7/28/2019 1991_Reid_The Protection Of Boilers.pdf

1/5

Proceedings of The South African Sugar Technologists' Association - June 1991

THE PROTECTION OF BOILERS FROM SUGARCONTAMINATION IN FEEDWATER

ByM. J. REID and A. DUNSMORESugar Milling Research Institute, Durban

AbstractSugar contamination of boiler feedwater by smallcontinuous traces is not as serious as that from large 'slugs' whichfrequently find theirway into the boilers. Reference ismadeto several incidents of sugar contamination in boilers toshowthat themost commoncauses area malfunction ofthefeedwater system or its mismanagement, and to show thatthe method of sugar detection is seldom at fault. Some ofthe devices for detecting sugartraces in useor proposed arementioned and their relative merits are discussed. In particular, the apparatuswhich was recently evaluated at the

Sugar Milling Research Insitute(SMRI) is discussed in somedetailand test results aregiven. Theapparatuswas designedto measure the change in pH and conductivity of a conden-sateafter being heatedto 265Cfor a few minutes. Sugges-tionsaregiven on the typeofmonitor to beusedin rawandrefined sugar operations, the design and operation of thecondensate handling system, and the options available forthe future. It is concluded that, subject to some importantprecautions, the conductivity monitoring method is satisfactory for protecting boilers in a raw sugar mill againstcatastrophic failure.

IntroductionTracesof sugar in boiler feedwater up to somearbitrarylevel, possibly as high as 50 parts per million (ppm), arebelieved to be quite safe for the performance and safety ofboilers, providedthe chemical treatmentof feedwater is adequate.However the occasional heavyslugs of sugar whichfind their way into the feedwater system can be verydangerous and it is thesehighdoseswhich needto beeliminatedbyan effective sugarmonitoring system to protect the boilers.In recent years the SMRIhas received many queries onthemethodsand procedures used to monitor the sugar content of condensates intendedfor boiler feedwater. Much hasbeenwrittenand researched on this subject yettherearestillseveral unanswered questions, for example: Whatistherealeffect of traces ofsugarinboilerfeedwaterand what costsare involved?

What level of sugarcontamination is acceptable? What method of detecting sugar in condensates is bestsuited for the protection of boilers?Most engineers would be preparedto spend large sumsofcapital and maintenance money to be able to accuratelymeasure the contamination down to levels of perhaps 20ppm. This is justified on the basis of the very high costsincurredin boilerdowntime, repairandmaintenance causedeither directly or indirectly by sugar contamination. How-ever thesehighcostsare incurredonlywhen sugar contam

ination is excessive and undetected. For boiler safety it ispreferable to detecthigh levels of sugarcontamination withabsolute certaintythanto expend greateffort on theaccuratemeasurement of small traces.208

Experiences of sugar in boilersGeneralThere have beenmany incidents of sugarcontaminationin boilerssince the sugar industrybegan to use steam. Mostoftherecentexperiences ofthisphenomenon inSouthAfricahave not beencatastrophic, and the boilers have survivedwith little or no permanent damage.Usually the incident starts with undetected entrainmentor leakage of sugar into condensate, which eventually revealsitselfby the characteristic smell of caramel emanating fromsteamdrains. The pH in the boiler falls sharply and mustbe controlled by slug dosing caustic soda into the feedwater.If the pH responds to this control the boilers can be kepton line, provided an adequate supply of uncontaminatedfeedwater is available.The following is a brief list of the causes and effects ofsome of the incidents which have come to the writers'attention:

A welding glove left in a juice line caused one effect ofthe evaporator to fill up and overflow into the calandriaof the next vessel. Several hoursof production were lostbut there was no permanentdamage Failure of the juice level control in an evaporatorcausedjuiceto fill thevessel to thevapouroutletand flow throughthe condensate system into theboilerfeedwater tank.Fourdaysof production were lost

Several boilershut downs have been caused by failure ofjuice heater tubes Extensive tube leaks in a Bpan caused sugarto reach theboilers three times in one season Rejected condensate overflowed froma sweetwater tankback into the feedwater tank and eventually entered theboilers The Kestner separator outlet became choked with scale,which caused juice to flow into the vapour lineand theninto the boiler feedwater tank . Excessive dumping was too much for the 'reject' pipingand backflow entered the boilerfeedwater tank.Thesemay be regarded as typical for the sugarindustry,and many engineers will recall similar experiences in theirmills. Themost important effect of theseincidents was thesevere loss of production and consequent financial losses,butinno instance was therepermanent damage totheboilers.In all of theseincidents and inmanyothersnot recorded,the amount of sugar contamination wasveryhigh, and fordifferent reasons was not detected by theconductivity monitoring system. In all cases however the reasons were notrelated to the typeofmonitor but to failures in the systemsuch asdirtyelectrodes, poorelectrical contacts, faulty valves,blocked pipes, or poor design of the dumping system.

