undertaken to estimate the first
cll'cct arca that would be required as a junction of vapour
where i ~-eprcsclltseach cl'l'ect and C is an arbitrary constant, blccd (V1 bleed only) varying between 25% and 35% on
termed here the arca efficiency criterion. clear juice flow, the best estimate being 3 1%. This was done
Sot- both Ihc cxisting coni'iguration ol'vessels in Ihe evapora-

Proc S A,fr Sug Teclri~olAss (1999) 73

Rob~lstde.sigtt of a11evaporcltor station as applied to Xitral~nrle DJ Love, DM Mertclows & RG Hoekstra

Table 2. Comparison of first effect area requirements for original and swopped sequences of evaporator vessels in the tail.

V1 bleed Original sequence of vessels Swopped sequence of vessels

quantity Area (m2):(as below): 703 : 402 : 877 Area (m2):(as below): 402 : 703 : 877
(% CJ) 1"' effect V1 bleed V1 throt. 1'' effect V1 bleed V1 throt.
area pressure A P area pressure AP
(m (kPa(a)) (k Pa) (m (kPa(a)) (kPa)
25 4 098 173,8 0,o 2 500 162,4 0,o

27 2 929 165,4 0,o 2 054 154,9 0,o

28,3 * ------ ------ ------ 1 850 150,O 0,o

29 2 316 157,3 0,o 1 875 150,O 2,s

30,9 + 1 957 150,O 0,o ------ ------ ------

31 1 962 150,O 0,5 l 950 150,O 9,7

33 2 037 150,O 7,8 2 026 150,O 16,5

35 2 114 150,O 14,8 2 103 150,O 23,l

Minimum first effect area for swapped sequence
+ Minimum first effect area for original sequence

tor tail and thac with thc duties of tlic second and third vcs- 3 gives details of thc cvaporator siniulations which require
sels swopped
- around. A minimum (V I ) bleed pressure of' 1.50 thc minimum first cfl'ccl arca.
kPa(a) was spccii.ied to provide for the requrremcnts of thc
pan floor. A summary of thcsc sin~ulationsis prcscntcd in Thc area efficicncy critcrion shows that the distribution of
Tablc 2 and Figure 2. thc area down the evaporator tail is not ideal with the pres-
ent arl-angcrnent of vcssels. Thc range ol'the area efficiency
TO dcmonstrate thc use of thc area efficicncy criterion, Table criterion in the tail is reduced from 8 1 to 36 by swopping the
duties of the sccond and third cl'l'cct vesscls, a significant
4500 improvemcnt.
The full benefit 01' an effcctivc distribution 01' heating sur-
facc in the tail might no1 have bccn evident if the sensitivity
to the design assumptions had no1 becn investigated. The
rcsults show clearly how thc rcquired I'irst effccl area is
p . ... .. dcpendcnt on thc design assumptions. In particular, if the
bleed ratc had bccn assumcd to be 31% of the clear juice
flow, a I'irst ell'ccl area ol' approximately 1 960 m? is
required, regal-dlcss of' whcthcr the scquencc of vessels in
the tails is left as at present or sltercd as indicated by the area
eff'icicncy crilcrion. The benefit ol' the swopped sequence in
the tail is however vcry clear when lowcr blecd rates are
considered. A design that is actequate for bleed rates as low
$ as 2.5%' on clcar juicc llow would be significantly more
20 cxpensivc if thc second and third vcssels were not swopped,
as it would recluirc approximately 1 600 m? additio~ialfirst
eflkct al-cil. Interprcling this diSferently. a design without
swopped sccond and third vcsscls that was adecluate at a
g blccd rate ol' 31'2' on clear juicc flow would be seriously
a
e under capacity at lower hlccd rates. Vapour blow-ol'f to
atmospl~erecould address this issue. but at the expense of
i.uel cfficicncy and with the loss ol'good quality condensale.
2
n While vapour blow-off might bc a uscful operational tech-
nique in an emergency, i t is unwisc to include i t as a neccs-
sary rcquircrnent during design.
An intcrcsting aspect of thc graphical pl-cscntation is that for
25% 26% 27% 28% 29% 30% 31% 32% 33% 34% 35% both cascs invcstigalcd (i.c. swopped and original sequences
v1 Bleed % Clear Juice of second and third cfl'cct vesscls) the graph shows two dis-
Figure 2. Effect of V1 bleed ?hclear juice on first effect area. linct scclions. At lower blccd lutes first effect area must be

Proc S A,fr Siig Techrrol Ass (1999) 73 215

R o b ~ ~desigtl
st oj'cl~reLjaporcltor station as c~ppliedto Xirlavane DJ Lol-e, DM Meadows & R(; Hoekstra

Table 3. Effectiveness of area distribution in evaporator tail with original and swopped sequences of vessels.

Evaporator Performance with Minimum First Effect Area 1

Arrangement Vessel characteristic First Second Third Fourth
of vessels effect effect effect effect
in tail
Area (m 2, 1957 703 402 877
Area efficiency (m2/K) 256 106 25 41
Original Heat transfer resistance (K/MW) 0,20 0,65 1,46 1,63
Sequence Ave. heat transfer resistance of tail (K/MW) 1,251
Area (m ') 1850 402 703 877
Area efficiency (m2/K) 241 33 69 40
Swopped Heat transfer resistance (KIMW) 0,22 1,13 0,84 1,63
Sequence Ave. heat transfer resistance of tail (K/MW) 1,20

installed to provide cxtra evaporation capacity, whilc at capacity can be gainfully used by increasing imbibition on
higher bleed rates, cxtra area must bc installed to providc thc the diffuser and thcrcby increasing cxtrac~ionif fuel supplies
required bleed pressure. permit.
Based on these results, thc selected design for a new vessel The configuration sclcctcd for a quintuplc effect evaporator
was to install a first effect with 2 200 m' of heating surface is:
and swop the duties of the present second and third effect
vcssels. If converting the semi-Kcstncr to a Kcstncr was cco- Effect Type Arca (m')
nomic, the 2 500 m' vessel that would result from installing First Kestner 2 000
standard 7,2 m long tubes would be more than adequate.
Second Semi-Kestner 1 523
Both of these options can be shown to be compatible with
designs for possiblc future expansion. Third Robert 401-
Fou rt l1 Robcrt 702
These results clearly dcmonstratc thc importance of dcsigns Fi l't h Robert S77
that are 'robust' with respect to design assumptions. The
robustness is shown as being rclativc to thc assumption of
blecd rate, but could be equally intcrprcted as robustness to This conl'iguration is also compatible with dcsigns for possi-
assumed first effect heat transkr cocn'icicnt. ble future expansion.

Fi~rrll.telectioll qf evaporator corfig~lrtrtio~i

Refilrbisld gl~irlticpleeffect e~~ri/~orcitoI-
'The final selection ol'cithcr a quintuple or quadruple evapo-
The design for the creation of a quintuple cvaporator was rator configuration for thc rcfurbishmcnt is dependent on
investigated on the basis of installing a ncw first cfkct detailed costings of'these alternatives and had not been made
Kestner vessel. The existing semi-Kcstncr and Robert vcs- at the time of writing this paper.
sels would be retained and their duties altered as necessary.
Computer simulations over a range of vapour bleeds did not Equipment design for the refurbishment of Xinavane
show the same dcsign sensitivity seen in the design of the evaporator station
quadruple effect evaporator. This is because in all instances
the first effect area that is required to be installed is defined Although this papcr is primarily about the principlcs of
by the need to provide the required bleed pressure and not by evapol-ator station dcsign, i t is also germanc to mcntion here
the need to supply evaporation capacity. An alternative way some of thc spccifics 01' evapol-ator equipment design as,
of interpreting this is that there is more than sufficient area without proper attention to these areas, the levels of per-
available in the existing vessels to be used for the sccond to fol-mancc assu~ncdin the system dcsign will not be met, and
fifth effects. The area efficiency criterion again indicates that the station will fall short of requircnicnts or exceed con-
the relative positions of the existing second and thil-d effect straints.
vessels should be swopped for the niost effective use of heat-
The juice l'ccd arrangements of the existing vcsscls was rudi-
ing surface although this change does not reduce the required
mentary and recluired rnoclii'ication. Proper distribution of
first effcct area (i.e. thcrc is more than sufficient area in the
the flash gcncrutccl when the juice enters the next effect in a
tail whether the vessel duties al-c swopped or not).
train significantly improves circulation in thc vessel and
'The excess capacity is shown i n simulations of performance therelhre heat tl-ansfer per-i'ormancc.A feed ring external to
at design conditions as a large pressure dl-op across a valve the vessel with short fccd stubs cnding in carefully sized ori-
throttling vapour to the third effect calandria, w ~ t hgreater ficcs ensures this distribution while allowing ease of clean-
throttling with the swopped sequence. It1 practice the extra ingldescaling as necessary.