Effects of sugar in boilersSucrose breaks down at high temperature into organicacids, which lower the pH and cause an increase in con-

7/28/2019 1991_Reid_The Protection Of Boilers.pdf

2/5

Proceedings of The South African Sugar Technologists' Association - June 1991

2.53r- - - - - - - - - - - - - - - - - - - - ,

12000~ 100 ppm sucrose

20 40 60 80Residence time (minutes)

FIGURE 2 Conductivity vs time.

The first version of the apparatus comprised a samplingsystem, which commenced with a filter, followed by a combined conductivity/pH cell and a high pressure pump thatpassed the sample through a stainless steel coil contained ina small thermostatically controlled oven. The sample wasthen cooled in a coil in a water bath, after which it passedthrough a pressure reducing valve and then into a secondcombined conductivity/pll cell.Preliminary experiments were carried out to determinethe best residence time by sealing a small sample in theheating tube and raising its temperature to 265C for different periods up to 100 minutes. It was found that therewasa linear relationship betweenchange in conductivity andresidence time up to the first 50 minutes and a tendency tolevel offwith longer times (seeFigure 2).The tests continuedwith a residence time of 20 minutes to find the effect ofsucrose concentration on pH starting from a neutral solutionand from a solution initially at a pH of 11,0. The results ofthese tests are shown in Figure I.

The pH change from neutral with sucrose concentrationsin the range 0 to 120 ppm is a fairly flat curve, whereas thatfrom pH 11,0follows a steeper curve, the shape of which isdifficult to explain. The major difference was that causticsoda was used to raise the initial pH to 11,0.At zero sucroseconcentration the pH changewas over one unit which couldpossibly be attributed to carbon dioxide absorption duringheating. .During these experiments it was found that the oven temperature control was erratic and heating was then carriedout in an oil bath. The fumes from the oil bath werehowevertroublesome and a third heating method was used, in whichthree electric heating elements, each of 300 watt, were attached to a straight 400 mm length of 6 mm diameter stainless steel tube and enclosed in an insulated housing. Thisheater worked well but the electrical connections were notvery secure because of the small dimensions and hightemperature.A diagram of the apparatus in its final form is shown inFigure 3. The sample was pumped at 8 MPa at a flow of0,55 Vhr, and heated to 270C. The overall residence timeof the sample was 2,5 minutes. Various safety devices wereincorporated as follows:

A pressure switch to cut out the heater if the pressuredropped below 7 MPa

100UlcQ) 80EQ)(j)6 60tig>... 40s;.;;o::3"0c:0 20o

00

12000

I- initial pH 11.0 -+- initial pH 7.0 I40 60 80Sucrose (ppm)20

FIGURE 1 Decrease in pH from static tests.

ductivity. This results in corrosion or excessive use of sodium hydroxide, and ultimately increases the total dissolvedsolids.The high temperature heating apparatus tested at the SMRIand described below has provided a means of estimating theexpected pH change in relation to the sugar content of asolution when heated to above 265C. This is shown in

Figure 1.At a sucrose concentration of 50ppm it was foundthat the pH could be expected to drop from 11,0 (which isthe required level for boiler operation) by approximately 2,9units. For a boiler house with a total evaporation rate of 150t/h of steam the cost of caustic soda required to restore thepH to 11,0 is negligiblecompared with the loss of sugar.If the presence of an excessive amount (i.e. > 200 ppm)of sugar in the boiler is not detected and dealt with, it cancause foaming, carry-over and fouling of strainers, steamtraps, control valves, turbine blades, etc. which can resultin very expensive repairs. In addition the evaporation process causes carbonaceous deposits in the boiler drum andon heating surfaces which reduce heat transfer, and couldcause blockages and corrosion. It is in these areas that ef

fective boiler water treatment is essential.The literature researched so far has not revealed any firmrecommendation for the maximum sugar contaminationpermitted in boilers in a sugar mill. It is suggested that afairly high level, of the order of 20 ppm, can be toleratedby a boiler for long periods provided sufficientcaustic sodais added to maintain the pH at the accepted level of 11,0and the water treatment chemicals are carefully controlled.Sugar detection apparatus

Several devices for the continuous monitoring of sugarcontamination have been proposed and several have beentested or used with success in sugar mills. A description ofthese instruments follows.Heating apparatus causing change in conductivityIn 1986 the SMRI carried out experiments to test thetheory that when a sucrose solution is heated to 265C forseveral minutes, there is a change in both pH and conductivity proportional to the sucrosecontent. The ideawasbasedon a Japanese paper (Takatori et al., 1975) in which experiments were carried out to test the theory. It was expectedthat an instrument could be designed around this principleand used to continuouslymonitor sugartraces in condensates.