Proc S Afr Silg Teclr~lolAss (1999) 73

 DJ Love, DM Merrtlo\vs & KG Hoekstrn

High performance entrainmcnt separators are essential to the basis of sccond effect Icvcl. The vapour throttle valve
ensuring a reliable supply of good cluality condensate for afcer the blccd is controllcd on clear juice tank level.
boiler feed. Thc cxisting evaporators at Xinavane were fitted
with ccntrif~~galvanc sepal-ato1-s(anglecl vanes around a cen-
Conclusions
tral hub), with indifkrcnt perl'orniance. Tongaat-Hulctt has
had success with an inhouse design ol'vertical Chevron Plate The redesign of the Xinavane evaporator station has reaf-
(VCP) separator, which typically achieves contamination I'irn~edthe place ol' thc evaporator station at the core of a
lcvels below 5 ppm, is free draining (nonclogging) and has sugar lhctory dcsign, filling as it does the simultaneous roles
been proven on all evaporator cft'ccts as well ;IS pans. It is ol'juicc concentrator, exhaust condenser, vapour utility sup-
modular and can bc arranged for clcaninginplace. New units plicr and boiler feedwatcr supplier, while playing a key part
ol'this design Ihr all vesscls is part of the rcfurbishment. i n the 1'~tel/ steam /power balance o f t h c f'actory. In addition,
the specil'ics ol'the Xinavane station have led to a revicw of
Correct sizing ol' the systems for I-cmoval of incondensiblc
the optimal distribution of hcating surl'acc in an evaporator
gases fl-om the cvaporator calancll-ias is crucial to achieve
train and the potential sensitivity ol' evaporator capacity to
complete scmoval without cxccssivc stcam wastage.
ch~unpcsin vapour bleed ratc.
Accurate design in this area allows evaporators to bc run
without manual throttling of incondcnsiblc vents, with its This cxa~nplc has demonstratecl that co~-rcctequipment
associated margin 1.01-error. dcsign and 'common sense' sugar engineering are necessary
but not sul'l'icient in clcsigning an optimum cvaporator station
The internals ol' the existing final cffcct external condenser
and that pitfalls may []-upthe unwary. A fully robust station,
arc simple ball'lcs. The rel'ul-bishment I-cplaccs these with a
able to lncct all o f the variecl station criteria under any rea-
raintray dcsign capable ol' achieving approach tempcl-aturcs
sonable set ol'operating circumstances. can be achieved only
of less than 3 K, minimising thc load on thc in-jection water
by delving decpcr into the detail.
systcm and ensuring steady final el'l'cct absolutc pressure.
Effective cooling of incondensiblc gasses in the condcnscr is
also important to avoid overloading the vacLlilm pump. REFERENCES
The latest design of Kcstncr type cvaporator (first effect for Ruczolich. A and Zatlori (1963). Ol>tim;rl distribution o f hcating surfilccs
either quin or quad) is fitted with Juice recycle, in recogni- in n tnulli-stage evaporator station. Z~lcXo.Vol. 16(5): 117-1 19.
tion of rcccnt work at the SMRI and Tongaal-Hulctt Sugar
Hockstra. R C ( 198 1 ). A compulcr program for simulating and cvalualing
(Walthcw ancl Whitclaw. 1996) which has confirmed the ~ilultiplccl'l'cct cvapor;ltors in the sugar industry. Proc S Afi- S~rg
positive cl'l'ecr on heat transfer perthrmancc of incl-'ascti 7i,c.l11rolA.c.7 55: 43-50,
tube wetting rates. 'The Kcstnel- separator juicc outlet will be
Hu~otE. ( 1986). I-l(~ritlbooX01' C(~ric,SLI~~II-l:',rgirieerirrg. Third Edition,
fitted with a weir systcln which ochievcs preferential recycle Elscvicr. Amsterdam.
up to a critical flowratc, beyond which rccyclc becomcs pro-
portional. This nvoicls the cornplexity 01' a p~lmpingor con- Rein. PW and Hocks~ra.RG (1994). Implications ul' cane diffusion for
cncrgy economy in a sugar mill. ISSCT comhinctl factorylenergy
trol system on the rccyclc line.
workshol~on cl'l'icicnt ~~roduction nncl uliliantion ol's~cumi n sugar t'ac-
'The I-cl'urbishrncnt includcs automation ol' the cvaporator torie.;. licltl at Punc. India. 11 347-3.5s.
station, using the 'cascaclc back' stl-atcgy. In this strategy, the Snlitii. I A ;~ntlTaylor. L A W (1% I ) . Sotnc data on heat 11-nnsleri n multi-
prefen-cd systcm I'ol- syrup bl-ix control is a radio I'rcqucncy ple effect cv:lporators. Proc S Afi. Sr~g7i~c1111o/ A.SS55: 5 1-55.
probe, on thc basis ol' which the ratc ol' syrup cxtl-action is
Spencer, C L and ~McudcGP ( 1945). Grrrr. S'LIS~I. Hor~clbook.Eighth
throttled. The level in each vcsscl cont~-olsthe ICccI rate to Edilion. .lolin Iliiley, New York.
that vcsscl, with the exception of thc fil-st cfkct, as thc scp-
W;ll~licw.D and Wlii~clawR W ( I996). Factors al'fcctin_~ the l~crfor~nance
aratol- is designed to operate without maintaining a juicc
PI.OC .S Afi. .SII~ Tc~chr~oI
ol' long tube clilnhinf filni cva~>o~.ators. A.c.5
lcvcl to minimise rcsidcncc time and therefore degradation
70: 220-224.
of the juice. Clcar juice flow rate is therefol-c controlled on

Prnc S Aj'r Sug Teclri~olAss (1999) 7-3

Robust design of an evaporator station as applied to Xitznvnne DJ Love, DM Meadows & RG Hoeksfrn

APPENDIX 1 rule of resistances i n series being additive clearly applies.