J: 2c-.s5l 1.5co

uQ)o

209

7/28/2019 1991_Reid_The Protection Of Boilers.pdf

3/5

Proceedings of The South African Sugar Technologists' Association - June 1991

A heat-sensitive fusebuilt into the heater to fail at about400C A thermocouple arranged to trip the heater if the coolingwater outlet temperature exceeded40C A flow detector on the coolingwater to trip the heater ifthe cooling water flow dropped below a pre-set limit An interlock between the heater and pump motor so thatthe heater could not operate while the pump was off.

However the various components of the apparatus werefragile and failed frequently. The worst problems were associated with the pump seals, pressure switch and heaters.It was also evident that the tube was being contaminatedwith scale and frequent cleaning was necessary to ensuregood repeatability. The relatively long reaction time (threeto five minutes)wasalso believedto be a disadvantage. Theapparatus was therefore considered unsuitable as a factorybased continuous sugar trace measuring device and wasdismantled.

PUMP (8 PlPAl

SAMPLEINL.fT

HEATINGEillIENT

Totalorganic carbonThe principle of this instrument is the conversion of organic carbon to carbon dioxide by irradiation with ultraviolet light after addition of persulphate. The quantity ofcarbondioxide ismeasuredbymeansofan infrared detectorand the output signal is proportional to the organic contentof the sample.The apparatus has been locally modified and improved

by the Watertech Division of the Council for Scientific andIndustrial Research (CSIR)and is now commerciallyavailable at about R30 000. It has not yet been tested in a sugarmill environment, but is claimed to be robust and reliableenough to survive in any industrial application. Reactiontime could be reduced to about one minute per samplestream, depending on the accuracy required.THERMOSTAT

COOLINGWATER

FIGURE 3 Diagram of the apparatus for determining sugar traces.:

Aftersomedifficulties the apparatuswassuccessfully testedas a continuous sampler and a definiterelationship betweenconductivity and sucrosecontent wasmeasured (See Figure4).

Auto-analyserIn this apparatus water samples containing trace amountsof sugar are pumped into a manifoldwhere mixingwith thereagents (resorcinol/hydrochloricacid) occurs. The mixtureis then continuously pumped into a heating bath where col

our development occurs. The colour, which is proportionalto the sugar content, is measured in a colorimeter. An advantage over other designs is that the reaction is sugar specific. A disadvantage is that the reaction time can vary from5 to 15 minutes, depending on the reagents used and theaccuracy required.Auto-analysers have been used successfully to monitorcondensates for boiler feed, and descriptions of two typesof auto-analyserused in monitoring refinery wastewaters aregiven by Fowler (1977). Schaffler (1978) described the useofan auto-analyserformonitoring entrainment. Continuoususe of the instrument in the factoryenvironment in this casewas not very successful.Auto-analysers have worked well in a clean laboratorywhere supervision isgood, but theyare not suited to on-linemonitoring in the harsh factory environment.

50r-------------------,

FIGURE 4 Conductivity increase using continuous apparatus.

The 'Spaldinlab Sugar-Tekior'using three reagentsThis device is described by Parker and Bond (1958) andBruijn (1962). It is essentially an auto-analyser measuringcolour developedin the reactionwith triphenyl tetrazoliumchloride (TTC). The instrument uses two other reagents,sodiumhydroxideand hydrochloricacid,and requiresa smallgas burner to provide the source of heat for reaction. Thereaction time is approximately 15minutes.The instrument is commerciallyavailable through BritishSugar Corp. The SMRI tested a version of the instrument(constructed in the SMRI workshop)at Hulett Refineries in1961 and described it as reliable. However the longreactiontime is a distinct disadvantage.