The effect of vapour bleed on evaporator capacity can be The lengthy, rigorous calculations to demonstrate the effect
demonstrated by the following analysis based on the simpli- of bleed on evaporator capacity can be avoided by the fol-
fying assumptions of evaporator behaviour and equations lowing approach:
detailed in the text of this paper. Consider a quadruple evap- An increase in bleed tlow will cause a drop in the vapour one
orator with vapour one bleed only, as shown in Figure 1 in
pressure, increasing the temperature difference across (and
the text of this paper. For simplicity (in line with the first
the evaporation in) the first effect. This increase in the value
principle of Rillieux) the quantity of evaporation taking of A T F , which is assumed to be dT, will cause an increase
place in the second, third and fourth effects is assumed to be dWF i n the first effect evaporation which is given by
equal. The temperature differences shown are the effective
temperature differences available for driving the heat trans-
fer. They are less than the available temperature difference
between the temperature of steam i n the calandria and that in
the vapour space as a result of the elevation in boiling point At the same time there will be an equivalent decrease in the
of the juice due to concentration and hydrostatic head value of A T T , (since the total effective temperature differ-
effects. ence across the evaporators ATE remains constant). This will
In this analysis we are particularly concerned with compar- result i n a decrease dWT in thc cvaporation over the tail,
ing the evaporation in the first effect with that in the rest (i.e. which is givcn by
the tail) of the evaporator station.
For the first effect, the evaporation W F can be expressed i n
terms of the effective temperature driving force ATF as

Clearly, if the increase in first effect cvaporation is greater

than the decrease in tail evaporation, then increasing the
blecd flow will increase evaporator capacity. This can be
For each effect in the tail the cvaporation W can be expressed
formalised by defining a 'bleed capacity factor', BCF, as
as
dWF
BCF = -
~ W T
Thus.

1
BCF =
1

Defining the effective temperature difference across the

whole tail ATT as I
Continuing with the concept of I,, being the resistance of
an evaporator to heat transl'cr, the BCF can be seen to be
simply thc ratio of the avcragc rcsistancc of the tail evapora-
it is possible to express ATT as tors to the resistance of the first effect.
This 'bleed capacity factor' can be used to identify the fol-
lowing three conditions:
BCF = 1 Increasing vapour bleed has no effect on evap-
and thus to express the evaporation in the tail WT as
orator capacity. (This will occur, for example,
W,=3.W if U, . A l = U 2 . A 2 = U, . A 3 = U 4 . A 4 ) .
and BCF > l Increasing vapour bleed increases evaporator
ATT capacity. Thisis thc most common situation in
W,=3. sugar Sactories.
1 1
A. (-
U 2 A 2 ++L)
U
~ 4A4 BCF< I Increasing vapour bleed decreases evaporator
I capacity.
If we consider o,to be the resistance of an evaporator to
heat transfer, in analogy with electrical circuits, the simple

218 Proc S Afr Silg Teclznol As.s (1999) 73

 THE USE OF SEQUESTERING AGENTS FOR CHEMICAL
CLEANING AT UBOMBO SUGAR

SD PEACOCK1,',DC WALTHEW1.',TH DE BEER' AND P NEEL'

'Sugar Milling Resear.clz Irlstitute, University o f Natal, Durbarl 4041.

'Illovo Sugar Limited, Uhon~boSugar Mill, Private Bag, Big Berld, Sw~rzilat~rl.
'NOH.at Totgnat-Hulett Sugar Lirnitcd, Private Bag 3. Glennslzley, 4022.