2005000Sucrose (ppm)50.:

o ' - - - - - - - - - ' - - - - - - - ' - - - - - - J - --'o

Qlg: _ 40Ql '".. c:U Ql.s E 30

. ~ J l> ,0u '" 20::::J .2"OE-u 10

210

7/28/2019 1991_Reid_The Protection Of Boilers.pdf

4/5

Proceedings of TheSouth African Sugar Technologists' Association - June 1990Pulsed amperometric detector (PAD)This instrument measures the organic content of a solu

~ i o n by.detecting the current flowing between two electrodesImmediately after application of a predetermined potential.After the detection phase, two short cleaning pulses are applied to the electrodes to remove substances plated ontothem. The procedure is extremely sensitive and detection ismstantaneous.This principle was tested at the SMRI both as a standalone unit as part of a high performance liquid chrornatography installation, Unfortunately an impurity presentin factory samples interfered with sugar analysis (Morel duBoil, 1989),detracting from its use as a sucrose detector andfurther work was discontinued.

FlamephotometerFlame photometers have been in use at several sugar millsfor many years and have worked well provided they havebeen well m a i n t a . i n e ~ and serviced, The use of a flame photometer for momtonng entrainment was described by Daleand Lamusse (1977). However the instrument does not detect. sucr?se directly but measures the concentration of potaSSIUm IOns that are assumed always to be present with thesugar. For this reason they are considered to be only slightlybetter than conductivity monitors.

Conductivity monitorsIn South Africa the use of conductivity monitors as described by Cargill (1962) and Douglas (1962) has becomealmost universal, even though it is appreciated that the conductivity of a solution is an imperfect measure of the sugarcontent. The suitability of such a monitor depends on thepresence of other impuri ties in the form of various salts

which always accompany the sugar. The relationship between sucrose content and conductivity is highly variablebecause it depends on the nature and quantity of these salts.This in turn depends on the source of the contaminationand on the constituents of the cane, which can vary throughout the season..The i n d ~ s t r i a l type ofmulti-channel conductivity monitorwah multiple outputs for recording, indicating or triggeringalarms has been commercially available for many years andsome models have proved very reliable. .In a typical raw sugar mill the conductivity monitoringsystem can be made to work satisfactorily as a sugar detectorprovided it is properly designed and installed and wellmaintained. The most important point is that the conductivity

method will certainly react to gross contaminat ion ofhundreds ofppm and higher, and it is these incidents whichplace the boilers in the greatest danger.The inherent inaccuracy of the monitors could result inmomentary undetected sugar contamination of boiler feedwater of perhaps 50 ppm, which may sound excessive butwill not result in any serious problem in the boilers providedthe chemical treatment is carefully monitored and the pHand total dissolved solids are well controlled.

RecommendationsSystem designWith the correct design of condensate storage and hand

ling systems, the level of sugar contamination can be controlled to a reasonable level. In particular, the exhaustcondensate from the calandria of the first effectis most probably uncontaminated and should comprise at least 75% ofthe total boiler requirement. The maximum level of con-211

tamination of the other condensates as .detected by a conductivity monitor may be as high as 50 ppm before themonitor reacts and dumps the offending stream but themixture ofthese condensates would then end up w'ith about13ppm of contamination, which isbelieved to be completelysafe. In any event experience with conductivity monitorshas shown that the usual level that can be easily detected isactually much lower than 50 ppm.Some recommendations to be incorporated into the design, specification and operation of a condensate handlingsystem to meet the above requirements are provided in theAppendix.

Refinery condensatesThe special case of contamination of condensates by refined sl;lgar should be mentioned. Refined sugar containsinsufficient salts to trigger a conductivity monitor. In thecase of a.refinery attached to a raw sugar mill, there shouldbe sufficient condensate from the raw house to meet theneeds of the boilers without using any refinery pan conden

sate. An independent refinery would have to consider oneof the alternative detectors listed above which more specifically measure sugar contamination. The most promising ofthese appears to be the total organic carbon analyser whichcan be obtained and serviced locally. In this case the initialcost and operating expenses would be justified.Future possibilities

Computer surveillanceThere is much to be gained from the incorporat ion of asurveillance computer for the condensate handling system.This could take the form of a PC or microprocessor programmed to expect certain signals from the monitors, flowmeters, valves, tank level indicators and controls, and othercomponents. It would also expect the manual sampling valvesto be operated every hour and the conductivity electrodesto be checked and recalibrated at regular intervals. It couldeven be arranged to carry out the physical recalibration ofelectrodes at pre-determined intervals. As soon as none ofthese events takes place, or when anything strange occurs,

it will trigger the alarm and the engineer can be notified.Laboratory analysisAt each sugar mill the laboratory has the task of routinehourly sampling and analysis of condensates. The installation of a continuous analyser in the laboratory would be aneffective means of eliminating this drudgery and saving labour. The reaction time of this instrument would at least be

better than sampling all streams once per hour, and it wouldbe an effective backup to the conductivity monitoring system. It would of course require all condensate streams to bepiped to the laboratory or a suitable room in the factory.The important point is that the analyser would operate ina laboratory environment, and would therefore be expectedto perform satisfactorily provided supervisory control wasintroduced.Conclusions