Abstract the cleaning proccss is actually expended in neutralising the

persistent caustic rcsidue in the evaporators, and does not
The chemical cleaning of certain evaporator types I-equires
contribute towards the cleaning ol'the heat transfer surfaces.
the use of an alkali cleaning agent, followed by an acid
In ordcr to avoid this wasteful use of acid, other methods of
cleaning agent. The use of acid under these circumstances is
cleaning were investigated, including the use of sequestering
expensive, as most of the acid used during the process is
agents, which arc compatible with the alkaline environment
consumed in neutralising the residual alkali present in the
which exists in the evaporators k>llowing the caustic soda
system following the caustic clean and watcr rinse. An alter-
cleaning psocess. Chelating agents (or complexing agents)
native to the use of acid in the chemical cleaning process is
al-e molecules which form multiplc stable attachments with
the use of sequestering agents (such as EDTA or sodium glu-
metallic ions. Sequestering agents are a sub-group of chelat-
conate), which are compatible with the alkaline envil-on-
ing agents which combine with mctal ions to produce solu-
ment present in the system and are thus not rendered less
ble complexcs in a solution (discussed at a later stage).
effective by the presence of caustic rcsidues. The use of
sequestering agents for chemical cleaning was tested at the The most versatile and widely used sequestering agent is the
Ubombo sugar mill in Swaziland, and the results of these tetrasodium salt ol' ethylencdiaminctctraacetic acid (this
factory trials arc discussed. Both EDTA and sodium glu- tetrasodium sal1 is commonly referred to as EDTA - also
conate were found to he effective at replacing acid during known as Ve~:reile).Trials wcrc carried out at the Ubon~bo
the cleaning proccdure. The economic advantages of using sugar mill to dcterminc the technical and cconomic ieasibil-
sequestering agents for cleaning are outlined. ity of using EDTA to rcplace the acid cleaning step i n the
chemical cleaning process. While the relative cost of EDTA
compared with phosphoric acid is about three times more (in
Introduction
terms of equivalent quantities of calcium removed), the inef-
Investigations in the local sugar industry havc indicated that ficient use of acid in the normal cleaning proccdure means
evaporator fouling is a major contributor towards poor heat that. in practice, significantly less EDTA would be required
transfer in evaporators (Walthew, 1994). As evaporation is to carry out the snmc degrec of cleaning. As EDTA is rela-
an important unit operation in a raw sugar mill, a broad tively cxpensive, the use of a cheaper sequestering agent,
investigation into Ihuling was undertaken over the past few sodium gluconatc, was also investigated.
years in the southern African sugar industry. Thc overall
aims of this study were to identify methods of reducing or Conventional chemical cleaning at the Ubombo
preventing scale formation and to improve thc efficiency of
sugar mill
cleaning of fouled evaporator surfaces.
The chemical cleaning of evaporators In the southern As part of an expansion psogranimc and a drive 1'0s energy
African sugar industry is discussed by Walthcw et al. (1997). cfficicncy, Uhombo cugar mill convcrtcd its cvaporator sta-
Most evaporators can bc cleaned using an alkali clcaning tion from a quadruple cffect to a quintuple cffect in 1996.
agent (such as caustic soda) only. However, thc presencc 01' Thc existing evaporator vcssels were l-carranged, and plate
significant quantltics of certain calcium salts (such as calci- evaporators were incorporated as the second and third
urn oxalate and calcium carbonate) which are essentially cfl'ects (DC Becr and Moult. 1998). Unl'ol-tunately, a very
insoluble i n caustic soda makes the use of an additional acid high I-atc of fouling was experienced in the plate evapora-
cleaning stcp necessary for the effective cleaning of certain tors, and an effective cleaning proccdure was I-equired.
evaporators. Furthermore, the use of acid cleaning in addi- Chemical cleaning is essential for the routine cleaning of the
tion to alkali cleaning leads to the breaking up of the evapo- platc evaporator packs becausc physical cleaning would take
rator scale into smaller pieces. In thc case of nart-ow gap too long to carry out1.Furthermore, the inter-plate gaskets
plate evaporators, the acid is thus beneficial in helping the
scale to be flushed from between the plates. However, in
' While the platr evaporator.; arc undergoing cleaning, n caustic soda solu-
tion is boiled in the other cvnporntors in ol.der to soften thc scale. The
these circumstances, a large proportion of the acid used in Robe~tsand clirnbing fill11 vessels arc, howeves, still cleaned ~nechanically.

Proc S Afi- S ~ i gTechriol Ass (1999) 73 219

The Lue of sequestering agerlts,for cheti~icnlcleat~itlgat Ubo171bn SD Pecrcnck, DC W~iltllei~:
TH De Beer & P Neel

undergo damage each time they are disturbed, and replacc- formed are so effective that the metals rcmain in solution
ment of these gaskets is expensive. Additionally, analysis of even in the presence of anions which would normally result
the evaporator scale formed at the Ubombo sugar mill has i n their precipitation from solution. This property can be
shown the presence of significant quantities of calcium car- exploitecl In the cleaning of ibulcd evaporators, as calcium
bonate, which is not effectively removed by caustic soda, but and magnesium salts arc generally a major component of
is attackcd by acid cleaning. evapol-atol-scale in the sugar industry. In the presence of a
sequestering agent, these salts readily dissolve, thc calcium
As a result of a series of chemical cleaning trials over an and magnesium forming soluble complexes.
extended pel-iod, the following cleaning procedure was
developed I'or the cffcctivc cleaning of the plate evaporators: The most widely used and versatile sequestering agent is the
Evaporatol- stops are scheduled fortnightly. tetrasodi~~msalt of cthylcncdiaminetetraacetic acid, also
When the flow of juice to the evaporators stops, the ves- known as EDTA or Versene. Thc normal complcxing reac-
sels are deswcetcncd by displacing the residual juice with tion which this compound undergoes with the calcium ion is
water. shown i n Figurc I (Bcrsworth Chemical Co., 1953). where
one molcculc of EDTA reacts with one calcium ion.
After displacing the juice, watcr is introduced into the
vessels and boiled under normal operating conditions for
one hour. The water is introduced into the second effect
and discarded alter the third effect.
The watcr boiling is followed by boiling a 12,5% caustic
soda solution under normal operating conditions for fivc
hours. To this solution is added 0,25 to 0,5% (by volume)
of wetting agent, which significantly increases the clean-
ing effectivcncss ol' the caustic. 'The caustic is circulated
from thc sccorid effect to the third effect and then back
again. The condensate from the vessels is returned to the
caustic solutions so as to maintain a constant caustic con-
centration during the cleaning proccss.
'The caustic solution is then drained 1'1-omthe evaporators
and replaced with water, which is boilcd under normal
operating conditions for one hour. The water is introduced
into thc second effect and discal-ded alter thc third effect.
Following the water boiling, the steam supply is isolated Figure 1. The general complexing reaction of EDTA with the calci-
um ion.
(and mechanical cleaning of the tubular evaporators takes
place).
Flushing of the vessels with water is maintained until the
EDTA in its acid lbrm (cthylc~~cdiaminetetraaceticacid) con-
pH of the discharged watcr drops to below 8. This typi-
tains l'our ionisable hydrogen atoms, and therefore the fol-
cally takes three to four hou1-s.
lowing cquilibria exist:
Phosphoric acid, of strength I 0 to 15% and containing 1
to 2% inhibitor, is then circulated through the plate packs G>
I'or a period of about foul- hours. As (lie steam supply is H4Y H+ + H 3 Y ~
unavailable, thc acid cannot be heated during circulation, H,Y- H+ + H2Y'
and the temperature typically does not cxcccd 40°C. This H2Y2- H+ + HYq-
is not idcal, as the acid should be circulated at 70 to 95OC. HY '~ H+ + Y J ~
Following the acid clean, the evaporators are rinscd and
where Yd. is shown in Figure 2.
returned to sel-vice.
The major disadvantage of the current cleaning pr-ocedure is
the acid cleaning step. The I-emoval of the rcsidual caustic
soda i n the system is extrenlely difl'icult and requires large
quantities of watel-. Large quantities 01' acid arc wasted in
neutralising the persistent caustic residue, rather than con-
tributing towards the cleaning 01' the evaporator surfaces.

Figure 2. The fully ionised EDTA molecule (Y4-).

Sequestering agents
Sequestering agents are molecules which I'orm stablc, solu- In an acidic solution, H,Y is the dominant species present.
ble complexes with heavy mctal or alkaline earth ions (for As thc pH is increased, hydrogen ions are removed, with Y"
examplc, calcium and magncsium ions). The complexes so cvcntual l y bccoming thc predominant species (typically

Pmc S Afr Siig Techtlol Ass (1999) 73

The use ofseqlresteritrg agents for cllerrlical cleatlir~gat Uboml:)o S D Pencock, DC Walthew, TH De Beer & P Neel

above a pH of I I or 12). The ions which form complexes

with the calcium ion arc YJ- and HY'- (Holland et a / , 1954).
Thus, in alkaline conditions, calcium will form complexes:

Ca2+ + Y 4 - CaY2-
Ca2+ + HY' > CaHY~
Under acidic conditions, however, whcre the Y4- and HY1-
ions are present in small concentrations, calciu~nis not com-
plexed. It is thus essential to maintain alkaline conditions for
the effective use of sequestering agents for chemical clean-
ing?. These reversible cquilibria associated with EDTA mean
that an EDTA cleaning solution can be regenerated by acidi-
fication with sulphuric acid. The acidification results in the
precipitation of calcium sulphate (gypsum), which can be fil-
tered off. Addition of caustic soda then returns the EDTA to
the tetrasodium salt form, which can be re-used for chemical
cleaning (Holland et al., 1954; Bennctt et al., 1955).
Figure 4. T h e sequestering power of sodium gluconate as a func-
Another effective sequestering agent is sodium gluconate. tion of caustic soda concentration.
The structure of this molecule is shown in Figure 3.

OH
I
7 OH H
l
OH
l /P
to complex the equivalent amount of calcium. This is rough-
ly equivalent to a requirement of 10 kg of EDTA per 100 m?
ol'surfacc asea. By comparison, in the beet sugar industry the
H-0-CC- I C--C--C-C,
I I
0-Na EDTA rec]ui~.emcnt is approximately 65 kg per 100 m'.
H OH H OH H However, the scale to he cleaned is rich in calcium oxalate
Figure 3. The structure of the sodium gluconate molecule. and calcium carbonate, and the cleaning is only carried out
once per season.