The boilers in a raw sugar mill can be effectively protectedfr0r:" the effec:ts of sugar contamination by using a properlydesigned installed condensate handling system in whichthe sugar IS detected by a good conductivity monitoring instrument. The following features and precautions are important for efficient operation: The. system should be designed to include all possiblecontingencies

7/28/2019 1991_Reid_The Protection Of Boilers.pdf

5/5

Proceedings of The South African Sugar Technologists'Association - June 1991 Adequate storage should beprovided foruncontaminatedcondensate Operation and maintenance should be well managed Regular inspection and testing of the system is necessaryto ensure that alarmsand failsafe procedures are effective Procedures should beadequately documented andallstaffthoroughly trained.

REFERENCESBruijn,J (1962). Automaticsugardetectionin boilerfeed water. Proc SAfrSug Techno/ Ass36: 49-51.Cargill, JM (1962). Theconservation of condensate. Proc SAfrSugTechno/Ass36:52.Dale, TB and Lamusse, JP (1977). Monitoring of entrainmentby vapoursampling and the use of a flame photometer. Proc S AfrSug Techno/ Ass51: 116-118.Douglas, WEO(1962). Application ofan old principle in thecontrolofsugarcontaminated condensates. Proc S AfrSug Technol Ass36: 53-55.Fowler, MJ (1977). Continuoussugardetection in refinery wastewaters. ProcSuglnd Tech 36: 219-231.Morel du Boil, PG (1989). Report for April 1989 October 1989. SMRIDivisionalReport 6/89, Chemical Division: 1820.Parker, WHand Bond,GM (1958). The 'Spaldinlab'Sugar-Tektor.lnt SugJ 60 (711): 71-74.Schaffler, KJ (1978). Sugarentrainmentmonitoring. Proc SAfrSugTechnolAss 52: 123-124.Takatori,Y,Toyama, Rand Takezaki, T (1975). Acontinuousdeterminationof sugars in condensed water. Proc Res Soclap SugRef Tech 25: 47-52(Japanese).

APPENDIXRecommendations for an effective condensate handlingsystem Condensates returningto the boiler feedtanks should be given prioritydependent on their expected extent of contamination and hence theirsuitability for boiler feed. Thus exhaust condensate from the first evap-

212

orator effect would have highest priority, and vapour condensates fromlatereffects a lowerpriority. Juiceheatercondensates aregenerally suspectbecause even small leaksin juice heatertubeswill enter the condensatedue to the pressure differential. Theyare therefore not recommended forboiler feedwater purposes. All pipingfrom these different sources should be brought to the samepoint so that monitoring can be carried out at one central point. Eachpipe shouldbe providedwitha separateconductivity monitor andassociated dump valve. The accepted condensate pipes into the boiler feedwater tank shouldbeprovidedwithfurtherconductivity monitorsand dumpvalvesso that allcondensates are monitored twice on the wayto the boilers. Dumpvalvesshouldbearranged to discharge rejected condensate directlyinto the reject tank with no possibility of the outlets becoming blockedor choked. Alldump valves shouldbe providedwith a positive indicationof theirposition. This indicationshouldbe displayed at the control panel. Thecontrolsystem should be arranged so that highpriority condensatesare the lastto overflow to the sweetwater tank. The sweetwater tank should be sited so that there is no possibility ofcontaminating the feedwater in the event ofan overflow. The conductivity monitors should be regularly checked by competentstaff. At least once per shift the set point should be set down until thedump valveoperates, and the operationshouldbe carefully observed toensurethat there is no problem. The conductivity reading at this pointshould be recorded in a log bookand a sampleshouldbe takento thelaboratory to be checked. The set point is then reset to a level which isestablished by the engineer from time to time.

Abulk storage tankwitha capacity ofat leastthree hoursat full steamingrate shouldbe providedand arranged so that it can be kept full and thewaterused continuously fromit during normaloperation. Acontrolpanelshouldbe providedina position, sayboilercontrolroom,where it is continuously supervised. The panel shouldindicatethe conductivity reading anddump valvepositionofeverystreamaswell as thevarious tank levels. Alarmsshould be provided to sound whenany setpoint is exceeded and whenany dump valve operates. Lowtank levelsshouldalso soundalarms according to the system'sdemands.