Sodium gluconate, bcing the sodium salt of gluconic acid, During the ncxt cleaning cyclc at the Ubombo sugar mill,
behaves in a similar fashion to EDTA with regard to the ion- EDTA was wbstitutcd Ihr thc pllosphoric acid. The 111ajor
isation of its hydrogen ion under alkalinc conditions. The advantage of using EDTA for chemical cleaning was that the
complexing power of this sequestering agent is thus also time consuming water boiling and flushing steps could be
very dependcnt on maintaining an alkaline environment dur- omitted, allowing the second stage ol'cleaning to take place
ing its use (typically a pH of above 8 is required), as shown whilst hcuting steam was still available. Aftel- the caustic
in Figure 4. boiling, mosl of thc caustic was drained out. and the seques-
tering agcnt was aclcled before the steam was isolated. The
chemical could be boiled, and i t remained hot during the
The use of sequestering agents for chemical cleaning recirculation period of four to i'ivc hours, evcn after isolation
at the Ubombo sugar mill of the steam supply.

Clear~itrgrvitlz EDTA A further reduction in cleaning time could be obtained if the

EDTA could be added at an earlier stage, with the boiling
The rirst step in [he use ol a seclucstertng agent for chemical
caustic soda. Howcvcr, this would rccluire that the scale
cleaning is to determine the quantity of thc chemical
solids be I-cmoved l'rom thc caustic belorc the EDTA is
required to carry out the cleaning proccss. For the trials
added, in order to prevent the expensive sequestering agent
under consideration here, samples of phosphoric a c ~ dwcre
from preferentially dissolving thc scale which has already
taken at the Ubombo sugar mill bcfore and after an acid
broken away from the heat transl'er surface.
clean of the plate evaporators, and analysed at the SMRI for
calcium content. From these analyses. it was found that 24
kg of calcium had been removcd I'rom the evaporators by the
During the I'ollowing fortnightly clean, the use of sodium
16 m' of acid used during the cleaning process. Thc total sur-
gluconate as a substitute Ibr acid cleaning was tested. The
face area cleaned was 3762 m'. Given that one molccule of
sodium gluconate was simply aclded to the caustic cleaning
EDTA is required to complex one calcium ion, it was calcu-
agcnt in the evaporators following four hours of caustic boil-
lated t h a ~approximately 250 kg of EDTA would be required
ing. A dosage rare of 6 kg of sodium gluconatc per 100 m' of
cvaporator surlhce area was used. A comparison between the
procedures used for the sequestering agent cleantng proccss-
P- P-

' The EDTA soltl 1.01. chcinicnl clcailing is L~suallysold as the tet~.asodiunl
CS and that used for the conventional acid cleaning process is
salt (i.e. h'a4Y), which is essentially ethylencdinniinetctmcetic acid neu-
tralised with caustic soda. shown in Figure 5 .

Proc S Afr S L LTechnol

~ Ass (1999) 73
The use of sequestering agents for chemical cleaning at Ubombo SD Peacock, DC Walthew, TH De Beer & P Neel

Steam

Available

Water boil
Ilour through EDTA Recirculate

4 to 5 boil / with

3 to 4 Water Once hours circulate colidensate

honrs flush through hot return

Water Once

flush through
4 Acid Isolated
Recirculate
hours clean
In service

Water Once

flush through

In service

Figure 5. Comparison between acid and sequestering agent cleaning procedures.

Results process was found to be light brown. This scale residue also
appeared to break up much more easily.
Unfortunately, it was impossible to measure the heat transfer
coefficient during the sequestering agent cleaning processes, Scale residue samples were collected from the Ubombo
as the condensate flow meters installed at the Ubombo sugar sugar mill following the acid and sodium gluconate cleaning
mill for this purpose were inoperative. Thus, the only assess- trials and analysed by x-ray fluorescence (Walthew and
ment of the cleanings which could be made was visual. Turner, 1995) and organic acid analysis. These samples of
Following the EDTA cleaning, it appeared that a lot of scale scale consisted of pieces which had broken free of the evap-
flakes had been dislodged, and partially blocked some of the orator plates during cleaning and had fallen to the bottom of
plate evaporator passages. It is possible that these flakes the plate cvaporators, where they were collected following
were scale which had not been effectively removed during the cleaning process. The estimated scale compositions
the previous acid cleaning cycles during the season. There based on these results are shown in Tables 1 and 2, which
was an obvious difference in the scale residue appearance indicate that the phosphate and silica components of the
when using sodium gluconate for cleaning. While acid clean- scale appcar to have been more readily attacked by the sodi-
ing resulted in the scale flakes being black in colour, the um gluconate than they were by the phosphoric acid. As cal-
scale which was removed by the sodium gluconate cleaning cium silicate is one of the most difficult scale components to

222 Proc S Afr S L LTechnol

~ Ass (1999) 73
The use of sequestering agents for chemical clearling at Ubombo SD Peacock, DC Walthew, TH De Beer & P Neel

Table 1. Scale compositions (in %) based on x-ray fluorescence and organic acid analyses.

Ca Loss on
SiO, CaO MgO P,O, Aconitatc
oxalate ignition

Acid clean scale residue 28,4 17,9 3,5 10,6 1,3 0,s 35

Sodium gluconate clean

14,l 18,s 0,3 0,9 10,4 0,o 66
scale residue

Table 2. Scale compositions (in %) based on x-ray fluorescence and organic acid analyses.

All1

orph Am
M Ca. Li
ous Si Mi OrP
Corn go Mg me
calci 0,. xed hou
POU aco hyd
LI m H, oxa S
nd H2 nita rat
phos 0 late orga
0 tc e
phat nic

e
Acid
36,
resid 28.6 5,l 1,3 0,5 6,4 21.1
9
ue

Sodi

uln

gluc
18, 10, 18,
onat 2,4 0,5 0,o 49,s
4 4 5
e

resid

Ue

remove during the evaporator cleaning process, it is encour- use of sodium gluconate for evaporator cleaning results in a
aging that this component is readily attacked by sodium glu- significant saving in cleaning cost. Over a 40 week season
conate. (comprising of 20 cleaning cycles), the projected savings
resulting from the use of sodium gluconate instead of phos-
Following the cleaning cycles, heat transfer measurements
phoric acid amount to R 80 000.
made during normal operation suggcst that the sequestering
agent cleaning processcs were as effective as the acid clean-
ing process. Thc Ubombo sugar mill plans, in the 199912000 Conclusions
season, to alternate between using phosphoric acid and sodi-
The use of sequestering agents for the chemical cleaning of
um gluconate for chemical cleaning of thcir plate evapora-
evaporators was tested at the Ubombo sugar mill i n
tors.
Swatiland. Both EDTA and sodium gluconate were found to
Cost implications bc effective when used to rcplace the acid cleaning step in
the Ubombo cleaning procedure. The use of sequestering
The chemical costs for the three cleaning processes evaluat- agents for evaporator cleaning holds an cconomic advantage,
ed in this study are shown i n Table 3. It can be seen that the and the Ubombo sugar mill plans 10 alternate the use of

Proc S Afr Scig Technol Ass ( l 999) 73 223

The use of seq~~esterirzg
agents for clzemical clearling at Uboinbo SD Peacock, DC Walthekv, TH De Beer & P Neel

Table 3. Chemical costs for evaporator cleaning at the Ubombo sugar mill.

Chemical cleaning agents used Cost of chenlicals Savings

1. Caustic soda -t wetting agent
R 10 018
2. Phosphoric acid + inhibitor

1. Caustic soda + wetting agent

R X 037 20 Yo
2. EDTA

1. Caustic soda + wetting agent

R 6 000 40 %
2. Sodiunl gluconate

phosphoric acid and sodium gluconate during the 199912000 REFERENCES

season. Bennctt, MC, Connolley, FH, Schmidt, NO, Wiggins, LF and Wise, WS
Further work needs to be carried out on the use of sequester- (1955). Further developments in the use of Versenc in evaporator
cleaning. Proc BWI S ~ l gTech Meeririg: 83-85.
ing agents for chemical cleaning, particularly with regard to:
the development of a reliablc analytical techniquc for the Bersworth Chemical Co. (1953). Thc Verscnes. Techr~icolB~illetirz 2.
measurement of the concentration of sequestering agent in Fmmingharn, Massachusetts.
solution, so that this may be monitored during the clcan- DC Bccr. TH and Moult, LM (1998). Experience with plate evaporators at
ing procedurc Ubornbo Ranches in Swaziland. Proc S Afi. Slrg Teclinol Ass 72: 228-
the effectiveness of sequestering agents for cleaning thc 233.
scale in a factory experiencing high silica concentrations Holland, ID, Massiah, BV, Mayers, JC, Schmidt, NO, Wiggins, LF and
the rate at which calcium-containing scale components Wise, W S (1954). Thc usc of Vcrscnc in cvaporator cleaning: A fac-
are attacked by sequestering agents, so that the length of tory trial. Proc BWI S L ITecll
~ Meetirlg: 155- 16 1.
time rcquircd for optimally effective cleaning can bc Walthew. DC (1994). Evaporator fouling literature review.
determined. the SMRI. Nurnber 159.
Cor~~~ii~~riictrtiorrs,fro~i~
Walthcw, DC and Turner, LM (1995). Analysis of scale from some South
African sugar mills. Proc S Afr S L ITechr~ol
~ AS.F69: 138- 143.
Walthcw, DC, Whitelaw, RW and Mohabir, R (1997). Chc~nicalcleaning
of evaporators. Proc S Afr S L ITechriol
~ Ass 7 l : 199-206.

Proc S Afr S L LTechnol

~ Ass (1999) 73